Ijcmi Vol2 No1 No2 2010

INTERNATIONAL JOURNAL OF
COMPUTATIONAL MATHEMATICAL IDEAS
ISSN: 0974-8652 Volume 2 / Number 1 / January 2010

--------------------------------------------------------------------------------------------------------------------
CONTENTS
Research Papers Page No
A Note on r - partitions of n in which The Least Part is k 6-12

K.Hanuma Reddy
Mixed Convection In A Porous Medium With Magnetic Field, Variable Viscosity And 13-21
Varying Wall Temperature
C.N.B.Rao, V. Lakshmi Prasannam , T.Raja Rani
On Completely Prime And Completely Semi-Prime Ideals In Γ-Near-Rings 22-27

Satyanarayana Bhavanari, Pradeep Kumar T.V, Sreenadh Sridharamalle, Eswaraiah Setty Sriramula
A Unified Frame Work For Searching Digital Libraries Using Document Clustering 28-32
Shaik Sagar Imambi, Thatimakula Sudha
Reducibility For The Fiorini-Wilson-Fisk Conjecture 33-42

S.Satyanarayana, J.Venkateswara Rao, V.Amarendra Babu
Perceiving Plagiarism Using Weighted Window Approach- Performance Analysis 43-47

Bobba Veeramallu, T. Pavan Kumar, Prof.V.Srikanth, Prof.K.Rajasekhara Rao
System Representation For Software Architecture Recovery 48-55

Shaheda Akthar, Sk.MD.Rafi
Experimental and theoretical evaluation of ultrasonic velocities in binary liquid mixture 55-59
cyclohexane + o-xylene at 303.15, 308.15, 313.15 and 318.15k.
Narendra K, Narayanamurthy P & Srinivasu Ch
ISSN: 0974-8652 Volume 2 / Number 2 / August 2010

--------------------------------------------------------------------------------------------------------------------
CONTENTS
Research Papers Page No
Over view to Implementation of robotics with Voice recognition 60-64

Ande Stanly Kumar, Dr.K.Mallikarjuna Rao, Dr.A.Bala Krishna, B.Venkatesh,
Novelty of Extreme Programming 65-72

Ch.V.Phani Krishna, S.Satyanarayana, K.Rajasekhara Rao
Radiation Effects On Mhd Free Convection Flow Past A Semi-Infinite Moving Vertical 73-81
Porous Plate With Soret And Dufour Effect
G.Venkata Ramana Reddy and Dr. A.Rami Reddy
Pareto Distribution - Some Methods Of Estimation 82-92

R. Subba Rao, R.R.L. Kantam, G.Srinivasa Rao
COMPUTATIONAL MATHEMATICAL IDEAS (IJCMI)
ISSN: 0974-8652 www.ijcmi.webs.com
Dr.S.Satyanarayana mathematics interface between applied mathematics,
EDITOR-IN-CHIEF –IJCMI numerical computation, and applications of systems -
E-Mail: ijcmi@in.com,sijcmi@gmail.com oriented ideas to the physical, biological, social, and
behavioral sciences. It includes but not limited to areas
of mathematics such as algebra (especially group
S.Rajeswara Rao theory), combinatory (especially graph theory),
MANAGING EDITOR geometry, number theory and numerical analysis;
computational complexity; cryptology; symbolic and
S.Satish Babu algebraic computation; optimization; the mathematical
CONSULTING EDITOR aspects of: models of computation; automata theory;
categories and logic in computer science; proof theory;
type theory; semantics of programming languages;
S.Srinivasara Rao
process algebra and concurrent systems; specification
PRODUCTION EDITOR and verification; databases; rewriting; neural nets and
genetic algorithms; computational learning theory;
HONARARY EDITORIAL TEAM theorem proving ,Applied Physics, Solid State Physics,
Nuclear Physics ,Theoretical Physics and more...
1. Prof.Frederick P. Brooks, Jr (U.S.A) Journal publishes research articles and reviews within
the whole field of Mathematical Sciences& Computer
2. Prof. John Thompson(U.S.A)
Science and Engineering, and it will continue to provide
3. Prof.S.R.Srinivasa Varadhan(U.S.A)
information on the latest trends and developments in this
4. Prof.EDMUND M. CLARKE (U.S.A)
ever-expanding subject.
5. Prof. Dr Bhavanari Satyanarayana(INDIA)
Publication Date and Frequency:
6. Prof. Dr.J.Venkateswara Rao(INDIA)
Three issues in a year.
7. Prof.Curtis T McMullen(U.K)
Submission: Authors are requested to submit their
8. Prof.Terence Tao(U.S.A) papers electronically sent to:ijcmi@in.com ,
9. Dr.Kuncham Syam Prasad(INDIA) sijcmi@gmail.comwith mention journal title (IJCMI) in
10. Prof. Dr T V Pradeep Kumar(INDIA) subject line.
11. Prof.S.Pallam Setty(INDIA)
12.Prof.Dr.J.Saibabu(INDIA) Annual Subscription Price:
13.Dr.D.Naga Raju(INDIA) INDIA US$
14.Dr.B.V.Appa Rao(INDIA) 1Year 2000 400
15.Prof.V.Kalyan Raju(INDIA) 3Years 5800 760
16. Prof.Dr.G.Murugusundaramoorthy(INDIA)
17. T.Madhu Mohan(INDIA) 5 Years 9500 1900
18. A.Sri Krishna Chaitanya(INDIA)
19. Dr. Sumanta Kumar Tripathy (INDIA) For Subscriptions contact :
20.Prof.Andreea S.Calude(NZ) S.SATISH BABU ,
21.Dr.A.Rami Reddy (INDIA) ( Consulting Editor , IJCMI ) ,
22.Dr.K.Srinivasa Rao (INDIA) D.No.3-250 , Pedakakani Post – 522509 ,
23.Dr.P.Srinivasa Rao (INDIA)
Guntur Dist , A.P. , INDIA , Cell : 0091-9908490506
About this Journal
The INTERNATIONAL JOURNAL OF Printed at and Published on behalf of
COMPUTATIONAL MATHEMATICAL IDEAS Satyan’s Publications , Dr.S.SATYANARAYANA ,
(IJCMI) is a refereed Mathematics &Computer science Editor-in-Chief –IJCMI , ISSN : 0974-8652 ,
and Engineering journal devoted to publication of D.NO:3-250.,PEDAKAKANI(POST) ,
original research papers, research notes, and review GUNTUR(DIST).,A.P.,INDIA ,(www.ijcmi.webs.com)
articles, with emphasis on unsolved problems and open
questions in mathematics &Computer science and
Engineering. All areas listed on the cover of
Mathematical Reviews, such as pure and applied
mathematics.
The INTERNATIONAL JOURNAL OF

(IJCMI) is an international research journal, which
publishes top-level work on computational aspects of
Editorial
We are pleased to announce the launch Fourth & Fifth issue of the INTERNATIONAL JOURNAL OF
COMPUTATIONAL MATHEMATICAL IDEAS (IJCMI).The IJCMI is a refereed Mathematics &Computer
science and Engineering journal devoted to publication of original research papers, research notes, and review
articles, with emphasis on unsolved problems and open questions in mathematics &Computer science and
Engineering. All areas listed on the cover of Mathematical Reviews, such as pure and applied mathematics &
Computer science and Engineering.
The IJCMI is an international research journal, which publishes top-level work on computational aspects
of mathematics interface between applied mathematics, numerical computation, and applications of systems -
oriented ideas to the physical, biological, social, and behavioral sciences. It includes but not limited to areas of
mathematics such as algebra (especially group theory), combinatorics (especially graph theory), geometry,
number theory and numerical analysis; computational complexity; cryptology; symbolic and algebraic
computation; optimization; the mathematical aspects of: models of computation; automata theory; categories
and logic in computer science; proof theory; type theory; semantics of programming languages; process algebra
and concurrent systems; specification and verification; databases; rewriting; neural nets and genetic algorithms;
computational learning theory; theorem proving, Applied Physics, Solid State Physics, Nuclear Physics
Theoretical Physics and more...
IJCMI Journal publishes research articles and reviews within the whole field of Mathematical
Sciences& Computer Science and Engineering, and it will continue to provide information on the latest trends
and developments in this ever-expanding subject.
The IJCMI will be published three issues in a year. The journal will be reviewed by two independent
reviewers. In addition, the Journal may occasionally publish special issues on various topics in the areas of
Mathematics &Computer science and Engineering, book reviews, conference reports, letters to the honorary
editors, conference announcements, etc.
Finally, The Editor-in-Chief and Honorary Editors wish to congratulate the authors of the published
papers in IJCMI. Satyans Publications to starting Three more International Journals such as International Journal
of Applied Sciences & Engineering ideas (IJASE), International Journal of Artificial Intelligence Ideas (IJAII),
International Journal of Entrepreneurship Ideas (IJEI) for the inspiration of IJCMI. Thanks are due to the
members of the Editorial Board for their precious feedback and advice. We hope that the new International
Journal of Computational Mathematical Ideas will serve Mathematics & Computer science engineering research
community as well as and this journal will be main media of presenting ideas and research work in their area.
Suggestions to improve our efforts in order to deliver a better journal to the authors, readers and subscribers of
this journal will always be appreciated.
Dr.S.Satyanarayana
Editor-In-Chief –IJCMI
E-Mail:ijcmi@in.com
ISSN: 0974-8652
INTERNATIONAL JOURNAL OF COMPUTATIONAL MATHEMATICAL IDEAS
ISSN:0974-8652
INTERNATIONAL JOURNAL OF COMPUTATIONAL

MATHEMATICAL IDEAS
ISSN:0974-8652 / www.ijcmi.webs.com / VOL 2-NO1-PP 6-12 (2010)
A NOTE ON r − partitions OF n IN WHICH THE LEAST PART IS k

K.Hanuma Reddy@
Lecturer in mathematics, Hindu College, Guntur, A.P-522002, India, Mail:
hanumareddy_k@yahoo.com
ABSTRACT
Partitions play an important role in Number Theory. It has wide applications in various fields. An attempt is
made to develop a theorem on the number of r − partiions of positive integer n in which the least part is k ,
a reduction theorem on r − partiions and some more results on r − partiions are derived.
Key words: p ( n ) , r − partiions , p r ( n ) , pr ( e ; n ) , pr ( o ; n ) and pr ( S ; n )

Subject classification: 11P81 Elementary theory of partitions.
1. Introduction: 1.7 pr ( S ; n ) : pr ( S ; n ) is the number of

1.1 Partition: A partition of a positive integer n partitions of a positive integer n having r parts
is a finite sequence of non-increasing positive in which each part is the element of the set S.
r
integers λ1 , λ2 , ... , λr such that ∑ λi = n 2. Theorem: Let r, n ∈ N ( r ≤ n) and
i =1
The number λi is called the i th part of the S = {am + b | a ∈ N , b ∈ Z and m = 1,2,..., n} b

partition. The partition is also denoted as e the set of positive integers. If a | n − br , then
n = (λ1 , λ2 , ... , λr ) .  n − br 
pr ( S ; n ) = pr   other wise
 a 
1.2 Partition function: The partition function
pr ( S ; n ) = 0 .
p ( n ) is the number of partitions of n.
1.3 r-partition: A partition containing r parts is [ 2.1]

called r − partiions . Proof: All parts in r − partiions of n multiplied
1.4 pr ( n) : pr ( n ) is the number of by a and added by b to get the partitions of n
r − partiions of a positive integer n . whose parts are elements of S .
Note: p ( n ) = p1 ( n) + p2 ( n ) + ... + pn ( n )
3. Theorem: Let r , n ∈ N and
1.5 pr ( o ; n ) : pr ( o ; n ) is the number of S = {am + b | a ∈ N , b ∈ Z and m = 1,2,..., n}
partitions of a positive integer n having r parts be the set of positive integers. Then, the highest
in which each part is odd number. least part of r − partiions of n in which the
parts are the elements of the set S is
1.6 pr ( e ; n ) : pr ( e ; n ) is the number of
 n 
partitions of a positive integer n having r parts a
 ar + b  + b
in which each part is even number.
A Note On r - partitions Of n In Which The Least Part Is k

6
ISSN:0974-8652
( {
∴ pr ( S ; n ) - pr −1 S ; n − a (1) + b })
[3.1]
= pr ( S ; n − ar )
Proof: Let λ1 , λ2 , ... , λr be the first,
second,…, 5. Theorem: Let r , n , k ∈ N and
th
r parts of the r-partition of ‘n’ respectively. S = {am + b | a ∈ N , b ∈ Z and m = 1,2,..., n}
So n = (λ1 , λ2 , ... , λr ) be the set of positive integers. then, the number
All the distinct r − partiions of n are arranged of r − partiions of n having the parts are
in such a way that all the parts and elements of S with least part k is
corresponding parts in each r − partiions are
monotonically increasing.
(
pr −1 S ; n − ( k − 1) ar − {a + b} )
  n    n 
where 1 ≤ k ≤
If possible, let λ1 = a + b + 1  ar + b 
  ar + b  
Since all the parts in each r − partiions are [5.1]
monotonically increasing order, the least Proof: Let λ1 , λ2 , ... , λr be the first,
possible value of each λ i for i = 2 to r is second,…,
  n   r th parts of the r − partiions of n respectively.
a  + b + 1 .
  ar + b   So
Then the sum of all parts in partition is n = (λ1 , λ2 , ... , λr )
  n    All the distinct r − partiions of n are arranged
r a
  + b  + 1 . in such a way that all the parts and
  ar + b    corresponding parts in each r − partiions are
  n    monotonically increasing.
But r a
  + b  + 1 > n Fixing λ1 = a (1) + b , the remaining value
  ar + b   
This is contradiction. n − {a (1) + b} of n can be expressed as the sum
 n  of the remaining r − 1 parts λ2 , λ3 , ... , λr in
Hence λ1 = a  +b is the highest
 ar + b  (
pr −1 S ; n − {a (1) + b} ways. )
integer.
i,e The number of r − partiions in which the
4. Theorem: Let r , n ∈ N and least part of the partition is λ1 = a (1) + b is
S = {am + b | a ∈ N , b ∈ Z and m = 1,2,..., n} pr −1 ( S ; n − {a + b} ) .
be the set of positive integers. Then prove that
( {
pr ( S ; n) - pr −1 S ; n − a (1) + b }) Fixing λ1 = a ( 2 ) + b , the remaining value
= pr ( S ; n − ar ) [4.1] n − {a ( 2 ) + b} of n can be expressed as the sum

of the remaining r − 1 parts λ2 , λ3 , ... , λr in
Proof:
The number of r − partiions of n whose parts ( )
pr −1 S ; n − {a ( 2 ) + b} ways.
are elements of S with least part a (1) + b is Since all the parts in each r-partition are non
equal to the number of ( r − 1) − partitions of

(
decreasing, pr − 2 S ; n − {a ( 3) + b ( 2 )} )
r − partiions with λ1 = a ( 2 ) + b , λ2 = a (1) + b
{
n − a (1) + b } whose parts are elements of
are to be eliminated from
S and the number of r − partiions of n whose
parts are elements of S with least part is ( {
pr −1 S ; n − a ( 2 ) + b }) r − partiions .Then,
not a (1) + b is equal to the number the number of the r − partiions in which the
of r − partiions of n − ar whose parts are
elements of S .

7
ISSN:0974-8652
least part of the partition λ1 = a ( 2 ) + b is  n 

where 1 ≤ k ≤  
 ar + b 
( {
pr −1 S ; n − a ( 2 ) + b })
- Corollary 5.1: Let n, r , k ∈ N . Then the number
( {
pr − 2 S ; n − a ( 3 ) + b ( 2 )}) of r − partiions of n with least part k is
= pr −1 ( S ; n − a ( r − 1) − {a ( 2 ) + b}) n
pr −1  n − ( k − 1) r − 1 where 0 ≤ k ≤  
r 
= pr −1 ( S ; n − ar − {a + b})
Proof: Put a = 1, b = 0 in [5.1]
Fixing λ1 = a ( 3) + b , the remaining value
{ }
n − a ( 3) + b of n can be expressed as the sum Corollary 5.2: Let n, r , k ∈ N . Then, the
number of r − partiions of n having the parts
of the remaining r − 1 parts λ2 , λ3 , ... , λr in
are even numbers with least part k is
( )
pr −1 S ; n − {a ( 3) + b} ways.
n
pr −1  e ; n − 2 ( k − 1) r − 2  where 0 ≤ k ≤  
pr − 2 ( S ; n − {a ( 4 ) + b ( 2 )}) r − partiions with  2r 
λ1 = a ( 3) + b , λ2 = a (1) + b and Proof: Put a = 2, b = 0 in [5.1]
(
pr − 2 S ; n − {a ( 5 ) + b ( 2 )} ) Corollary 5.3: Let n, r , k ∈ N . Then the
-
number of r − partiions of n having the parts
(
pr −3 S ; n − {a ( 6 ) + b ( 3)} ) are odd numbers with least part k is
r − partiions with λ1 = a ( 3) + b , λ2 = a ( 2 ) + b pr −1 o ; n − 2 ( k − 1) r − 1
are to be eliminated  n 
where 0 ≤ k ≤  
(
from pr −1 S ; n − {a ( 3) + b} r − partiions . )  2r − 1 
Then, the number of the r − partiions in Proof: Put a = 2, b = −1 in [5.1]
which the least part of the partition
λ1 = a ( 3) + b is 6. Reduction theorem for pr ( S ; n) :
Let r , n, k ∈ N and
(
pr −1 S ; n − {a ( 3) + b} ) S = {am + b | a ∈ N , b ∈ Z and m = 1, 2,..., n}
(
- pr − 2 S ; n − {a ( 4 ) + b ( 2 )} ) be the set of positive integers. then,
-  n 
 ar +b 
 
 (
 pr − 2 S ; n − {a ( 5 ) + b ( 2 )} ) 
 pr (S; n) = ∑ pr −1 ( S ; n − ( k − 1) ar − {a + b} )
 
 (
− pr −3 S ; n − {a ( 6 ) + b ( 3)} ) 
k =1
= pr −1 ( S ; n − a ( r − 1) − {a ( 3) + b}) [6.1]
and
− pr − 2 ( S ; n − a ( r − 2 ) − {a ( 5 ) + b})
 n 
= pr −1 ( S ; n − ar − {a ( 2 ) + b}) n  ar +b 
− pr − 2 ( S ; n − ar − {a ( 3 ) + b}) p(S; n) = ∑∑
r =1 k =1
pr −1 ( S ; n − ( k −1) ar − {a + b} )
= pr −1 ( S ; n − a ( r − 1) − ar − {a ( 2 ) + b})
[6.2]
= pr −1 ( S ; n − 2 ar − {a + b} ) Proof: From [5.1] we can observe it
By induction we observe that the number of Corollary 6.1: Prove that
r − partiions of n having the parts are elements n
of S with least part k is n  r 
pr −1 ( S ; n − ( k − 1) ar − {a + b} )
p (n) = ∑∑ p ( n − ( k − 1) r − 1)
r =1 k =1
r −1

8
ISSN:0974-8652
Proof: Put a = 1, b = 0 in [ 6.2] Proof: Case 1: Let n = 2m for m ∈ N

p(n) = p(2m)
= p1 (2m) + p2 (2m) + p3 (2m) + ...
Corollary 6.2: Prove that
+ pm −1 (2m) + pm (2m) + pm +1 (2m) + ...
n + p2 m − 2 (2m) + p2 m −1 (2m) + p2 m (2m)
n  2 r 
p(e; n) = ∑ ∑ p ( e ; n − 2 ( k − 1) r − 2)
r =1 k =1
r −1 = { p1 (2m −1)}
+{ p1 (2m − 2) + p2 (2m − 2)}
Proof: Put a = 2, b = 0 in [ 6.2] +{ p1 (2m − 3) + p2 (2m − 3) + p3 (2m − 3)} + ...
Corollary 6.3: Prove that +{ p1 (m +1) + p2 (m +1) + ... + pm−1 (m +1)}
 n 
n  2 r −1 
+ p(m) + p(m −1) + ... + p(2) + p(1) +1
p(o; n) = ∑∑
r =1 k =1
pr −1 ( o ; n − 2 ( k − 1) r − 1)
= { p1 (2m −1)}
+ { p1 (2m − 2) + p2 (2m − 2)}
Proof: Put a = 2, b = −1 in [ 6.2] + { p1 (2m − 3) + p2 (2m − 3) + p3 (2m − 3)} + ...
+ { p1 (m + 1) + p2 (m + 1) + ... + pm−1 (m +1)}
Theorem 7: If r , n ∈ N and r < n , then
+ { p1 (m) + p2 (m) + ... + pm−1 (m) + pm (m)}
n
pr (n) = p(n − r ) for ≤2 + { p1 (m −1) + p2 (m −1) + ... + pm−1 (m −1)} + ...
r
+ { p1 (3) + p2 (3) + p3 (3)}
+ { p1 (2) + p2 (2)} + p1 (1) +1
Proof:
Case:1 Let
n
=2 =1+ ∑ pi ( j ) + ∑ pi ( j ) +…
r i+ j =2m i + j = 2 m −1
⇒ n = 2r
pr (n) = p1 (n − r ) + p2 (n − r ) + ... + pr (n − r )
+ ∑
i + j =3
pi ( j ) + ∑
i+ j =2
pi ( j )
= p1 (r ) + p2 (r ) + ... + pr (r ) 2m
= p(r )
= 1+ ∑
i+ j=2
pi ( j )
= p( n − r ) n
n
= 1+ ∑
i+ j=2
pi ( j )
Case:2 Let < 2
r Case 2: Let n = 2m + 1 for m ∈ N
n p (n) = p (2m + 1)
Since r < n and < 2
r = p1 (2m + 1) + p2 (2m + 1)
⇒ r < n < 2r
⇒ 0< n−r < n + p3 (2m + 1) + ... + pm (2m + 1)
pr (n) = p1 (n − r ) + p2 (n − r ) + ... + pm +1 (2m + 1) + pm + 2 (2m + 1) + ...
+ pn −r (n − r ) + pn− r +1 (n − r ) + ... + pr (n − r ) + p2 m −1 (2m + 1) + p2 m (2m + 1)
+ p2 m +1 (2m + 1)
= p1 (n − r ) + p2 (n − r ) + ...
+ pn −r (n − r ) + 0 + ... + 0
= p( n − r )
n
Hence pr (n) = p(n − r ) for ≤2
r
Theorem 8: Let n, i , j ∈ N , then

n
p ( n) = 1 + ∑
i+ j=2
pi ( j )

9
ISSN:0974-8652
= { p1 (2m)}
+ pm +1 (n) +…+ p2 m -2 (n) + p2 m −1 (n) + p2 m (n)
+ { p1 (2m − 1) + p2 (2m − 1)}
+ { p1 (2m − 2) + p2 (2m − 2) + p3 (2m − 2)} + ... = p1 (2m) + p2 (2m) +…+ pm −1 (2m) + pm (2m)
+ { p1 (m + 1) + p2 (m + 1) + ... + pm (m + 1)} + + pm +1 (2m) +…+ p2 m -2 (2m) + p2 m −1 (2m)
p(m) + p(m − 1) + ... + p(2) + p(1) + 1 + p2 m (2m)
= { p1 (2m)}
+ { p1 (2m − 1) + p2 (2m − 1)} = pm −1 (2m + m − 1) + p(m) + p(m -1) +…+ p(2)
+ { p1 (2m − 2) + p2 (2m − 2) + p3 (2m − 2)} + ... + p(1) +1
+ { p1 (m + 1) + p2 (m + 1) + ... + pm ( m + 1)}
=1+ pm −1 (2m + m -1) + { p(1) + p(2) + ... + p(m)}
+ { p1 (m) + p2 (m) + ... + pm ( m)}
m
+ { p1 (m − 1) + p2 ( m − 1) + ... + pm −1 (m − 1)} + ... = 1+ pm −1 (3m -1) + ∑ p(i)
+{ p1 (3) + p2 (3) + p3 (3)} i =1
n
+ { p1 (2) + p2 (2)} + p1 (1) + 1 2
 3n 
∑ ∑
= 1+ p n  -1 + ∑ p(i)
2  
=1+ pi ( j ) + pi ( j ) +…+ −1 2
i =1
i + j = 2 m +1 i + j =2m
2 m +1
If n is odd
= 1+ ∑
i+ j=2
pi ( j ) Let n = 2m + 1 for m ∈ N
p ( n) = p1 (n) + p2 (n) +…+ pm −1 (n)
n
= 1+ ∑
i+ j=2
pi ( j ) + pm (n) + pm +1 (n) + pm + 2 (n) +…
+ p2 m −1 (n) + p2 m (n) + p2 m +1 (n)
n
Hence p ( n) = 1 + ∑
i+ j=2
pi ( j )
= p1 (2m +1) + p2 (2m +1) +…+ pm (2m +1)
Corollary 8.1: Let n be a natural number, then + pm+1 (2m +1) + pm+2 (2m +1) +…
p(n) − p(n − 1) = ∑
i+ j=n
pi ( j )
+ p2m-1 (2m +1) + p2 m (2m +1) + p2m+1 (2m +1)
= pm (2m + m +1) + p(m) + p(m -1) +…
n
+ p(2) + p(1) +1
Proof: Since p(n) = 1 + ∑
i+ j=2
pi ( j ) m
= 1+ pm (3m + 1) + ∑ p (i )
n −1 i =1
∴ p(n − 1) = 1 + ∑ pi ( j )
 3n -1  2
n −1
+ ∑ p (i )
i + j =2
= 1+ p n -1 
2 
2  i =1
Hence p(n) − p(n − 1) = ∑
i+ j=n
pi ( j )
Hence
Theorem 9: If n ∈ N then
Theorem 10:
m  m + 1
If   = 1 ,then   = 1 for m, r ∈ Z
r  r +1 
Proof:
If n is even m
Proof: Since   = 1
Let n = 2m for m ∈ N r
p ( n) = p1 (n) + p2 (n) +…+ pm −1 (n) + pm (n) ⇒ r ≤ m < 2r
⇒ r + 1 ≤ m + 1 < 2r + 1

10
ISSN:0974-8652
Since r + 1 ≤ m + 1 < 2r + 1 and 2r + 1 ≤ 2r + 2 Proof:

∴ r + 1 ≤ m + 1 < 2r + 2 From theorem 11
m +1 pr (m) + pr (m + r + 1) + ... + pr (m + tr + t )
⇒1≤ <2
r +1 m
 m + 1 = pr +1 (m + tr + t ) when   = 1
⇒ r
 =1
 r +1  Put m = r and
m  m +1 pr ( r ) + pr (2r + 1) + ... + pr  nr + ( n − 1) 
Hence   = 1 ⇒   = 1 for m, r ∈ Z
 
r  r +1 
= pr +1 [ n( r + 1) ]
m
Theorem 11: If   = 1 , then
r ACKNOWLEDGEMENT
pr (m) + pr (m + r + 1) + ... + pr (m + tr + t )
We would like to express our thanks to referees
= pr +1 (m + tr + t ) for valuable comments that improved the paper.
Proof: Let t = 1
m REFERENCES
Since   = 1
r [1]. George E. Andrews, The theory of
There fore partitions, Encyclopedia of Mathematics and its
 m + r + 1 + 1  m + 1 r + 1 Applications, Vol. 2, Addison-Wesley
 = + = [1 + 1] = 2
 r + 1   r + 1 r + 1  Publishing Co., Reading, Mass.-London-
Amsterdam, 1976.
Since [2]. G. E. Andrews and F. G. Garvan, Dyson’s
pr +1 (m + r + 1 + 1) = pr (m + r + 1) crank of a partition, Bull. Amer. Math. Soc.
(N.S.) 18 (1988), 167–171.
+ [ pr (m + r ) − pr (m + r − 1)]
[3]. G. E. Andrews, The number of smallest
= pr (m + r + 1) + pr (m + r − r ) parts in the partitions of n, J. Reine Angew.
= pr (m + r + 1) + pr (m) Math., to appear.
[4]. G. E. Andrews and F. G. Garvan, Dyson’s
We assume that it is true for t = s
crank of a partition, Bull. Amer. Math. Soc.
There fore
(N.S.) 18 (1988), 167–171.
pr (m) + pr (m + r + 1) + ... + pr (m + sr + s) [5]. T. M. Apostol, Modular Functions and
= pr +1 (m + sr + s + 1) Dirichlet Series in Number Theory, Springer-
Verlag, NewYork, 1976.
Adding both sides by pr (m + sr + s + r + 1) [6]. A. O. L. Atkin and F. G. Garvan, Relations
pr (m) + pr (m + r + 1) + ... between the ranks and cranks of partitions,
Ramanujan J. 7 (2003),343–366.
+ pr (m + sr + s ) + pr (m + sr + s + r + 1) [7]. K. Bringmann, F. G. Garvan and K.
= pr +1 (m + sr + s + 1) + pr (m + sr + s + r + 1) Mahlburg Partition statistics and quasiweak
Maass forms, Internat. Math.Res. Notices, to
= pr +1  m + ( r + 1) s + 1 + pr  m + ( r + 1)( s + 1) 
appear.
= pr +1  m + ( r + 1) s + 1 + pr  m + ( r + 1) s + 1 + r  [8]. K. Bringmann and K. Ono, Dyson’s ranks
and Maass forms, Ann. of Math., to appear.
= pr +1  m + ( r + 1)( s + 1) + 1 [9]. K. S. Chua, Explicit congruences for the
It is true for t = s + 1 partition function modulo every prime, Arch.
There fore our statement is true for all t ∈ N Math. (Basel) 81 (2003).
Hence [10]. F. J. Dyson, “Selected papers of Freeman
pr (m) + pr (m + r + 1) + ... + pr (m + tr + t ) Dyson with commentary,” Amer. Math. Soc.,
Providence, RI, 1996.
m [11]. F. J. Dyson, Some guesses in the theory of
= pr +1 (m + tr + t ) when   = 1 partitions, Eureka (Cambridge) 8 (1944), 10–15.
r
[12]. F. Garvan, D. Kim, D. Stanton, Cranks
Corollary 11.1: Prove that and t-cores, Invent. Math. 101 (1990), 1–17.
pr ( r ) + pr (2r + 1) + ... + pr  nr + ( n − 1)  [13]. G. Gasper, M. Rahman, Basic
hypergeometric series, Encyclopedia of
= pr +1 [ n( r + 1) ] Mathematics and its

11
ISSN:0974-8652
applications v.35, Cambridge, 1990.

[14]. K.Hanuma Reddy: Note on convex
polygons, proc of Joint Conference on
Information
Sciences 2007, USA, 1691-1697.
[15]. G. H. Hardy and EM. Wright “An
Introduction to the Theory of Numbers,” Oxford
Univ. Press, London, 1979.
[16]. S. Ramanujan, “The lost notebook and
other unpublished papers,” Springer-Verlag,
Berlin, 1988.

12
ISSN:0974-8652

MATHEMATICAL IDEAS
MIXED CONVECTION IN A POROUS MEDIUM WITH MAGNETIC FIELD,

VARIABLE VISCOSITY AND VARYING WALL TEMPERATURE
C.N.B.Rao$,V. Lakshmi Prasannam# , T.Raja rani*

$
S.R.K.R. Engineering College, Bhimavaram, Andhra Pradesh, India.
#
P. B Siddhardha P.G.College of arts and science, Vijayawada,
*
Sri Vishnu Engineering College for Women, Bhimavaram. E-Mail: rajarani72@gmail.com
ABSTRACT
Effects of magnetic field and variable viscosity on similarity solutions of mixed convection adjacent to a
vertical flat plate in a porous medium are studied numerically. Excess of plate temperature over the
ambient temperature and the free stream are assumed to vary as power functions of x , where x is the
distance measured along the plate. The flow and heat transfer quantities of the similarity solutions are
found to be functions of C, λ , γ µ , RP where C is magnetic interaction parameter, λ is power of
index of the plate temperature, γ µ is viscosity variation coefficient and RP, mixed convection
parameter is ratio of the Rayleigh number to the Pe ′clet number. The cases of assisting flow and
opposing flow are discussed. Dual solutions are found for negative values of RP, and ranges of values of
RP are found for which either a unique solution, no solution or dual solutions exist. Skin friction and heat
transfer coefficients are observed to diminish as the intensity of the magnetic field increases (or C takes
diminishing values). The range of negative values of RP over which solutions exist is observed to
increase with decreasing values of C as well as with increasing values of λ and γ µ .
Key Words: Mixed Convection; Variable viscosity; Magnetic field; varying wall temperature
Mathematics Subject Classification Codes: 76S05; 76R10; 76R05; 76DXX
NOMENCLATURE Rax – Rayleigh number,

B0 = Magnetic flux
ρ ∞ K g β (To − T∞ ) x
C - Magnetic interaction parameter
f - Dimensionless stream function
µ f αm
g - Acceleration due to gravity Ra x
K – Permeability RP - Mixed convection parameter,
Pe x
L2
K * - Porous parameter, T0 - Temperature of plate
K
B02 L2σ T∞ - Ambient temperature
M2 – Hartmann number,
µf Tf - Reference temperature
p – Pressure u, v - Velocity components in x - and y -
U∞ x directions
Pe x – Pe ′clet number,
αm U∞ - Free stream velocity
x , y – Cartesian coordinates
Mixed Convection In A Porous Medium With Magnetic Field, Variable Viscosity And Varying Wall Temperature
13
ISSN:0974-8652
Greek Symbols λ and ε where λ is the power of index of

αm - Effective thermal diffusivity of the the plate temperature and ε , the mixed
porous medium convection parameter is the ratio of the
β - Coefficient of thermal expansion Rayleigh number to the P e ′ clet number. Both
assisting flow and opposing flow were
γ µ - Viscosity variation coefficient discussed. Ranges of the values of ε for
η - Similarity variable different values of λ were presented for which
θ - Dimensionless temperature either a unique solution, dual solutions or no
µ - Dynamic viscosity solution exist. The effects of λ and ε on the
ρ - Fluid density flow and heat transfer characteristics were
discussed. Reference [4] discussed mixed
ψ - Stream function convection boundary layer flow over a vertical
λ − Power of index of plate temperature surface for the Darcy model when viscosity
varies inversely as a linear function of
Subscripts temperature. Results of both assisting flow and
0 – condition at the plate opposing flow were presented which were
∞ - condition at infinity discussed as functions of the mixed convection
f - condition at reference temperature parameter ε and variable viscosity
u - upper solution parameter θ c . In the opposing flow case, the
l - lower solution existence of dual solutions and boundary layer
separation were noticed.
I.INTRODUCTION There has been increasing attention to
Heat transfer studies in porous media find the study of magnetic field on convection flows
applications in several Engineering and in porous media as pointed out in [8]. Reference
technological systems (ref. [2]). In mixed [7] studied free convection at a vertical plate in
convection flows, when the temperature of the a porous medium in the presence of magnetic
plate varies as a power function of distance, field, variable physical properties and varying
similarity exists only if the free stream velocity plate temperature. Magnetic field effects on the
also varies according to the same power free convection and mass transfer flow through
function of distance as that of the plate a porous medium with constant suction and
temperature. In mixed convection flows, there constant heat flux has been discussed in [1]. The
arise four cases, of which two correspond to effect of magnetic field and varying plate
assisting flow and two to opposing flow (refer temperature on convective heat transfer past a
[10]), depending on the ambient and plate vertical plate in porous medium has been
temperatures and the direction of the free discussed in [9]. Magneto hydrodynamic mixed
stream. They are (i) hot plate assisting flow (ii) convection flow has been analyzed in an
hot plate opposing flow (iii) cold plate assisting annular region filled with a fluid saturated
flow (iv) cold plate opposing flow. Of these porous medium in [3]. A transverse magnetic
four cases only two (i), (iv) are taken into field which acts radially is created by a
consideration in this study. stationary electric current that flows through a
Reference [6] discussed the effect of cylindrical shaped electrical cable present in the
variable viscosity on convective heat transfer in annular region. The effect of non uniform
three different cases of natural convection, magnetic field on the flow and heat transfer of
mixed convection and forced convection, taking the Darcy model is discussed. Magneto
fluid viscosity to vary inversely with hydrodynamic free convection in a horizontal
temperature. However, the authors have cavity filled with a fluid saturated porous
confined their attention to the assisting flow medium with internal heat generation has been
case only. Reference [2] studied mixed studied in [5]. Assuming that the magnetic field
convection boundary layer flow on a vertical is inclined at angle γ with the horizontal plane,
surface in a porous medium, when both the
temperature of the plate and the free stream the flow and heat transfer are discussed as
velocity vary as the same power function of functions of inclination angle γ , Hartmann
distance along the plate. Similarity solutions number Ha, Rayleigh number Ra and aspect
were found to be functions of two parameters ratio a.
14
ISSN:0974-8652
A viscosity variation coefficient γµ is

introduced as
1  dµ 
γµ =   (T 0 − T ∞ ).
µ f  dT  f
Density of the fluid is assumed to be a
function of temperature only in the body force
term. The ambient fluid flows with a velocity
U∞ parallel to the vertical plate, the flow
being vertically upwards. The physical model
and coordinate system are presented in
figure1.The governing equations of the present
analysis and the boundary conditions are well
In the present paper the effects of magnetic known and are not presented here.
field, variable viscosity and varying plate Taking the free stream velocity as U ∞ = b x λ
temperature on mixed convection at a vertical where b is a constant, introducing
plate in a porous medium are studied. The plate P e ′ clet number ( Pe x ) and nondimensional
temperature and the free stream velocity are
assumed to vary as power functions of distance functions f , θ together with a similarity
( x ) along the plate, viscosity is assumed to variable η through the relations
vary as a linear function of temperature and a 
magnetic field is assumed to act normal to the 
plate. Similarity solutions are obtained for the 
problem and both assisting flow and opposing U x 
flow are discussed. In the opposing flow case, Pe x = ∞

dual solutions (referred to as upper and lower α m 
solutions) are obtained for certain values of the 
mixed convection parameter RP and ranges of 
values of RP are also obtained for which either ψ 
f (η ) = 1 
a unique solution, dual solutions or no solution 
( 2α x λ +1 ) 2
exist. Significant differences are noticed m

between the flow and heat transfer quantities T − T∞ 
related to the upper and lower solutions. θ (η ) = 
II. FORMULATION AND SOLUTION T0 − T∞ 
Let a flat plate be embedded vertically in a 1 
porous medium saturated with a viscous y  x λ +1 2 
η =   
incompressible homogeneous fluid. The porous x  2α m  
medium is assumed to be homogeneous and is
in thermal equilibrium with the surrounding (1)
fluid. Let a magnetic field of uniform strength
be applied in a direction normal to the plate.
Let X-axis be taken along the plate and Y-axis the governing equations in the mixed
perpendicular to it. The temperature of the plate convection case are obtained as
  1 
( T0 ) is assumed to vary as a power function of 1 + C γ µ  θ − 2   f ′′ + C γ µ f ′ θ ′ = C ( RP ) θ ′
  
λ
distance along the plate, as T0 = T∞ + A x (2)
where T∞ is temperature of the ambient fluid, θ ′′ − 2 λ f ′ θ + (1 + λ ) f θ ′ = 0
(3)
A is a constant and λ is a real number. Fluid
where
viscosity is assumed to be a function of
temperature as µ = µ f s µ (t ) , where µ f is K2
C= ,
viscosity evaluated at the film temperature, K2 +M 2
 dµ  B02 L2 σ
s µ (t ) = 1 +  (
 T −Tf . ) and
M = 2
,
 dt  f µf
 T + T ∞  is the film temperature.
Tf = 0  ρ K g β (T0 − T∞ ) x
 2  Ra x = ∞ , and
µf α
15
ISSN:0974-8652
Ra x values for opposing flow. When RP is zero, the

RP = . results correspond to the forced convection
Pe x case. Calculations are done for a wide range of
positive and negative values of RP. Enhanced
The boundary conditions in terms of f and θ flow can correspond to an increase in the value
of RP, as an increase in the value of the
are
parameter can be due to an increase in the
at η = 0, θ = 1, f = 0,  temperature difference ( T0 −T∞ ).

as η → ∞, θ → 0, f ′ → 1 To determine certain important values for λ ,
(4) the total heat convected in the flow, Q( x) at
Equation (2) can be integrated once using the
any down stream location x is considered.
condition on f ′ at infinity to get ∞
 Cγµ 
1 −  + C ( RP) θ
Q (x) = ∫ρC p β (T − T ∞ ) u dy .
 2  0
f′= 3 λ +1
  1  2
This can be seen to be proportional to x
1 + Cγ µ θ − 2  ,
   like in the free convection case(ref. [7]). For
uniform heat flux surface, Q ( x ) should vary
1
linearly with x and so λ = . For an adiabatic
(5) 3
Evaluating at η = 0, we get the slip velocity surface, Q ( x ) should be independent of x and
f ′(0) as −1
so λ = . Zero value of λ corresponds to
2 (1 + C ( RP ) ) − C γ µ 3
f ′( 0 ) = . the isothermal case. In this study solutions are
2 + Cγ µ found for the values
− 0.3, − 0.2, 0, 0.3, 0.5 and 1 of λ .
When A is positive, an increase in the value of
If γ µ = 0 then f ′(0) = 1 + C ( RP) λ can correspond to an increase in the
III.a PARAMETERS OF THE PROBLEM temperature of the plate, and, in a broader sense,
AND THEIR EFFECT ON THE FLOW it can result in enhanced flow.
AND HEAT TRANSFER: When there is no magnetic field, the
The flow and heat transfer depend on parameter C takes the value unity and for
the parameters γ µ , λ , C and RP where RP is increasing intensity of the magnetic field, the
the ratio of the Rayleigh number to the parameter takes values smaller than unity. In the
present study, solutions are found for the values
P e ′clet number.
0.1, 0.5 and 1 of C. Reduced flow can be
The constant A appearing in the expression for expected for smaller values of C or for
the temperature of the plate can take positive as increased intensity of the magnetic field as the
well as negative values and, as a result, the magnetic field lines obstruct the flow.
temperature of the plate can be higher or lower The effect of simultaneous variation of the
than the ambient temperature. In the present values of the parameters on the flow and heat
work, these correspond to assisting flow and transfer are presented in the discussion.
opposing flow respectively. For liquids (except III.b. NUMERICAL SOLUTION: The
0
for water near 4 C ) the parameter γ µ takes equations for f and θ , i.e., equations 3,5 are
negative values when T0 > T∞ and takes integrated numerically subject to appropriate
boundary conditions by Runge-Kutta-Gill
positive values when T0 < T∞ , while for gases method (Ref. [10]), together with a shooting
technique. The accuracy of the method is tested
it is vice versa. Irrespective of the values of T0 by comparing appropriate results of the present
and T∞ , zero value of γ µ corresponds to analysis with available results. Our results for C
=1 and γ µ =0 (i.e., no magnetic field, constant
constant viscosity case. In this paper, solutions
are found for the values -1, 0, and 1 of γ µ . viscosity) are in very good agreement with
those of in ref. [2]. Also our results for C =1,
The mixed convection parameter RP takes
positive values for assisting flow and negative
γ µ =0 and λ =0 (i.e., no magnetic field,
16
ISSN:0974-8652
constant viscosity and isothermal plate) agree case ( λ = 0 ) while it increases with γ µ when
very well with those of ref. [4 ].
IV. DISCUSSION OF THE RESULTS: λ takes positive values.
Qualitatively interesting results related to In the following, more attention is paid to the
the shear stress, heat transfer coefficient, discussion of the dual solutions of the opposing
velocity and temperature are presented, some of flow case. For a given value of RP, the solution
them in the form of tables I,II and others in the corresponding to a relatively larger value of
form of figures 2 to 11. Quantities such as the f ′′ ( 0 ) is referred to as the upper solution
Nusselt number and drag coefficient can be and the one corresponding to a smaller value of
readily obtained from the heat transfer f ′′ ( 0 ) as the lower solution.
coefficient and skin friction. Variations in
f ′ ( 0 ), f ′′ ( 0 ) and '−θ ′( 0 ) ’ for positive
The changes in skin friction with negative
values of RP are presented in table I. Skin values of the mixed convection parameter RP
friction f ′′ ( 0 ) can be observed to be negative are shown in figures 2(a),2(b) for different
for positive values of RP for all values of the values of the parameters C, λ and γ µ . The
other parameters under consideration. Absolute
value of f ′′ ( 0 ) decreases with increasing corresponding changes in heat transfer
coefficient are shown in figures 3(a),3(b)
values of RP and γ µ while it increases with respectively. One curve each corresponding to
increasing values of C.. ref. [2] are presented in figures 2(a),2(b) and
one curve corresponding to ref.[4] in fig.3(a).
From the figures the range of values of RP over
which solutions exist can be seen to be more
when fluid viscosity is taken to be temperature
dependent than when it is constant. Similarly
the range is more in the presence of magnetic
field than in its absence. In the isothermal case,
when viscosity is a constant as well as variable
and in the presence as well as absence of
magnetic field, f ′′ ( 0 ) is observed to be
positive. For C = 0 . 5 , λ = 0 , γ µ = 0 single
solution exists for − 2.1 ≤ RP ≤ 0 , dual
− 2.7 ≤ RP ≤ − 2.0 and
solutions exist for
no solution for RP ≤ − 2.7 . Like the skin
friction f ′′ ( 0 ) , the heat transfer coefficient also
takes positive values when λ = 0 (see figures
2(a) and 3(a)). Except for magnitude, behaviour
of skin friction and heat transfer coefficient
when γ µ = 1 are similar to the corresponding
ones when γ µ = 0.
Heat transfer coefficient ‘ − θ ′(0) ’ takes

increasing values with increasing values of RP
and C while it takes decreasing values with
increasing values of γ µ . In table II are
presented the ranges of values of RP for
which either no solution, a single solution or
dual solutions exist. The range of values can be
seen to be more for gasses than for liquids. The
range can also be seen to increase with
increasing values of λ . The range decreases
with increasing values of γ µ in the isothermal
17
ISSN:0974-8652
18
ISSN:0974-8652
example λ = 0.05) , f ′′ ( 0 ) is observed to

take both positive and negative values with
changing negative values of RP, and dual
solutions exist for a wide range of values of RP.
For C =0.5, λ =0.05 and γ µ =0 the range over
which
solutions exist is − 2 . 7 ≤ RP ≤ − 0 . 1 . Like
the skin friction, heat transfer coefficient also
takes both positive and negative values with
changing values of RP, when λ = 0.05 .
Plots of shear stress for the upper and lower
solutions for different values of the parameters
are shown in the figures 4(a),4(b),5(a) and 5(b).
Considerable differences can be noticed in the
behaviour of the shear stress for the upper and
lower solutions (see figures 4(a) and 4(b)).
Curves of figure 4(a) correspond to those for
liquids while those of figure 4(b) correspond to
gases. From figures 5(a) and 5(b) and also from
numerical results, it can notice that, for positive
values of RP, the shear stress at the plate
becomes negative thereby indicating separation
of the boundary layer.
Fluid velocity profiles for the two solutions of
the opposing flow case are presented in figures
6(a), 6(b) (for liquids) and in figures 7(a) and
7(b) (for gases). It can be observed that the
hydrodynamic boundary layer thickness of the
lower solution is much larger than that of the
upper solution. Qualitative differences between
the two solutions can also be observed in the
vicinity of the plate. Fluid temperature profiles
corresponding to the upper and lower solutions
are presented in figure 8 for certain negative
values of RP. It can be noticed that thermal
boundary layer thickness of the lower solution
is much larger than that of the upper solution.
Variations in the lower solutions with changing
Unlike in the isothermal case, when the plate values of the parameters are significant than
temperature is variable (for those in the other solution. From figure 9,
f ′′ ( 0 ) can be seen to
19
ISSN:0974-8652
results of the two works for C = 1, λ =0 and

γ µ =0.
VI.CONCLUSIONS:
Assisting flow (RP Positive)
1. For fixed values of λ , C , γ µ , as RP increases

there is an increase in the magnitudes of
f ′′ ( 0 ) and ‘ − θ ′(0) ’. Increase in ‘RP’ can
mean increase in the buoyancy force and and
this can cause an increase in fluid velocity and
hence an increase in the skin friction and heat
transfer coefficient.
2. For fixed values of γ µ , RP & λ ,

f ′′ ( 0 ) as well as ‘ − θ ′(0) ’ decrease as ‘C’
decreases ( i.e., as the intensity of the magnetic
field increases).
Opposing flow (RP Negative) 1.

‘ − θ ′(0) ’ decreases with diminishing values of
RP. This may be due to the buoyancy force that
works against the flow and hence the retardation
in the heat transfer process.
2. f ′′ ( 0 ) takes positive as well as negative

values for certain values of the parameters.
Positive values of f ′′ ( 0 ) imply that the fluid
exerts a dragging force on the surface and
negative values imply the opposite.
3. Dual solutions exist for certain values of RP.

diminish as γ µ changes from 0 to -1, and Significant differences are observed between
takes smaller values in the absence of the upper and lower solutions. Ranges of values of
magnetic field. From figures 10,11 heat transfer RP for which unique solution or dual solutions
coefficient ( − θ ′(0) ) and slip velocity exist is observed to change considerably with
changing values of the parameters.
( f ′(0) ) can be seen to increase as γµ
changes from 0 to -1, and both the quantities ACKNOWLEDGEMENTS
assume larger values in the absence of magnetic
T.Raja Rani wishes to thank the
field.
authorities of Sri Vishnu Engineering College
V.COMPARISION WITH AVAILABLE
for Women and also authorities of S.R.K.R
RESULTS:
Engineering College for their encouragement
Results of the present analysis agree well with
and also for providing the facilities for
appropriate results of references[2] and [4]. In
figures 2 and 3 of our analysis are shown curves research. T.Raja Rani also conveys her thanks
for C = 1, λ =0 and γ µ =0 which coincide with to Mr.K.S.Sreenivasa babu of
S.R.K.R.Engineering College for his help in
those presented in references [2] and [4].
numerical computations.
Comparison of numerical results of our analysis
with those presented in table 2 of reference [4]
has revealed excellent agreement between the
20
ISSN:0974-8652
REFERENCES [10] White,F.M, Viscous Fluid Flow, McGraw-

Hill Inc., New York, 1974.
[1] Acharya.M, Dash.G.C and Singh.L.P.,
Magnetic field effects on the free convection [11] Vafai,K., Hand Book of Porous Media,2nd
and mass transfer flow through porous medium Ed, Taylor & Francis, New York.pp.379-
with constant suction and constant heat flux, 381,2005.
Indian J pure appl. Math., vol.31(1),pp.1-
18,2000.
[2] Aly,E.H, Elliot,L, Ingham,D.B, Mixed

convection boundary-layer flow over a vertical
surface embedded in a porous medium,
European Journal of Mechanics B/Fluids.,
vol.22, pp. 529-543, 2003.
[3] Barletta.A, Lazzari.S, Magyare.E and Pop.I,

Mixed convection with heating effects in a
vertical porous annulus with a radially varying
Magnetic field, Int. J. Heat. Transfer., vol.51,
pp.5777-5784, 2008.
[4] Chin,K,E, Nazar,R, Arifin,N.M, Pop,I,

Effect of Variable Viscosity on mixed
convection boundary layer flow over a vertical
surface embedded in a porous medium,
International Communications in Heat and Mass
Transfer., vol.34,pp. 464-473, 2007.
[5]Gorsan.T, Revnic.C, Pop.I and Ingham.D.B,

Magnetic field and internal heat generation
effects on the Free convection in a rectangular
cavity filled with a porous medium,
International Communications in Heat and Mass
Transfer, vol 52, pp.1525-1533,2009.
[6] Lai,F.C, and Kulacki,F.A,(1990), The Effect

of Variable Viscosity on Convective Heat
Transfer along a Vertical Surface in a Saturated
Porous medium, Int.J.Heat Mass Transfer.
vol.33, pp.1028- 1031,1990.
[7] Lakshmi Prasannam.V, Raja Rani,T and

C.N.B.Rao, Free convection in a porous
medium with magnetic field, variable physical
properties and varying wall temperature,
accepted for publication in International Journal
of Computational Mathematical ideas.
[8] Nield.D.A, Bejan.A, Convection in Porous

Media, Third ed., Springer, New York,2006.
[9] Sobha.V.V and Ramakrishna.K, Convective

Heat Transfer Past a Vertical Plate Embedded in
a Porous Medium with an Applied Magnetic
Field, IE(I) Journal-MC, 84, pp.133-134, 2003.
21
ISSN:0974-8652

MATHEMATICAL IDEAS
ISSN: 0974-8652 / www.ijcmi.webs.com / VOL 2-NO1-PP 22-27 (2010)
ON COMPLETELY PRIME AND COMPLETELY SEMI-PRIME

IDEALS IN Γ-NEAR-RINGS
Satyanarayana Bhavanari@, Pradeep Kumar T.V.#, Sreenadh Sridharamalle$,
Eswaraiah Setty Sriramula^
@
Department of Mathematics, Acharya Nagarjuna University, Nagarjuna Nagar-522
510,A.P. India. e-mail: bhavanari2002@yahoo.co.in
#
Department of Mathematics, ANU College of Engineering, Acharya Nagarjuna University
$
Department of Mathematics, S.V. University, Tirupathi, A.P. India.
^
Department of Mathematics, SGS College, Jaggaiahpet, Krishna DT., A.P., India.
ABSTRACT
In this paper we considered the algebraic system Γ-near-rings that was introduced by Satyanarayana.
“Γ-near-ring” is a more generalized system than both near-ring and gamma ring. The aim of this short
paper is to study and generalize some important results related to the concepts: completely prime and
completely semi-prime ideals, in Γ-near-rings. We included examples when ever necessary.
AMS Subject Classification: 16 D 25, 16 Y 30, and 16 Y 99

Key Words: Gamma near-ring, γ-ideals, γ-semi prime ideals
1. Introduction 1.1 Definition: An algebraic system (N, +, .) is

called a near-ring (or a right near-ring) if it
In recent decades interest has arisen in algebraic satisfies the following three conditions:
systems with binary operations addition and (i) (N, +) is a group (not necessarily
multiplication satisfying all the ring axioms Abelian);
except possibly one of the distributive laws and (ii) (N, .) is a semigroup; and
commutativity of addition. Such systems are (iii) (n1 + n2)n3 = n1n3 + n2n3 (right
called “Near-rings”. A natural example of a distributive law) for all n1, n2, n3 ∈ N.
near-ring is given by the set M(G) of all
mappings of an additive group G (not In general n.0 need not be equal to 0 for all n in
necessarily abelian) into itself with addition and N. If a near-ring N satisfies the property n.0 = 0
multiplication defined by for all n in N, then we say that N is a zero-
(f + g)(a) = f(a) + g(a); and symmetric near-ring.
(fg)(a) = f(g(a)) for all f, g ∈ M(G) and a ∈ G.
The concept Γ-ring, a generalization of ‘ring’ 1.2. Definitions: A normal subgroup I of (N, +)
was introduced by Nobusawa [ 4 ] and is said to be
generalized by Barnes [1]. Later, Satyanarayana (i) a left ideal of N if n(n1 + i) – nn1 ∈ I for all i
[8, 9], Satyanarayana, Pradeep Kumar & ∈ I and n, n1 ∈ N
Srinivasa Rao [14] also contributed to the (Equivalently, n(i + n1) – nn1 ∈ I for all i ∈ I
theory of Γ-rings. A generalization of both the and n, n1 ∈ N);
concepts near-ring and the Γ-ring, namely Γ-
near-ring was introduced and studied by (ii) a right ideal of N if IN ⊆ I; and
Satyanarayana [ 9, 11, 12 ], and later studied by
several authors like: Booth [2 ], Booth & (iii) an ideal if I is a left ideal and also a right
Groenewald [ 3], Syam Prasad [16]. ideal.
Now, we collect some existing fundamental If I is an ideal of N then we denote it by I ⊴ N.

definitions and results which are to be used in
later sections.
On Completely Prime and Completely Semi-prime Ideals in Γ-near-rings

22
ISSN:0974-8652
1.3. Definitions: (i) An ideal (left ideal) P of N f1g1f2 + f1g1f3. To see this, fix 0 ≠ z ∈ G and u ∈
(with P ≠ N) is said to be a prime (prime left) X. Define Gu: G → X by gu(x) = u for all x ∈
ideal of N if it satisfies the condition: I, J are G and fz:X → G by fz(x) = z for all x ∈ X. Now
ideals (left ideals) of N, IJ ⊆ P, implies I ⊆ P or for any two elements f2, f3 ∈ M, consider
J ⊆ P. fzgu(f2+ f3) and fzguf2 + fzguf3. For all x ∈ X,
[fzgu(f2+ f3)] (x) = fz[gu(f2(x) + f3(x))] = fz(u) = z
(ii) An ideal P of N is said to be completely and
prime if for any a, b ∈ N, ab ∈ P ⇒ a ∈ P or b [fzguf2 + fzguf3](x) = fzguf2(x) + fzguf3(x) = fz(u) +
∈P fz(u) = z + z.
(iii) An ideal S of N is said to be semi-prime if Since z ≠ 0, we have z ≠ z + z and hence
for any ideal I of N, I2 ⊆ S implies I ⊆ S. fzgu(f2+ f3) ≠ fzguf2 + fzguf3.
Thus we have that M is a Γ-near-ring which is
(iv) An ideal S of N is said to be completely not a Γ- ring.
semi-prime ideal if for any element a ∈ N, a2∈
S implies either a ∈ S. 1.7. Definition: Let M be a Γ-near-ring. Then a
normal subgroup I of (M, +) is called
1.4. Definitions: (i) For any proper ideal I of N, (i) a left ideal if aα(b + i) - aαb ∈ I for all a, b
the intersection of all prime(Completely Prime, ∈ M, α ∈ Γ and i ∈ I;
respectively) ideals of N containing I, is called (ii) a right ideal if iαa ∈ I for all a ∈ M, α ∈
the prime(Completely Prime, respectively) Γ, i ∈ I; and
radical of I and is denoted by P-rad(I) (C-rad(I) (iii) an ideal if it is both a left and a right ideal.
, respectively).
Let M be a Γ-Near-ring and α ∈ Γ.
(ii) The Prime (Completely Prime, respectively)
Satyanarayana [ 11 ] defined a binary operation
radical P-rad(0)(C-rad(0) , respectively) is also
“*α” on M by a *α b = aαb for all a, b ∈ M.
called as Prime (Completely Prime,
respectively) radical of N and we denote this by Then (M, +, *α) is a near-ring. So we may
P-rad(N) (C-rad(N), respectively). consider every element α ∈ Γ as a binary
operation on M such that (M, +, *α) is a
For some other fundamental definitions and near-ring. Also for any α, β ∈ Γ, we have (a *α
results, we refer Pilz [5], Satyanarayana [9, 13], b) *β c = a *α (b *β c) for all a, b, c ∈ M.
Satyanarayana and Syam Prasad [15].
Conversely, if (M, +) is a group and Γ is a set of
1.5. Definition: (Satyanarayna [9, 11, 12, 15]): binary operations on M satisfying
Let (M, +) be a group (not necessarily Abelian)
and Γ be a non-empty set. Then M is said to be (i) (M, +, * ) is a near-ring for all * ∈ Γ; and
a Γ-near-ring if there exists a mapping M × Γ ×
M → M (the image of (a, α, b) is denoted by (ii) (a *1 b) *2 c = a *1 (b *2 c) for all a, b, c ∈M
aαb), satisfying the following conditions: and for all *1, *2 ∈ Γ, then (M, +) is a Γ-near-
(i) (a + b)αc = aαc + bαc; and ring.
(ii) (aαb)βc = aα(bβc) for all a, b, c ∈ M and
α, β ∈ Γ. 1.8. Remark: (i) If *α, *β are operations on M
with a *α b = a *β b for all a, b ∈ M, then the
M is said to be a zero-symmetric Γ-near-ring if functions *α, *β are one and the same. So in this
aα0 = 0 for all a ∈ M and α ∈ Γ, where 0 is the case, we have *α = *β.
additive identity in M.
(ii) Suppose that (M, +) is a Γ-near-ring and
A natural example of Γ-near-ring is given also (M, +) is a Γ*-near-ring with the following
below: property: α ∈ Γ implies there exists β ∈ Γ* such
that a *α b = a *β b for all a, b ∈ M. Then we
1.6 Example (Satyanarayana [11]): Let (G, +) may consider this case as α = β and so Γ ⊆ Γ*.
be a non - abelian group and X be a non-empty
set. Let M = {f / f: X → G}. Then M is a group 1.9. Definition: Let (M, +) be a group. A Γ-
under point wise addition. near-ring M is said to be a maximal Γ-
near-ring if M cannot be a Γ*-near-ring for any
Since G is non-abelian, then (M, +) is non - Γ ⊂ Γ* (Here it is assumed that the restriction of
abelian. Let Γ be the set of all mappings of G the mapping M × Γ* × M → M to M × Γ × M is
into X. If f1, f2 ∈ M and g ∈ Γ, then, obviously, the mapping M × Γ × M → M).
f1gf2 ∈ M. But f1g1(f2 +f3) need not be equal to
23
ISSN:0974-8652
1. 10. Theorem (Th. 1.3 of Satyanarayana [ 11 1.14. Definition: Let I be an ideal of N. Then a
]): Let (M, +) be a group and P = {* / * is a prime (completely prime, respectively) ideal of
binary operation on M such that (M, +, *) is a N containing I is called a minimal prime
near-ring and M * M = M}. Then there exists a (minimal completely prime, respectively) ideal
partition {Γi / i ∈ I} of P such that (M, +) is a of I if P is minimal in the set of all prime
maximal Γi-near-ring for all i ∈ I. Conversely, if (completely prime, respectively) ideals
{Γj}j∈J be a disjoint collection of sets such that containing I.
(M, +) is a maximal Γj-near-ring for each j ∈ J
with M * M = M for all * ∈ Γj and for all j ∈ J, 1.15. Theorem ( Th. 1.4 of [ 13 ]): Let I be an
ideal of a near-ring N. Then I is a semi-prime
UΓ
j∈J
j ideal of a N ⇔ I is the intersection of all
then ⊆ P. Moreover (Say property B: minimal prime ideals of N ⇔ I is the
If Γ is a nonempty set such that (M, +) is a intersection of all prime ideals containing I.
maximal Γ-near-ring implies Γ = Γj for some j
∈ J). 1.16. Theorem (Cor. 5.1.10 of Satyanarayana [
9 ]) : Let N be a near-ring and A an ideal of N.
UΓ
j∈J
j Then A is completely semi-prime ideal if and
If property B holds, then = P. only if A is the intersection of completely prime
ideals of N containing A.
1.11. Definition: Let M be a Γ-near-ring and γ
∈ Γ. A subset A of M is said to be a γ-ideal of 1.17. Theorem (Theorem 2.2(b) of
the Γ-near-ring M if A is an ideal of the near- Satyanarayana [ 13]): An ideal P of N is prime
ring (M, +, *γ). and completely semi-prime ⇔ it is completely
prime.
1. 12. Observations: (i) Let (N, +, *) be a near-
ring which is not zero symmetric. Then there 1.18. Theorem (Lemma 2.7 of Satyanarayana [
exists a ∈ N such that a * 0 ≠ 0. Write Γ = {*}. 13 ]): Every minimal prime ideal P of a
Then N is a Γ-near-ring with aα0 ≠ 0 for some a completely semi-prime ideal I is completely
∈ N, α ∈ Γ. Therefore, in this case, N cannot prime. Moreover, P is minimal completely
be a zero symmetric Γ-near-ring. prime ideal of I.
(ii) Let M be a Γ-near-ring and (I, +) a normal
subgroup of (M, +). It is clear that I is an ideal 1.19.Theorem (Theorem 2.8 of Satyanarayana [
13 ]): Let I be a completely semi-prime ideal of
of the Γ-near-ring M if and only if I is an ideal
N. Then I is the intersection of all minimal
of the near-ring (M, +, *α) for all α ∈ Γ. In
completely prime ideals of I.
other words, I is an ideal of the Γ-near-ring M if
and only if I is a γ-ideal of M for all γ ∈ Γ. 1.20. Theorem (Theorem 2.9 of Satyanarayana
[ 13 ]): If P is a prime ideal and I is a
(iii) Let M be a Γ-near-ring. For any Γ* ⊆ Γ completely semi-prime ideal, then P is minimal
we have that M is a Γ*-near-ring. Every ideal I prime ideal of I if and only if P is minimal
of the Γ-near-ring M is also an ideal of Γ*-near- completely prime ideal of I.
ring M, but the converse need not be true.
1.21. Corollary: (Corollary 2.10 Satyanarayana
To see this, we observe the following example. [ 13 ] ): If I is a completely semi-prime ideal
of N, then I is the intersection of all completely
1. 13. Example: Consider G = {0, 1, …, 7} the prime ideals of N containing I.
group of integers modulo 8 and a set X = {a,
b}. Write M = {f / f: X → G such that f(a) = 0} 2. γ-Completely Prime and γ-Completely
= {fi / 0 ≤ i ≤ 7} where fi: X → G is defined Semi-prime γ-Ideals.
by fi(b) = i, fi(a) = 0 for 0 ≤ i ≤ 7. Consider
two mappings g0, g1 from G to X defined by Throughout this section we consider only zero-
g0(i) = a for all i ∈ G, and gi(i) = a if i ∉ {0, 3}, symmetric right near-rings, and M denotes a Γ-
g1(3) = g1(7) = b. near-ring.
Write Γ = {g0, g1} and Γ* = {g0}. Now M is a
Γ-near-ring and also Γ*-near-ring. 2.1 Definition: Let γ ∈ Γ. A γ-ideal I of M is
Now Y = {f0, f2, f4, f6} is an ideal of the Γ*- said to be
near-ring M but not an ideal of the
Γ-near-ring M (since f2 ∈ Y and f3g1(f1 + f2) - (i) γ-completely prime if a, b ∈ M, aγb ∈ I ⇒ a
f3g1f1 = f3 ∉ Y). ∈ I or b ∈ I.
24
ISSN:0974-8652
(ii) γ-completely semi-prime if a ∈ M, aγa ∈ I respect to γ ∈ Γ) if AγB ⊆ P for any two γ-

⇒ a ∈ I. ideals A, B of M implies A ⊆ P or B ⊆ P.
2.2 Note: Let M be a Γ-near-ring and γ ∈ Γ. (ii). A γ-ideal S of a Γ-near-ring M is said to be

Write N = M. Now (N, +, *γ) is a near-ring. a γ-semi-prime γ-ideal of M (with respect to γ
Let I be a γ-ideal of M. ∈ Γ) if AγA ⊆ S for any γ-ideal A of M implies
A ⊆ S.
(i) I is a γ-completely prime γ-ideal of M if and
only if I is a completely prime ideal of the near- 2.8 Note: Let P be an γ-ideal of a Γ-near-ring M
ring (N, +, *γ). and γ ∈ Γ. Then we have the following:
(ii) I is a γ-completely semi-prime γ-ideal of M (i). P is a γ-prime γ-ideal of the Γ-near-ring M

if and only if I is a completely semi-prime ideal ⇔ P is a prime ideal of the near-ring (M, +,
of the near-ring (N, +, *γ). *γ).
2.3 Remark: Every γ-completely prime γ-ideal (ii). P is a γ-semi-prime γ-ideal of the Γ-near-
of M is a γ-completely semi-prime γ-ideal of M. ring M ⇔ P is semi-prime ideal of the near-ring
[Verification: Let I be a γ-completely prime γ- (M, +, *γ).
ideal of M. Let a ∈ M. Suppose aγa ∈ I. Since
I is γ-completely prime, we have that a ∈ I. (iii).Suppose that S is a γ-ideal of M. Then (by
Thus I is a γ-completely semi prime γ-ideal of Theorem1.15) we have that S is γ-semi-prime γ-
M.] ideal of M ⇔ S is the intersection of all γ -
prime ideals P of M containing S.
2.4 Corollary: Let M be a Γ-near-ring, γ ∈ Γ
and A be a γ-ideal of M. Then A is γ- The following corollary follows from Theorem
completely semi-prime γ-ideal if and only if A 1.17.
is the intersection of γ-completely prime γ-
ideals of M containing A. 2.9 Corollary: A γ-ideal P of a Γ-near-ring M is
γ-prime and γ-completely semi-prime ⇔ it is
Proof: A is γ-completely semi-prime γ-ideal γ-completely prime.
⇔ A is completely semi-prime ideal of the
near-ring (M, +, *γ) (by Remark 2.3 ) ⇔ A is 2.10 Definitions: Let I be a γ-ideal of a Γ-near-
the intersection of all completely prime ideals of ring M for γ ∈ Γ.
the near-ring (M, +, *γ) containing A (by I is called a minimal γ-prime (γ-Completely
Theorem 1.16) ⇔ A is the intersection of all γ- Prime, respectively) γ-ideal of M if it is minimal
completely prime γ-ideals of M containing A. in the set of all γ-prime (γ-Completely Prime,
The proof is complete. respectively) γ-ideals containing I.
2.5 Definition: Let A be a proper ideal of M. The following corollary follows from Theorem
The intersection of all γ-completely prime γ- 1.18.
ideals of M containing A of M, is called as the
γ-completely prime radical of A and it is 2.11 Corollary: Let P be a γ-ideal of a Γ-near-
denoted by C-γ-rad(A). The γ-completely prime ring M for γ ∈ Γ. Every minimal γ-
radical of M is defined as the γ-completely prime γ-ideal P of a γ-completely semi-prime γ-
prime radical of the zero ideal, and it is denoted ideal I is a γ-completely prime γ-ideal. More
by C-γ-rad(M). over P is a minimal γ-completely prime γ-ideal
of I. The following corollary follows from
2.6 Note: From Theorem 1.16, and Theorem 2.4 Theorem 1.19.
we conclude the following:
(i) An ideal A of a near-ring is completely semi- 2.12 Corollary: Let γ ∈ Γ. If I is γ-completely
prime ⇔ A = C-rad(A). semi-prime γ-ideal of M, then I is the
intersection of all minimal γ-completely prime
(ii) A γ-ideal A of a Γ-near-ring M is γ- γ-ideals of I.
completely semi-prime ⇔ A = C-γ-rad(A).
2.13 Corollary: Let γ ∈ Γ and P be a γ-ideal of
2.7 Definitions: (i). A γ-ideal P of a Γ-near- M. If P is a γ-prime γ-ideal and I is a γ-
ring M is said to be a γ-prime γ-ideal of M (with completely semi-prime γ-ideal, then P is a
25
ISSN:0974-8652
minimal γ-prime γ-ideal of I if and only if P is a Proof: Write S = P-γ-rad(A).

minimal γ-completely prime γ-ideal of I. (i). Since S = P-γ-rad(A) is equal to the
Let γ ∈ Γ. By applying the Corollary 1.21 to intersection of all γ-prime γ-ideals of M
the near-ring (M, +, *γ) we get the following. containing S, by Note 2.8(iii), it follows that S
is a γ-semi-prime γ-ideal. Thus we conclude
2. 14 Corollary: Let γ ∈ Γ. If I is a γ- that the γ-prime radical of a γ-ideal A (that is, P-
completely semi-prime γ-ideal of M, then I is γ-rad(A)) is a γ-semi-prime γ-ideal.
the intersection of all γ-completely prime γ-
ideals of M containing I (that is, I = ∩ {P / P is (ii). Follows from (i), by taking A = (0).
a γ-completely prime γ-ideal of M such that I ⊆
M} = C-γ-rad(I)). ACKNOWLEDGEMENTS
The first author acknowledges the financial
2.15 Example: Let us consider the Example assistance from the UGC, New Delhi under the
2.11 of Satyanarayana [13]. In this example, grant F.No. 34-136/2008(SR), dt 30-12-2008.
(G, +) is the Klein four group where G = {0, a, The authors thank the referee for valuable
b, c}. We define multiplication on G as comments that improved the paper.
follows:
. 0 a b c REFERENCES
0 0 0 0 0 [1] Barnes W.E. “On the -rings of
a a a a a Nobusawa”,Pacific J. Math 18 (1966) 411- 422.
b 0 a b c
c a 0 c b [2] Booth G.L. “A note on Γ -Near- rings”,
Stud. Sci. Math. Hunger 23 (1988) 471-
This (G, +, .) is a near-ring which is not zero 475.
symmetric. The ideal {0, a} is only the
nontrivial ideal and also it is completely prime. [3]Booth G. L. & Groenewald N. J. “On
Radicals of -near-rings”, Math. Japan. 35
(i) Write M = G, the Klein four group and G = (1990) 417-425.
{0, a, b, c}. Define multiplication on G as
[4] Nobusawa “On a Generalization of the Ring
above. If we write Γ = {.}, then M is a Γ-near-
ring, which is not a zero symmetric theory”, Osaka J. Math. 1 (1964) 81-89.
Γ-near-ring (because aγ0 = a.0 ≠ 0). It is clear
[5] Pilz .G “Near-rings”, North Holland,
that for γ ∈ Γ, the γ-ideal {0, a} of M is only the 1983.
nontrivial γ-completely prime γ-ideal. The γ-
ideal (0) of M is γ-completely semi-prime γ- [6] Ramakotaiah Davuluri “Theory of Near-
ideal, but not γ-completely prime γ-ideal rings”, Ph.D. Diss., Andhra univ.,1968.
(because cγa = c.a = 0 and a ≠ 0 ≠ c). Hence
the γ-completely semi-prime γ-ideal (0) can not [7] Sambasivarao.V and Satyanarayana.Bh.
be written as the intersection of its minimal γ- “The Prime radical in near-rings”, Indian J.
completely prime γ-ideals. Pure and Appl. Math. 15(4) (1984) 361-364.
From this example 2.15, we can conclude that if
M is not a zero symmetric Γ-near-ring, then the [8] Satyanarayana Bh. "A Note on
corollary 2.14 need not be true. Γ-rings", Proceedings of the Japan
Academy 59-A (1983) 382-83.
2.16 Notation: Let A be a γ-ideal of M. The
intersection of all γ-prime ideals containing A is [9] Satyanarayana Bhavanari. “Contributions to
called the γ-prime radical of A and it is denoted Near-ring Theory”, VDM Verlag Dr Mullar,
by P-γ-rad(A). The γ-prime radical of M is Germany, 2010 (ISBN: 978-3-639-22417-7).
defined as the γ-prime radical of the zero ideal
(0). So P-γ-rad(M) = P-γ-rad(0). [10] Satyanarayana Bhavanari “A Note on g-
prime Radical in Gamma rings”,
2.17 Theorem: Let A be an ideal of M. Then Quaestiones Mathematicae, 12 (4) (1989)
415-423.
(i). P-γ-rad(A) is a γ-semi-prime γ-ideal.
[11] Satyanarayana Bhavanari. “A Note on Γ-
(ii). The γ-prime radical of M is a γ-semi-prime
near-rings”, Indian J. Mathematics
γ-ideal.
(B.N. Prasad Birth Centenary commemoration
volume) 41(1999) 427-433.
26
ISSN:0974-8652
[12] Satyanarayana Bhavanari "The f-prime

radical in Γ-near-rings", South-East Asian
Bulletin of Mathematics 23 (1999) 507-511.
[13] Satyanarayana Bhavanari “A Note on

Completely Semi-prime Ideals in Near-
rings”, International Journal of Computational
Mathematical Ideas, Vol.1, No.3 (2009) 107-
112.
[14] Satyanarayana Bhavanari, Pradeep Kumar

T.V. and Srinivasa Rao M. “On Prime left
ideals in Γ-rings”, Indian J. Pure & Appl.
Mathematics 31 (2000) 687-693.
[15] Satyanarayana Bhavanari & Syam Prasad

Kuncham “Discrete Mathematics and Graph
Theory”, Printice Hall of Inida, New Delhi,
2009.
[16] Syam Prasad K. “Contributions to Near-

ring Theory II”, Doctoral Dissertation
Acharya Nagarjuna University, 2000.
[17] Venkata Pradeep Kumar T. “Contributions

to Near-ring Theory III”, Doctoral
Dissertation, Acharya Nagarjuna University,
2008
27
ISSN:0974-8652

MATHEMATICAL IDEAS
A UNIFIED FRAME WORK FOR SEARCHING DIGITAL

LIBRARIES USING DOCUMENT CLUSTERING
Shaik Sagar Imambi@, Thatimakula Sudha#
@
Dept. of Computer Science, TJPS College, Guntur, A.P., India
#
Sri Padmavathi Mahila Univesity, Tirupathi, A.P., India
Abstract
The increasing interest in processing larger collections of documents from digital libraries has led to a
new emphasis on document clustering problem. Document clustering is a technique for identifying
clusters or groups of documents which share some common features or have overlapping content.
These groupings of documents can be useful in document retrieval from digital libraries. We have
developed the retrieval frame work for searching digital libraries, called UFDC. It is an advanced, an
efficient and effective search facility for digital information. It combines conventional information
retrieval and full-text searching techniques. Automatically linked Top Ranked Document(TRD)
clusters are generated from the digital library information . By grouping together TRDs that share a
common topic, UFDC provides an effective means of finding and tracking documents.
Keywords: Data mining, Digital library, Document clustering, UDFC frame work.
graph partitioning (Karypis, 2002), mixture of

INTRODUCTION vMFs (Banerjee et al., 2003), information
Digital Library systems are software systems bottle-neck (IB) clustering (Slonim and
which help in the management of metadata and Tishby, 2000), and co-clustering using
data. They also provide end-user services for bipartite spectral graph partitioning (Dhillon,
activities such as submission, discovery and 2001).
retrieval of digital objects. In most cases these One of the challenging research issues in
systems have developed out of the needs of Digital Libraries is the facilitation of efficient
libraries to manage digital equivalents of their and effective access to large amounts of
holdings. The existing generation of such available information. Digital library system
software tools – systems such as Greenstone architecture is shown in fig 1. Document
(Witten and Bainbridge, 2002), DSpace clustering [1] and automatic text
(Tansley, et al., 2003) and Eprints (University summarisation [2] are two methods which
of Southampton,2006) – have made it possible have been used in the context of information
for non-programmers to easily set up and access in digital libraries.
manage a digital archive.
The increasing interest in processing larger

collections of documents from digital libraries
has led to a new emphasis on designing more
efficient and effective techniques, leading to an
explosion of diverse approaches to the
document clustering problem. Some of the
approaches are (multi-level) self-organizing
map (Kohonen et al., 2000), spherical k-means
(Dhillon and Modha, 2001), bisecting k-means
(Steinbach et al., 2000), mixture of
multinomials (Vaithyanathanand Dom, 2000; Figure 1. Digital library system
Meila and Heckerman, 2001), multi-level
A Unified frame work for searching Digital libraries Using Document Clustering
28
ISSN:0974-8652
1. DOCUMENT CLUSTERING clustering algorithms. The unified framework

Document clustering is defined as the for searching data from digital libraries is
automatic discovery of document based on the clustering data and models. In
clusters/groups in a document collection, short, the scope of this paper is to help users to
where the formed clusters have a high degree evaluate the quality and feasibility of using
of association (with regard to a given similarity cutting edge clustering methods implemented
measure) between its members. Members from for digital libraries. In order to achieve this, we
different clusters will have a low degree of designed a unified frame work for document
association (ref 5). Document clustering clustering. This frame work should be aimed at
generates groupings of potentially related clustering experiments on medium to large
documents. By taking into account inter scale digital library websites which are already
document relationships; users have the indexed .
possibility to discover documents that might
have otherwise been left unseen in the digital
Challenges
libraries.(ref 3). Document clusters,
effectively, reveal the structure of the Although commercial information retrieval
document space. systems utilizing existing clustering
algorithms, document clustering is far from a
This space however may not help users trivial or solved problem. The clustering
understand how their search terms relate to the process is filled with challenges like:
retrieved documents. Therefore, the
information space offered by document Selecting appropriate features of the
clusters to users is essentially not documents that should be used for clustering.
representative of their queries. Clusters are Selecting an appropriate similarity measure
represented as probabilistic models in a model between documents.
space that is conceptually separate from the Selecting an appropriate clustering method,
data space. For partitioned clustering, the view utilising the above similarity measure.
is conceptually similar to the Expectation Implementing the clustering algorithm in an
Maximization (EM) algorithm. For efficient way that makes it feasible in terms of
hierarchical clustering, the graph-based view required memory and CPU resources.
helps to visualize critical/important Finding ways of assessing the quality of the
distinctions between similarity-based performed clustering.
approaches and model-based approaches. Finding feasible ways of updating the
clustering if new documents are added to the
2. TEXT SUMMARIZATION collection.
Finding ways for applying the clustering to
improve the information retrieval task at hand.
Text summarization, in the context of
information access, offers short previews of
the contents of documents, so that users can
3. UDFC FRAMEWORK
make a more informed assessment of the
usefulness of the information without having
to refer to the full text of documents (ref 2 The proposed framework includes query
,4). Particular classes of summarisation generation view, generating top ranked
approaches, query-oriented or query-biased documents, view of Clustering Top Ranked
approaches, have proven effective in providing Document (TRD) and a view of hardware that
users with relevance clues (ref 4). Query- leads to several useful extensions.
biased summaries present to users textual parts The four main views of Unified frame work of
of documents (usually sentences) which highly Document clustering (UFDC).
match the user’s search terms. The
effectiveness of such summaries in the context
of interactive retrieval on the World Wide Web
has been verified by (ref 4).
Goal: In this paper we present Unified Frame

work for Clustering Documents (UFDC) by
comparing its effectiveness at providing access
to useful information. The framework also
suggests several useful variations of existing
29
ISSN:0974-8652
or query of each of the required documents.

Individual TRD are linked to the original
documents (or to representations of the original
documents, such as titles, summaries, etc.) in
which they occur, so that users can access the
original information. User interaction with
TRS clusters, individual documents and other
document representations can be monitored.
The information collected can be used to
recommend new documents to users, and to
select candidate terms to be added to the query
from the documents and clusters viewed. We
use a standard iterative clustering technique to
compute N clusters of documents. The N seeds
Figure 2. UFDC frame work for the initial cluster centers are obtained by a
full hierarchical clustering of the best-ranked
100 documents resulting from the query, in
Generating query TRS clusters. This type of implicit feedback
The overall objective of our approach is to has been used by (ref 6) in order to utilise
utilise information resulting from the information from the interaction of users with
interaction of users with in the personalized query based document summaries, and is
information space. Users can access effective in enabling users to access useful
documents, or other shorter representations of information.
documents such as titles and
query-biased summaries, by selecting D. Hardware view:
individual sentences. The hardware was designed in a modular
framework. This allows for more complicated
Generating TRD: operations to be created from smaller and
simpler operations. For instance, the cosine
distance module is created by linking together
Sentences within document will discuss query a controller, a dot product, and a normalization
terms in the context of the same (or similar) circuit.
topics. This can assist users in better
Data Source Pd Pg P Pc Balance
NG17-19 Overlapping 2998 15810 3 999 0.998
groups
Classic CACM/CISI 7094 41681 4 1774 0.323
Obacal OHSUM 11162 11465 10 1116 0.437
Klb Wabace 2340 21839 6 390 0.013
La1 LA times 3204 31472 6 534 0.290
La2 Latimes 3075 31472 6 1047 0.282
understanding the structure and the contents of
the information space which corresponds to the 4. EXPERIMENTAL RESULTS
top-retrieved documents. This may be We used the 20-digital libraries data and a
especially useful in cases where users have a number of datasets from the CLUTO toolkit2.
vague, not well-defined information need. These datasets provide a good representation
of different characteristics: number of
Clustering TRD view: documents ranges from 204 to 19949, number
of words from 5832 to 43586, number of
The main function of TRD clusters is to classes from 3 to 20. A summary of all the
provide effective access to retrieved datasets used in this paper is shown in Table 1.
documents by acting as an abstraction of the
query information space. Essentially, TRD Table 1 : Summary of text datasets(for each
clusters form a second level of abstraction, data set d is the total number of documents , g
where the first level corresponds to summaries the total number of words , K the number of
30
ISSN:0974-8652
classes, C the average number of documents architectures. A dynamic, temporal or

per class and Balance the state ratio of the permanent retrieval of the images between the
smallest classes to the large class nodes balances the workload. Moreover,
suitable operators for dynamic feature
For each dataset, we assess the results by extraction and similarity metrics are necessary.
locating the result clusters that are affiliated
with the user’s area of interest. We then ACKNOWLEDGEMENT
calculate precision for each located cluster
defined as the number of relevant documents We would like to express our thanks to
compared to the total number of documents in referees for valuable comments that improved
the cluster. We thereafter calculate the the paper.
combined recall of the clusters defined as the
number of relevant documents found in these REFERENCES
clusters compared with the total number of [1]. T. Heskes, ” Self-organizing maps,
relevant documents in the search result. For vector quantization, and mixture
each case, we have gone through the entire modeling.” IEEE Trans. Neural
search result and identified the top ranked Networks, 12(6):1299–1305,
documents that were relevant to the area of November 2001.
interest based on the way the search word was [2]. Young wang et al , “Document
used in the document. The result of this is a list clustering with semantic Analysis”
of relevant and irrelevant pages with duplicates ,Proceeding of 39th Hawai
removed. The times taken for searching same International conference on System
data set with the same given query, in normal Science,2006
search and in experimental search are tabled in [3]. T. S. Jaakkola and D. Haussler,
table2. “Exploiting generative models in
discriminative classifiers”, Advances
in Neural Information Processing
Systems, volume 11, pages 487–493.
MIT Press, 1999.
[4]. David M Blei Andrew Yng and
Michael I JD , “ Latent Dirichlet
allocation”, Journal of Machine
Learning Research 3, 993-1022 , Jan-
2003.
[5]. K. Jain, M. N. Murty, and P. J.
Flynn,”Data clustering: A review”,
ACM Computing Surveys, 31 (3):
Table 2: Normal and Experimental Search 1999: 264–323,
component Times in Seconds [6]. Tapher H Haveliwala, “Topic
Sensitive page Rank – A context
Conclusion: Document clustering has been sensitive ranking algorithms for web
studied intensively because of its wide search” , IEE transactions on
applicability in areas such as web mining, Knowledge & Data Engineering
search engines, information retrieval, and 15(4)-784-796 ,2003.
topological analysis. Most traditional [7]. White, R.W., Ruthven, I., Jose, J.M, “
clustering methods do not satisfy the special A task-oriented study on the
requirements for document clustering. We influencing effects of query-biased
presented an unified frame work for document summarisation in web searching”.
clustering technique which can be Journal of Information Procsessing &
implemented in searching text from digital Management , 2003:350-362.
libraries. The novelty of this approach is that it [8]. Zamir, O., Etzioni, O, “ Web
exploits top ranked documents from digital document clustering: A feasibility
libraries; organize the cluster hierarchy, and demonstration”, In: Proceedings of
reducing the dimensionality of document sets. the 21st Annual ACM SIGIR
The experimental results show that our Conference, Melbourne, Australia
approach outperforms its competitors in terms (1998) 46–54.
of accuracy, efficiency, and speed. Feature
work includes the development of suitable
scheduling strategies for document cluster
31
ISSN:0974-8652
[9]. White, R.W., Ruthven, I., Jose, J.M, “

Finding relevant documents using top
ranking sentences: an evaluation of
two alternative schemes.”, In:
Proceedings of the 24th Annual ACM
SIGIR Conference, Tampere, Finland
(2002) 57–64
[10]. Radev, D.R., Jing, H., Budzikowska,
M, ” Centroid-based summarization
of multiple documents: sentence
extraction, utility-based evaluation,
and user studies”. In: Proceedings of
the ANLP/NAACL Workshop on
Summarization, Seattle, U.S.A.
(2000).
[11]. O. Kao, “Towards Cluster Based
Image Retrieval”, In Proceedings of
the International Conference on
Parallel and Distributed Processing
Techniques and Applications
(PDPTA), CSREA Press ,2000, pp
1307-1315,
[12]. Reuter, “Methods for parallel
execution of complex database
queries”, Journal of Parallel
Computing, Volume 25, 1999:pp
2177-2188,
[13]. Zha, H.,”Generic summarization and
key phrase extraction using mutual
reinforcement principle and sentence
clustering” , In: Proceedings of the
25th Annual ACM SIGIR
Conference, Tampere, Finland (2002)
113–120.
[14]. Kural, Y., Robertson, S.E., Jones,
S.,” Deciphering cluster
representations.” Information
Procsessing & Management, Volume
37 (2001) 593–601.
[15]. O. Kao, I. la Tendresse, “CLIMS -
A system for image retrieval by using
colour and wavelet features” ,
Proceedings of the First Biennial
International Conference on
Advances in Information System
(2004).
32
ISSN:0974-8652

MATHEMATICAL IDEAS
REDUCIBILITY FOR THE FIORINI-WILSON-FISK CONJECTURE

S.Satyanarayana@, J.Venkateswara Rao#, V.Amarendra Babu$
@
Department of Mathematics, Sri Venkateswara Institute of Science & Information Technology,
Tadepalligudem-534101, E-mail:s.satyans1@gmail.com
#
Professor of Mathematics, Mekelle University Main Campus, P.O.Box No.231, Mekelle, Ethiopia,
Email: venkatjonnalagadda@yahoo.co.in
$
Assistant Professor of Mathematics, Acharya Nagarjuna University,
Email:amarendravelisela@ymail.com
ABSTRACT
In this paper we define what we mean by reducibility for the Fiorini-Wilson-Fisk Conjecture and outline the
logic used to establish the reducibility of every configuration in U. Of course, a computer actually verifies the
reducibility of each configuration, as it would be too difficult using the present techniques to do so by hand.
Essentially, reducibility for the Fiorini-Wilson-Fisk Conjecture is just a strengthening of reducibility for the
Four Color Theorem, and in fact many of the configurations that were reducible for the Four Color Theorem
are also reducible for the Fiorini-Wilson-Fisk Conjecture.
Keywords:- Tricolorings, Contracts, Colorings of Rings, Reducibility.
Mathematics Subject Classification:- 05C15, 05C35, 05C90
1.1.1 Tri colorings and Notation Theorem 1.1.1 (Tait) Let G is a triangulation or
Recall that two functions c and c| with identical near-triangulation. The following statements are
domain and range={1, 2,…,K} are equivalent if equivalent,
{c-1({1}), c-1({2}),…, c-1({k})} = {c|-1({1}), c|- (i) The vertices of G can be 4-colored.
1
({2}),…, c|-1({k})}. We will use this frequently (ii) The drawing G has a tri coloring.
when the functions represent colorings. If A is a (iii) The dual of G can be edge-3-colored.
set of functions with domain D and range R =
{1,…, k}, then η (A) will denote the set of all 1.1.1.1 Tri colorings and Contracts
functions with domain D and range R that are
equivalent to some coloring in A. Let T be a The reducibility part of the recent Robertson et al.
triangulation or near-triangulation, and let F(T) proof of the Four Color Theorem essentially
denote the set of all faces of T that are bounded proceeds by induction. Without going into
by exactly 3 edges. A tri coloring of T is a details, the contraction of edges is critical in their
function c :F(T) → {-1, 0, 1} such that for every argument to produce smaller graphs. To avoid
f ∈ F(T), and for any two distinct edges I and j notational difficulties, they introduced the idea of
incident to f, c(i) ≠ c(j). The next theorem a tri coloring of T modulo X, where T is a
establishes a connection between Tri colorings triangulation or near-triangulation and X is a set
and vertex colorings of a graph and edge of edges in T. As the definition will reveal, the
colorings of the dual of the graph. set X represents the set of edges to be contracted.
Following the definitions of Robertson et al. [98],
a set X ⊂ E(T) is said to be sparse if no two
edges of X are incident to a common finite face
Reducibility For The Fiorini-Wilson-Fisk Conjecture
33
ISSN:0974-8652
of T, and if T is a near-triangulation, then no edge k (e) = 1 if {c(Vi), c(Vj)} = {1, 4} or {c(Vi),

of X is incident to the infinite face of T. If X is c(Vj)} = {2, 3}.
sparse, then a tri coloring of T modulo X is a Claim. k is a tri coloring of T modulo X. To
coloring k:E(T)- X → {-1, 0, 1} such that for prove this, let r be a triangular
every finite face r ∈ F(T) face in F(T) incident to the vertices {x, y, z} and
1.) If r does not have any edges in common with edges e = {x, y}, f = {x, z} and
X then k assigns distinct colors to the three edges g = {y, z}. If {e, f, g} I X = φ then x, y and z
of r. are in three distinct vertices of H,
2.) If r has exactly one edge in common with X, say x ∈ Vi, y ∈ Vj and z ∈ Vk. In the vertex-4-
then k(e) = k(f) for the other two edges e and f of coloring of H which defines k, Vi, Vjand Vk
r. receive different colors and thus k can be seen to
Recall that a counterexample is defined to be a assign difierent colors to e,f,and g. If one of the
planar graph which is not a vertex Fiorini- edges incident with r is in X, say g ∈ X, then the
Wilson-Fisk graph and has at most one vertex-4- vertices y andz are in the same vertex, say Vi of
coloring. A minimum counterexample is a H. If x were in the same component of T(X) as
counterexample with a minimum number of yor z, then since e, f ∉ X, there would be a
vertices. The following theorem, adapted from circuit C in T such that |E(C) - X| = 1.Since X is
[48], captures the idea that if one can contract sparse, x is in a distinct vertex. Hence, k (e) and k
edges in a minimum counterexample so that no (f) are well defined andequal to each other. This
loops are created, then the resulting graph has a completes the proof of the claim that k is a tri
tri coloring. coloring ofT modulo X and hence completes the
Theorem 1.1.2 Let T be a minimum proof of the theorem.
counterexample, and let X ⊂ E(T) be a
If X ⊂ E(T) is sparse and |E(C) - X| ≥ 2 for all
non-empty, sparse set such that there is no circuit
circuits C in T then we say
C of T for which |E(C) – X| = 1.
that X is contractible in T.
Then there is a tri coloring of T modulo X.
Proof. Let T(X) be the sub graph of T consisting
of the vertices of T and the edges 1.1.2 Colorings of a Ring
of X. Let V1,V2,…, Vp be the vertex sets of the
Let S be a free completion of a configuration K
components of T(X). Let H be the
with ring R. If c is a tri coloring
graph obtained by deleting multiple edges in the
of S, the restriction of c to the ring defines a
graph with vertex set {V1,…, Vp}
coloring of that ring. A basic part of
and with Vi adjacent to Vj if and only if there is
the theory of reducibility is the consideration of
an edge in E(T) − X which joins
these ring colorings. Let the vertices of R be
two vertices vi and vj with vi ∈ Vi and vj ∈ Vj .
1,2,…,r and the edges of R be e1,e2,…,er, where ei
Claim. H is loopless. If there was
has endpoints i and i + 1 for i = 1,…, r - 1 and er
a loop f joining the vertex Vi to itself then there
has endpoints r and 1. A coloring of R is a
would be an edge f’ ∈ E(T) - X
which joins two distinct vertices x, y of T that are function k : E(R) → {-1, 0, 1}. Let C*(R) denote
both in Vi. Since Vi is a vertex the set of colorings of R. We will sometimes
set of a component of the graph T(X), there is a abbreviate C*(R) by C*. By a restriction to R of a
path in T joining x and y and consisting entirely tri coloring c of S, we mean the function c|R with
domain E(R) and range {-1,0,1} that agrees with
of vertices in T. The circuit P U {f’} violates the
c on the edges of R. Also, if k . E(R) → {-1,0,1}
condition that |E(C) – X| ≠ 1 for every circuit C is a coloring of R, then we define an extension of
in T. Thus H is loopless. Since X is nonempty, p k into a tri coloring of S, to be a tri coloring of S
< |V (T)| and since T is a minimum which agrees with fi on E(R). We let C(S) denote
counterexample and H is loopless, H is either a the set of restrictions to R of Tri colorings of S. A
vertex Fiorini-Wilson-Fisk graph or H has at least restriction to R of a tri coloring of S has either
two vertex-4-colorings that are not permutations one or more extensions into Tri colorings of S.
of one another. Either way, H has a vertex-4- Let the set of restrictions to R of Tri colorings of
coloring c. Use the standard Tait coloring to S which have exactly one extension into a tri
define a coloring k . E(T) - X → {-1,0,1}, that is coloring of S be denoted by U(S) or just U if the
for an edge e of E(T) - X with endpoints u ∈ Vi free completion S is understood from the context.
and v ∈ Vj, define. If T is a triangulation that is uniquely vertex-4-
k (e) = -1 if {c(Vi), c(Vj)} = {1, 2} or {c(Vi), colorable, and if the free completion S appears in
c(Vj)} = {3, 4}. T then it follows that the restriction to R of the
k (e) = 0 if {c(Vi), c(Vj)} = {1, 3} or {c(Vi), corresponding tri coloring of T must be an
c(Vj)} = {2, 4}. element of U.
34
ISSN:0974-8652
The following definitions are taken from Therefore α ’ does not γ ’- fit k, because if it
Robertson et. al. [98] A match is a an unordered did, condition ii) of consistency would imply that
pair {e,f} of distinct edges of E(R). A matching is α ’ ∈ C. This establishes that C is A - critical and
a nonempty set of completes the proof of Lemma 1.1.1.
matches {{e1, f1},{e2,f2},…,{ek, fk}} such that for Consistency and criticality are defined in terms of
any i ≠ j, the edges ej and fj colorings of a circuit, but the near triangulations
are in the same component of R – {ei,fi}. Finally, to which we want to apply the ideas of
a signed matching is a collection consistency may have their infinite face bounded
of ordered pairs {({e1,f1}, µ 1), ({e2, f2}, by something other than a circuit. This does not
µ 2),…, ({ek, fk}, µ k)}, where the collection turn out to be a serious obstacle to using
{{e1, f1},{e2, f2},…,{ek, f k}} is a matching, and consistency as we shall now see. Let R be a
where µ i ∈ {-1, 1} for 1 ≤ i ≤ k.The sign of a circuit with vertices {1, 2,…, r} and edges e1,
e2,…, er where edge ei joins vertex i to vertex i +
match is used to differentiate whether both ends
of a kempe chain have the same color or distinct 1 for 1 ≤ i < r, and edge er joins vertex r to
colors. vertex 1. Let H be a near triangulation with outer-
facial walk W = v1, f1, v2, f2, v3,. . . , vr, fr, v1,
If θ ∈ {-1, 0, 1} and k is a coloring of R we say where v1, v2, . . . , vr are vertices, not necessarily
that k θ -fits a signed matching distinct and where {f1, f2,…, fr} are edges
M = {({e1, f1}, µ 1), ({e2, f2}, µ 2), … , ({ek, fk}, such that fi joins vi and vi+1 for 1 ≤ i ≤ r - 1 and
µ k)} if where fr joins the vertices vr and v1. Let φ . E(R)
(i) E(R) - U i ≤ i ≤ k {ei, fi} = {e ∈ E(R) . k (e) = → {f1, f2,…, fr} be defined by φ (ei) = fi. Also
θ } and suppose that k is a tri coloring of H and define a
(ii) For each ({ei, fi}, µ i) ∈ M, k (ei) = k (fi) if function λ on the edges of the circuit E(R) by
and only if µ i = 1. λ (e) = k( φ (e)). Following [98], we say that
A set C of colorings of R is consistent if for every φ wraps R around H and that the coloring λ of
k ∈ C and every θ , θ ’ ∈ {-1, 0, 1} E(R) is a lift of k.
there is a signed matching M such that The next theorem is an important result which
i) k θ -fits M. uses ideas of both Kempe and Birkhoff.
ii) C contains every coloring of R that θ ’-fits M. Theorem 1.1.3 Let H be a near triangulation with
Let A ⊂ c * . A set of colorings C of R is said to outer facial walk W as above,
and let φ wrap the circuit R around H. The set C
be A-critical if for every k ∈ C and every θ , θ ’
∈ {-1, 0, 1}, there is a signed matching M such of all lifts of Tri colorings of H is
that consistent.
i) k θ -fits M., Proof; Let k ∈ C and let θ ∈ {-1, 0, 1}. We
ii) C contains every edge that θ ’-fits M, and will construct a signed matching M = M(k, θ )
iii) there are not two colorings α , α ’ ∈ A and such that k θ -fits M. Following Robertson et. al.
integers γ , γ ’ ∈ {-1, 0, 1} such that both α γ - [98], we define a θ - rib to be a sequence
fits M and α ' γ ’-fits M and α is not equivalent g0,r1,g1,r2,…,rt,gt such that
(i) g0, g1, … , gt are distinct edges of H.
to α ’.
(ii) r1,r2,. . . ,rt are distinct finite faces of H.
(iii) If t > 0 then g0,gt are both incident with the
Lemma 1.1.1 If |A I C| ≤ 1, then C is A - infinite face of H, and if t = 0 then g0 is incident
critical if and only if C is consistent. with no finite face of H.
(iv) For 1 ≤ i ≤ t, ri is incident with gi-1 and with
Proof; If C is A - critical, then it is clearly gi.
consistent. Conversely, let C be a consistent set.
(v) For 0 ≤ i ≤ t, k(gi) ≠ θ .
We must show that under the hypothesis, C is
Any two distinct θ -ribs ρ = g0, r1,g1,r2,g3,…,gt
critical. Let k ∈ C and θ ∈ {-1, 0, 1}. Since C
is consistent, there is a signed matching M such and ρ ’ = g’0,r’1,g’1,…,r’t’ ,g’t’ must have {g0, g1,
that k θ -fits M and C contains every coloring . . . , gt} I {g’0, g’1,… , g’t’ }= θ , and {r0, r1,…,
that θ ’- fits M. Now let α , α ’ ∈ A, and let, rt} I {r’0, r’1, …, r’t’} = θ . To see this, note two
γ , γ ' ∈ {-1, 0, 1}. Since |A I C| ≤ 1, it things. First, if ρ and ρ ’ share a common non-
follows that one of either α or α ’ is not in C. infinite face r, then ρ and ρ ’ must also share
Without loss of generality, assume that α ’ ∉ C. both of the unique (because each of the finite
35
ISSN:0974-8652
faces is a triangle) edges incident to r that are si+1} = {sj , sj+1}, it follows that the edge ej-1 is in
colored with the colors in {-1, 0, 1} – { θ }, one of the faces si or si+1. Therefore ej ∈ {ei-1, ei,
because of (iv). Second, if ρ and ρ ’ share a ei+1}. Now the choice of j insures that ej-1 ∉{e0,
common edge g then ρ and ρ ’ also share both e1, . . . , ej-2}. Thus, it must be that i + 1 = j-1, so j
= i + 2 and si+1 = sj-1. Thus ei = ej is incident to si,
of the faces that are incident to g, because of (iv) si+1 = sj-1 and si+2 = sj and since each edge is
and (v). Using these two facts, we can show that incident to at most two faces, two of the three
if any two θ - ribs share either an edge or a finite faces si,si+1 and si+2 must actually be the same
face, then the two θ - ribs are identical. face. By the construction, si ≠ si+1 which forces si
Because of (iii), unless a rib consists of a single = si+2. But then ej-1 = ei-1 which contradicts the
edge, it contains at least two edges incident to the choice of j. Thus, every edge of the sequence is
infinite face. Because of (ii) and (iv), a rib does distinct. Since the graph is finite, this means that
not contain more than two edges which are the sequence must terminate on an edge ek other
incident to an infinite face. Thus, if a rib is not a than e0 which is incident to the infinite face. The
single edge, then it has exactly two edges that are sequence e0,s1,e1,s2,. . .,sk,ek is a rib which
incident to the infinite face and colored with contains e0, as desired.
colors in {-1, 0, 1} - { θ }. This shows that each θ - rib ρ defines a pair of
Conversely, we claim every edge that is incident edges {e ρ , f ρ } which are both incident to the
to the infinite face and is colored with a color in infinite face and which both receive colors from
{-1, 0, 1} – { θ } is in some rib. To see this, let e0 {-1, 0, 1} - { θ }. We can thus use ρ to define a
be an edge incident to the infinite face which is signed match, namely ({e ρ , f ρ }, µp ) where
colored α ∈ {-1, 0, 1}-{ θ } and let γ ∈ {-1,
µp = -1 if k(e ρ ) ≠ k (f ρ ) and µp = 1
0, 1} - { θ , α }. If e0 is not incident to a finite otherwise. Now we will show the set of ribs
face, then e0 is itself a rib by (iii). So suppose that { ρ 1 ,… ρp } defines a signed matching. First of
e0 is incident to a unique finite face s1 that is a
all, the set { ρ 1 ,… ρp } defines a set of signed
triangle. Because the given coloring is a tri
coloring, s1 has exactly one edge e1 ≠ e0, which matches M = M(k, θ ) = {({e p1 , f p1 }, µp1 ),…,
receives the color . This edge is incident to a face ({e ρp , f ρp }, µρp )} in the manner defined
s2 ≠ s1. If s2 is the infinite face, then e0,s1,e1 is a above. Because of the planarity of the graph, and
rib and we have proven that e0 is in a rib. If s1 is the fact that two ribs are either disjoint or
not an infinite face, then because the coloring is a
identical, it must be that for every i ≠ j, 1 ≤ i, j
tri coloring and because s1 is a triangle, there is
an edge e2 ≠ e1 which receives the color θ and
≤ p, 1) {e ρi ,f ρi } I {e ρj , f ρj } = θ , and 2) the
is incident to a face s3 ≠ s2. In this way we removal of {e ρi ,f ρi } from the ring could not
generate an alternating sequence of edges and separate e ρj from f ρj . Thus M is a signed
faces e0,s1,e1,s2,. . .. If the sequence ever selects matching. Now we show that k θ - fits M. First,
an infinite face sk+1 it terminates. We now show because every edge incident to the infinite face
that the construction of this sequence guarantees and receiving a color in {-1, 0, 1} – { θ } must be
that all of the edges e0, e1,…, are distinct. If not,
in a rib, it follows that {e ρ 1 ,…, e ρp , f ρ 1 ,…,
there would be integers 0 ≤ i < j such that ei = ej
f ρp } equals {f ∈ E(R) .k(f) ∈ {-1, 0, 1} –
. Of all such pairs (i, j) choose one with the
smallest j and subject to that choose among those { θ }}. The definition of µρi also shows that for
the one with the largest i. As a first case, assume i every integer i (1 ≤ i ≤ p), k(e ρi ) = k (f ρi ) if
= 0. Thus ej is incident to the infinite face. It
and only if µρi = 1. This proves that k θ - fits M
follows that s1 = sj and e1 = ej-1. By the choice of
j, it cannot be that ej-1 ∈ {e0, e1, . . . , ej-2} and so = M(k).
j - 2 = 0. It follows that s1 = sj = s2. This however, We now finish the proof that C is consistent.
contradicts the construction. Thus we have shown First, for every k ∈ C and every θ ∈ {-1, 0, 1},
that i > 0, and in addition that if e0,…, ej-1 are our construction using ribs has produced a signed
distinct and ej is incident to the infinite face, then matching M = M(k, θ ) such that k θ - fits M.
e j ≠ e 0. So let θ ’ ∈ {-1, 0, 1} and let k’ be another
Of the two faces incident to ei, the face si-1
precedes the face si+1 is the sequence. Similarly, coloring that θ ’- fits M(k, θ ) = M. Define the
the face sj precedes the face sj+1 and both are coloring k’’ as follows. k ‘’(e) = θ if k ‘(e) =
incident to ei = ej . Clearly {si, si+1} = {sj, sj+1}. θ ’, k ‘’(e) = θ ’ if k ‘(e) = θ and k ‘’(e) = k ‘(e)
The face sj is incident to an edge ej-1 ≠ ej that if k ‘(e) 2 {-1, 0, 1} – { θ , θ ’}. It follows that k
receives a color in {-1, 0, 1} - { θ }. Since {si, ‘’ θ - fits M. Let c be the coloring of H whose lift
36
ISSN:0974-8652
is k and let { ρ 1 ,… ρp } be the set of θ - ribs Let k ∈ C1 U C2 and let θ ∈ {-1, 0, 1}.
induced by c which define M. If ({ei, fi}, µ i) is Without loss of generality, we may assume k ∈
the signed match associated with the θ - rib ρi , C1. Therefore there is a matching M that k θ - fits
and such that every other coloring k’ which θ ’-
then the fact that k’’ θ ’- fits implies that
1a) Either k ‘’(ei) = k ‘’(fi) ≠ k (ei) or fits M is in C1 ⊂ C1 U C2.
1b) k ‘’(ei) = k ‘’(fi) = k (ei) or Now suppose by way of contradiction that there
1c) k ‘’(ei) ≠ k ‘’(fi), and k ‘’(ei) ≠ k (ei) or are two non-equivalent colorings α 1, α 2 and
1d) k ‘’(ei) ≠ k ‘’(fi), and k ‘’(ei) = k (ei). two integers λ 1, λ 2 ∈ {-1, 0, 1} such that
where either 1a) or 1b) hold if µ i = 1 and either α 1 λ 1 fits M and α 2 λ 2- fits M. This however
1c) or 1d) hold of µ i = -1. In the θ - rib ρi = violates the A - criticality of either C1 or C2.
g0,r1,g1,r2,…,rt,gt, we have k (g0) = k (g2) = k (g4) This proves that C1 U C2 is A - critical. If |A| ≤

= … = k (g 2 2t ) = α and k (g1) = k (g3) = k 1 the set C is A - critical if and only if it is
consistent and so by choosing A = φ , we deduce
(g1) = … = k ( g 2 t −21  + 1 ) = β where { α , β } that the union of two consistent sets is consistent.
= {-1, 0, 1} – { θ }. There is another tri coloring Now let B ⊂ C*(R). The union of all A - critical
γ of H that can be obtained by exchanging the subsets of B is A - critical and certainly contains
every A - critical subset of B. This completes the
colors α and β along H, namely γ (e) = k (e) proof of Lemma 1.1.2.
if e is not in ρi , γ (e0) = γ (e2) = γ (e4) =… = Lemma 1.1.3 If |A I C| ≤ 1, then the maximal
 
γ ( e 2  2t  ) = β and γ (g1) = γ (g3) = γ (g1) = consistent subset of C equals the maximal A -
critical subset of C.
… = γ ( g 2 t −21  + 1 ) = α . Using this idea, we
Proof: Let MCSA(C) denote the maximal A -
define a new tri coloring c’’ of H by exchanging critical subset of C and let MCS φ (C) denote the
the colors α , β ∈ {-1, 0, 1} – { θ } along each maximal consistent subset of C. We know that
rib ρi for which either 1a) or 1c) holds. The lift MCSA(C) is consistent so MCSA(C) ⊂
of c’’ will be k’’ and thus k’’ ∈ C. Moreover, by MCS φ (C). Also, by Lemma 1.1.1 and the fact
defining a coloring c’ of H from the coloring c’’ that |CA| ≤ 1, MCS φ ,(C) is A - critical. Hence
by swapping the colors θ and θ ’, that is MCS φ (C) ⊂ MCSA(C), and thus the theorem
defining c’(x) = θ if c’’(x) = θ ’, c’(x) = θ ’ if holds.
c’’(x) = θ and c’(x) = c’’(x) otherwise, we see 1.2 Proving Reducibility
that c’ is also tri coloring of H whose lift equals 1.2.1 Using a Corresponding Projection
k’. Thus k’ ∈ C as desired. Let K be a configuration that appears in a
This shows that C is consistent and completes the triangulation T and has free completion S and
proof of Theorem 1.1.3. ring R. In general S will not appear in T, but
Lemma 1.1.2 Let R be a ring and let A ⊂ suppose for illustration that it does. The ring R
C*(R). The empty set is an A – critical set. Also, will naturally split T up into two near
the union of two A - critical sets is an A - critical triangulations, one of them S and the other which
set and in particular, the union of two consistent we denote by H. However, it may be the case that
sets is consistent. Finally, for any subset B of S does not appear in T. The next lemma is a
colorings of R, the maximally A - critical subset technical result to show that we may still in a
of B exists, that is, there is a subset of B which is certain sense decompose T into the near
A - critical, and such that every other A - critical triangulation H and the free completion S.
subset of B is contained in it. Lemma 1.2.1 Let K be a configuration which
Proof: Let A ⊂ C*(R). The statement that the appears in a triangulation T and has free
empty set is A - critical is vacuously true. Let C1, completion S with ring R and let φ be the
C2 ⊂ C*(R) be two A - critical sets, and assume corresponding projection of S into T. Let H be
that for i = 1, 2, and any θ , θ ’, γ , γ ’ ∈ {-1, 0, the graph obtained from T be deleting the vertex-
1} and any k ∈ Ci, there is a signed matching M set φ (V (G(K))). Then
such that 1) H is a near triangulation and φ wraps R
i) k θ - fits M and around H.
ii) every k’ that θ ‘- fits M is in Ci and 2) If X ⊂ E(S) is sparse in S, then φ (X) is
iii) no two non-equivalent colorings α i ∈ A γ I sparse in T.
- fit M for both i = 1 and i = 2. Proof; Since φ fixes G(K) and G(K) is
connected, all of G(K) lies in the same face of the
37
ISSN:0974-8652
drawing T -V (G(K)). This and the fact that T is a 3) The edges φ (gi), φ (hi) and φ (gi-1) are all
triangulation implies that H = T - φ (V (G(K))) = incident to the finite face ri in T, for 1 ≤ i ≤ p.
T - V (G(K)) is a near triangulation. All of this implies that φ (g0), φ (g1),…, φ (gp)
Let V (R) = {r1, r2,…, rq} and E(R) = {e1, e2,…, is a portion of any clockwise listing in T of edges
eq} where for i = 1, 2, . . . , q - 1, ei has endpoints
incident to φ (v) in T. Moreover, from the claim
ri and ri+1 and eq = {rq, r1}. Suppose that r1,r2,…,rq
is the clockwise order of appearance of the above, φ (ri) is a face of T that is contained in
vertices of V (R). Consider the alternating the infinite face of H. Hence, the edges φ (g1),
sequence W of vertices and edges in T . φ (r1), φ (g2),…, φ (gp-1) are all edges of the infinite
φ (e1), φ (r2), φ (e2),…, φ (rq), φ (eq), φ (r1). face of H, and it therefore follows that φ (gp)
By property (iii) of projections, φ (ri) is incident follows φ (g0) in any clockwise listing in H of
to φ (ei) in T for each i ∈ 1,…, q - 1 because ri is edges incident to φ (v).
incident to ei in S for each i ∈ 1,…, q - 1. For the We now show that the sequence φ (r1), φ (e1),
same reason φ (eq) is incident to φ (r1). Thus W
φ (r2), φ (e2),…, φ (rq), φ (eq), φ (r1) is a
is a closed walk in H.
We now prove some things that will help in facial walk in H that bounds the infinite face.
establishing that W is a facial walk. We claim From what we have just shown, φ (ei-1) follows
that every finite face r ∈ F(S) - F(G) has the φ (ei) in the clockwise listing in T of edges
property that φ (r) is contained in the infinite incident to φ (ri) for 1 ≤ i ≤ q (and where when
face of H. It suffices to show that φ (r) is i = 1, we interpret ei-1 as er). This completes the
incident to some vertex of V (G), because the proof that W is a facial walk of the infinite face
infinite face of H will contain that vertex in its of G and thus shows that φ wraps R around H.
interior. By property 2) of known Lemma r is Now let X ⊂ E(S) be a sparse set of edges. Each
incident to a vertex x ∈ V (G). By property (iii) edge in X must have at least one endpoint in V
of projections, φ (r) is incident to φ (x) = x in T. (G). Therefore, every edge of X is in the domain
This proves that claim. of the natural edge function of Lemma 4.11.
Let u ∈ V (R). By property 6) of Lemma 4.2.1, a Since φ is an extension of the natural edge
clockwise listing in S of edges incident to u is g0, function, φ preserves that embedding at every x
g1,…, gp where g0, gp ∈ E(R), p ≤ 2 and g1,…gp-
∈ V (G), which implies that if e, f ∈ X share a
1 2 E(G). We assume that the endpoints of gi in V
common endpoint x ∈ V (G), but do not share a
(S) - v are xi for 0 ≤ i ≤ p. Also, for 1 ≤ i ≤ p,
common face, then φ (e) and φ (f) have
we label the unique finite face of S that is
incident to gi-1 and gi as ri and the unique edge in common endpoint x in T but are not in a common
E(S) that is incident to ri as hi. Thus, for 1 ≤ i ≤ face of T. This implies that if φ (e) and φ (f) are
p - 1, hi+1, gi, hi is a portion of a clockwise listing in the same face r’ of T for some distinct edges e,
in S of edges incident to xi. From the fact that φ f ∈ X, then φ (e) and φ (f) do not have a
is the extension of the natural edge function, and common endpoint in V (G). However, φ (e) and
the fact that the natural edge function preserves φ (f) both have endpoints in V (G) since e and f
the embedding in S at xi, it follows that φ (hi+1), do. Let xe denote the endpoint of e in V (G) and
φ (gi), φ (hi) is a portion of the clockwise listing xf the endpoint of f in V (G), and let z be the
in T of edges incident to φ (xi) = xi for 1 ≤ i ≤ common endpoint of φ (e) and φ (f) in T.
p - 1. Also, the fact that ≤ is a projection, implies We digress briefly to show that there is an r ∈
that φ (ri) is a face of T that is incident to the F(S) such that φ (r) = r’. Now r’ is adjacent to a
vertex in V (G), (in fact two, xe and xf ). Property
edges fi(gi-1), fi(hi), fi(gi) for 1 ≤ i ≤ p. We
4) of the natural edge function guarantees that φ
claim that for every 1 ≤ i ≤ p, φ (gi) follows φ
restricted to the edges of S which have at least
(gi-1) in the clockwise listing in T of edges
one endpoint in V (G) is a one to one and onto
incident to φ (v). This however follows from function into the set of edges T with at least one
three facts. endpoint in V (G). From the way φ was defined
1) φ (hi) follows φ (gi) in any clockwise listing
for faces, this implies that φ (r) = r’.
in T of edges incident to xi for 1 ≤ i ≤ p. Let g be the edge in S which has endpoints xe
2) φ (gi-1) follows φ (hi) in any clockwise listing and xf and which is incident to the face r. Because
in T of edges incident to xi-1 for 1 ≤ i ≤ p. φ is a projection, φ (g) is incident to r’. Also,
38
ISSN:0974-8652
since G is an induced sub drawing of S, g ∈ such that |X| = k, φ (X) is contractible and no tri
E(G) and so φ (g) = g. Suppose without loss of coloring of S modulo X is in the
generality that g follows φ (e) in every set MCS φ (C* - CS).
clockwise listing in T of edges incident to xe, and 3. If u ∈ U and u ∉ MCS(u) then we say that u
that φ (f) follows g in every clockwise listing in is D-removable.
T of edges incident to xf . Now the edge e either 4. If u ∈ U and there is a sparse set X ⊂ E(S)
precedes or follows the edge g in any clockwise such that φ (X) is contractible and
listing in S of edges incident to xe. If e follows g MCS(u) I CS(X) = φ , then we say that u is C-
in every clockwise listing in S of edges incident removable.
to xe then φ (e) would both precede and follow 1. A configuration K is U-reducible if
φ (g) in every clockwise listing in T of edges 1) every u ∈ U, is either D-removable or C-
incident to xe. This however is impossible removable.
because dT (xe) = γ K(xe) ≥ 1. Therefore, it must 2) At least one u ∈ U is C-removable or the
configuration K is either D-reducible
be the case that e precedes g in any clockwise or C(4)-reducible.
listing in S of edges incident to xe. For similar Notice that each of the above types of reducibility
reasons, f must follow g in any clockwise listing depends only on R and S and not on H. This has
in S of edges incident to xf . Thus e and f share a the very practical application that if /V (R)/ is
triangular face with each other and with g, which relatively small, say 14 or 11, it is feasible
contradicts that X is sparse. This completes the computationally to calculate Maximum Critical
proof of the lemma. Subsets like MCS(u). This coupled with the
1.2.2 Defining Various Types of Reducibility following observations which relate CH to
We now introduce the notation that will be used MCS(u) are the key to reducibility because
for the rest of this chapter. Let K be a calculations on a small piece of the triangulation
configuration with free completion S and ring R. T yield information about the rest of the
Suppose that K appears in the triangulation T, triangulation which could be immense. Let us
that φ is a corresponding projection of S into T, recall that the operator fi was defined at the
that H is the near triangulation T - V (K(G)), and beginning of the chapter.
that φ wraps R around the outer facial walk of 1) If it is not the case that CH I CS ⊄ η ({u})
H. Finally, let X be a sparse subset of E(S). We for some u ∈ U, then it can be
now define various sets of colorings of R. Let C* shown that T has at least two vertex-4-colorings.
be the set of all colorings of the ring R, let CS be 2) If CH I CS ⊂ η ({u}) for some u ∈ U, then
the set of restrictions to R of Tri colorings of S,
CH ⊂ MCS(u) by Lemma 1.1.3.
and let U ⊂ C be the set of colorings of R which
Thus if T is a minimum counterexample, then CH
extend to a unique tri coloring of S. Note that
fi MCS(u). This will turn out to be valuable
C*(R) - CS is the set of colorings of
because the induction hypothesis can be used to
R which do not extend into S. The set CS(X) will
color H.
denote the set of restrictions to R of tri coloring
Robertson et al. used D-reducibility and C(k)-
of S modulo X. Also, let CH denote the set of lifts
of Tri colorings of H. By Lemma 1.1.2, for any B reducibility for 1 ≤ k ≤ 4 to prove the Four
⊂ C, there is a maximal U - critical subset of B Color Theorem [98]. Notice that reducibility for
which we denote by MCSU (B) or just MCS(B) the Four Color Theorem (Types 1. and 2.) is
for short. The notation MCS φ (B) will denote the defined for entire configurations while
reducibility for the Fiorini- Wilson-Fisk
maximal consistent subset of B. Finally, for u ∈
Conjecture must first be defined in terms of
U we denote the set MCSU ((C* - CS) U {u}) by individual colors in U (types 3. and 4.) and only
MCS(u). With these definitions in place, we now then defined for an entire configuration (type 1.)
define various types of reducibility, the first two This means that proving reducibility for the
of which appear in the literature and are suficient Fiorini-Wilson-Fisk conjecture will tend to be
to prove the Four Color Theorem, and the third, more computationally intensive than proving it
fourth and fifth of which are introduced to prove for the Four Color Theorem because in principle,
the Fiorini-Wilson-Fisk Conjecture. each color in U needs to be considered.
1. The configuration K is D-reducible if MCS φ 1.2.3 Proving Reducibility
(C* - CS) = φ . As noted above, S will not, in general, appear in
2. The configuration K is C(k)-reducible if there T, but suppose again for illustration that it does.
exists a sparse set set X ⊂ E(S) The ring R will appear in T and will naturally
split T up into two near triangulations, one of
them S and the other which we denote by H.
39
ISSN:0974-8652
Denoting by CS the set of restrictions to R of tri I {e, f, g} ≠ φ , say e ∈ X, then

cX(g)}. If X
coloring of S and CH the set of restrictions to R of
Tri colorings of H, it is clear that T will have a tri
φ (e) ∈ φ (X), so cX(f) = c( φ (f)) = c( φ (g)) =
cX(g). Thus cX is a tricoloring of S modulo X.
coloring if and only if CS I CH ≠ φ . Many of
From the definitions of cX and cH, cX(e) =
the results of this section will use this simple c( φ (e)) = cH( φ (e)) for e ∈ E(R). Hence the
principle in one way or another. The next
theorem proves that this simple principle can be restriction of cX to R (which equals the restriction
applied even when only a projection φ of S of cH to φ (E(R)) is in the set CS I CH. This
appears in T. completes the proof of Lemma 1.2.4
Lemma 1.2.2 Let d be a coloring of R. Then d ∈ Let A ⊂ C*. Generalizing Robertson et al. we
say that a set X ⊂ E(S) - E(R) is an A - contract
CS I CH if and only if T has a
if it is a nonempty, sparse set and if no tricoloring
tri coloring whose restriction to φ (R) is d.
modulo X of S is in the set MCS( (C* - C) U A
The proof of this is straightforward and we omit ). If A = {a} we call an A-contract simply an a -
it.
contract. If A = φ , then we say that X is a
The usefulness of our definitions of reducibility
hinge on the following lemma, as contract.
was alluded to in Section 4.2.2. The free completion S of a configuration does not
Lemma 1.2.3 Either T has at least two non- necessarily appear in the triangulation T even if
equivalent vertex-4-colorings or there is a u ∈ U the configuration does. However, Theorem 4.3.1
such that CH ⊂ MCS(u). shows that there is a projection of S into T. It is
conceivable that a contract X in S might produce
Proof. By Lemma 1.2.2, if |CH I CS| ≥ 2, or if
loops if the corresponding edges were contracted
CH I (CS - U) ≠ φ , then T has at in T and Theorem 1.1.2 would not be applicable.
least two distinct vertex-4-colorings. Hence, we The following method of Robertson et al. gives
may assume there is a u ∈ U such that CH ⊂ an easy to check sufficient condition for a
(C* - CS) S U ({u}). Now MCS(u) ⊂ (C* - CS) contract X ∈ S not to produce loops after being
projected into T.
U η ({u}) and also MCS(u) is consistent by
An edge e is said to face a vertex v if v is not an
Lemma 1.1.3. Thus Theorem 1.1.3 implies that endpoint of e and both v and e are incident to a
CH ⊂ MCS(u) since CH is consistent. common face. A vertex v ∈ V (S) is a triad for X
Lemma 1.2.4 Let φ be a corresponding if
projection of S into T. If c is a tricoloring of T (i) v ∈ V (G(K))
modulo φ (X), then there are functions cX and cH, (ii) There are at least three vertices of S adjacent
such that cX is a tricoloring of S modulo X and cH to v and incident to a member of X
is a tricoloring of H. In addition, cX(e) = c( φ (e)) (iii) If γ K(v) = 1, then there is an edge of X that
for every e ∈ E(S), and cH(e) = c(e) for all e ∈ does not face v.
E(H). Finally, the restriction of c to φ (E(R)), the
Theorems 1.2.1 Let K be a configuration with
restriction of cX to E(R) and the lift of cH by φ free completion S and ring R and suppose that K
are all the same ring coloring,and this ring appears in an internally 6 connected
coloring is in CH I CS. triangulation T. Let φ be a corresponding
Proof: Note that by Lemma 1.2.1, H is a near- projection of S into T and let X ⊂ E(S) be a
triangulation, φ wraps R around H and φ (X) is sparse subset with |X| = 4 such that there is a
vertex of G(K) which is a triad for X. Then for
sparse in T. Since E(H) I φ (X) = φ , c
every circuit C in T, |E(C) - φ (X)| ≥ 2 or there
restricted to H is a tricoloring of H, which we
henceforth denote cH. The tricoloring c also is a short circuit in T.
defines a tricoloring cX of S modulo X as follows.
cX(e) = c( φ (e)) for e ∈ E(S). By property (iii) Proof: Let Y = φ (X). Lemma 1.2.1 guarantees
of projections, if r ∈ F(S), and r is incident to the that Y is sparse in T. Let C be a circuit in T.
distinct edges e, f, g ∈ E(S) then φ (r) is a face Since T is loopless, |E(C)| > 1. If |E(C)| = 2 then
because all faces are triangles, C cannot bound a
in F(T) which is incident to edges the edges φ face and must therefore be a short circuit. If
(e), φ (f) and φ (g). If X T {e, f, g} = φ , then |E(C)| = 3 then C must bound a face, for
φ (X) I { φ (e), φ (f), φ (g)} = , so {1, 0, 1} otherwise it would be a short circuit. Thus, the
sparseness of Y implies that Y has at most one
= {c( φ (e)), c( φ (f)), c( φ (g))} = {cX(e), cX(f), edge in common with E(C) and so the desired
inequality holds. If |E(C)| = 4 and C = {x1, x2, x3,
40
ISSN:0974-8652
x4} then either C is a short circuit or some pair of Theorem 1.2.3 Let T be a minimum
diagonally opposite vertices of C, say x1 and x3 counterexample. Then no configuration
are adjacent to each other and {x1, x2, x3} and isomorphic to one in Appendix 1appears in T.
{x3, x4, x1} form triangular faces in T. Since Y is Proof: Let T be a minimum counterexample, and
sparse, Y has at most one edge in common which suppose that K is a configuration in Appendix 1
each of these two faces and so |E(C) I Y | ≥ 2 which appears in T. By Theorem 3.3.1, we know
and the inequality follows. If |E(C)| ≥ 6, then |X| that T is internally 6 - connected.
Let H and S be as at the beginning of Section
≥ 4 implies |E(C) - φ (X)| ≤ 2. So we may
1.2.2. We first notice that if CH I CS includes
assume |E(C)| = 1 and thus that |X| = 4. Let C =
x1, x2, x3, x4, x1. We may assume |E(C)-Y | = 1 two non-equivalent colorings, or if k ∈ CH I
and so all 4 edges of Y are in E(C). Let int(C) C S for some k ∈ CS -U, then Lemma 1.2.2
denote the sub drawing of T induced by the implies that T would have at least two non-
vertices in one of the arc-wise connected equivalent vertex-4- colorings. Therefore, we
components of Σ - C and let ext(C) denote the may assume that C H I C S ⊂ (C* - C S U
)
the sub drawing of T induced by the vertices in η ({u}), for some u ∈ U.
the other arc-wise connected component of Σ - Assume first that no u ∈ U is C-removable.
C.We may assume that either |V (int(C))| ≤ 1 or Therefore, every u ∈ U is D-removable, which
|V (ext(C))| ≤ 1, or else C is a short circuit. If |V implies that u ∉ MCS(u) for every u ∈ U. Since
(int(C))| = 0 or |V (ext(C))| = 0, then there are K appears in Appendix 1, Theorem 1.2.2 implies
only edges in one of the two disjoint regions of that K is either D-reducible or C-reducible. We
the sphere defined by C, but this will create first consider the case that K is D-reducible, and
triangular faces containing two edges of Y , a therefore that MCS (C* - C) = φ . Since T is a
violation of the sparseness of Y . Thus we must minimum counterexample, H has a vertex-4-
have |V (int(C))| = 1 or |V (ext(C))| = 1. By
coloring and thus CH ≠ φ ,. If CH I CS = φ ,
symmetry, we may assume the former and we
will let y denote the vertex for which V (int(C)) = then CH ⊂ (C* - CS) which implies that CH ⊂
{y}. Note that y has degree 1 and faces all the MCS(C* - CS) since Lemma 1.1.3 implies the
edges of Y and so cannot be a triad for Y . Since latter is consistent. This however, is a
there is a triad v for Y , v ∈ V (ext(C)), v is contradiction. Assume then that CH I CS = η
incident to at least three vertices xi1 , xi2 , xi3 ⊂ ({u}) for some u ∈ U. This also gives rise to a
{x1, x2, x3, x4, x1} which are, in turn endpoints of contradiction, because then CH ⊂ (C* - CS) U
edges in Y . By relabeling, we may assume xi1 = η ({u}) which implies CH ⊂ MCS(u) since by
x1 and xi2 = x2 and that i3 ∈ {3, 4}. If i3 = 4 then
{v, x4, y, x1} form a short circuit. So assume that Lemma 1.1.3, MCS(u) equals the maximal
i3 = 3, and deduce that {v, x3, x4, x1, x1} is either critical subset of (C* - CS) U η ({u}) and CH is
a short circuit, or there is a degree 1 vertex w that critical by Theorem 1.1.3. Thus u ∈ CH ⊂
is adjacent to {v, x3, x4, x1, x1}. We may assume MCS(u) which contradicts that u is D-removable.
{v, x3, x4, x1, x1} is not a short circuit in T, so the Now consider the case that K is C(4)-reducible,
later holds and v has neighbors {x1, x2, x3,w} and let X be a sparse subset of S such that φ (X)
which form a short circuit. This completes the
is contractible in T and that CS(X) I MCS(C* -
proof of the theorem.
We will call any A - contract X with |X| = 4 and CS) = φ . By Theorem 1.1.2, there is a tricoloring
for which X has a triad a safe contract. of T modulo X which we denote by c. Let cH and
Theorem 1.2.2 Every configuration in Appendix cX be the colorings that are guaranteed to exist by
1 is U-reducible. Moreover, for every u ∈ U that Lemma 1.2.4, let cH(R) be the lift of cH by φ
is C-removable, there is a safe u-contract X. and let cX(R) be the restriction to R of the
Proof: Let K be a configuration in Appendix 1. coloring cX. Lemma 1.2.4 says that cH(R) = cX(R)
The computer verifies that for every u 2 U, u is and that cX(R) ∈ CH I CS. Also, we know that
either D-removable or C-removable. When the
color is C-removable, the computer finds a u- c X(R) ∈ CS(X). Therefore cX(R) ∈ CH I CS I
contract X and verifies that X is safe. After CS(X). From this and our assumption that CH ⊂
showing that every u ∈ U is either D-removable (C* - CS) U η ({u}) for some u 2 ∈ , it follows
or C-removable, the computer verifies that the that cX(R) ∈ U and that CH ⊂ MCS(cX(R))
configuration K is U-reducible. If at least one of since by Theorem 1.1.3, CH is consistent and by
the u 2 U was C-removable, then U-reducibility Lemma 1.1.3, MCS(cX(R)) equals the maximal
for K is immediately established. Otherwise, the consistent subset of (C* -CS) U {cX(R)}. Since
computer verifies that K is either D-reducible or
we are assuming that every u ∈ U is D-
C(4)-reducible
41
ISSN:0974-8652
removable, it follows that cX(R) is D-removable,

and hence that cX(R) ∉ MCS(cX(R)). This [4] Akbari, S.: Behzad, M.; Haijiabolhassen,
however is a contradiction because we know that H.Mahmoodian Uniquely total colorable graphs.
cX(R) ∈ CH ⊂ MCS(cX(R)). This completes the Graphs Combin 13, (1997) 305-314 .
proof of Theorem 1.2.3 in the case when no u ∈
U is C-removable. [5] S.Satyanarayana. “Complete Works on Four
We may assume then that there is a u ∈ U which Colour theorem (Research Note Book)”, Satyan’s
is not D-removable and hence is C-removable. Publications in Progress.
By Theorem 1.2.2, we know that there is a safe u
- contract X. We now show that we may assume [6] S.Satyanarayana “Programming Approach on
CH ⊂ (C* - CS) U η ({u}). If not, then from Discharge on Four Colour Theorem”,
“International Journal of Computational
our previous assumptions we know that there is a
Mathematical Ideas, Vol 1 No 1, (2009) PP 5-
u’ ∈ U with u ≠ u’ such that CH ⊂ (C* - CS)
212.
U η ({u’}). Now CH ⊄ C* - CS, otherwise CH
⊂ C* - CS ⊂ (C* - CS) U η ({u}). It follows [7] S.Satyanarayana “Programming Approach on
then that u’ ∈ CH ⊂ MCS(u’), so u’ is not D- Reduce on Four Colour Theorem”, “International
removable. Therefore u0 is C-removable, by Journal of Computational Mathematical Ideas,
Theorem 1.2.2. Thus, we could let u’ play the Vol 1 No 1, (2009) PP 213-236.
role of u. This proves that that we may assume u
[8] S.Satyanarayana, Dr.J.Venkateswara Rao,
∈ CH ⊂ MCS(u) ⊂ (C* - CS) U η ({u}).
“On Planar Coloured Graphs”, “International
Since X is a safe contract, Theorem 1.1.2 Journal of Computational Mathematical Ideas,
guarantees that T has a tricoloring modulo X Vol 1 No 2, (2009) PP 6-8.
which we denote by c. Using Lemma 1.2.4 and
its notation, we write cH(R) for the lift of cH by [9] S.Satyanarayana, Dr.J.Venkateswara Rao,
φ , and cX(R) for the restriction to R of the Dr.A.Rami Reddy, “Theorems on Structure of
coloring c Lemma 1.2.4 guarantees that cX(R) = Minimum counter example to the Fkorini-
cH(R) and that cX(R) ∈ CH T CS(X). Since CH Wilson-Fisk Conjecture”, “International Journal
⊂ MCS(u), it follows that cX(R) ∈ MCS(u) I of Computational Mathematical Ideas, Vol 1 No
CS(X). This is a contradiction however, because 2, (2009) PP 20-25.
X is a u-contract implies that CS(X) I MCS(u)
[10] S.Satyanarayana, Dr.J.Venkateswara Rao,
=φ . “Configurations, Projections and Free Completions
This completes the proof of Theorem 1.2.3. on Uniquely Planar Coloured Graphs”,
“International Journal of Computational
Mathematical Ideas”, Vol 1 No 3, (2009) PP 80-86.
ACKNOWLEDGEMENTS:
[11] Xu, Shaoji, The size of Uniquely colorable
We would like to express our thanks to referees for graphs. J.Combin. Theory (B) 50, (1990) 319-320.
valuable comments that improved the paper. The
first Author wish to thank Principal &
Management, Sri Venkateswara Institute of
Science & Information Technology,
Tadepalligudem for their encouragement and
cooperation in the preparation of their research
paper.
REFERENCES:
[1] G.Chartrand, D.Geller. “On uniquely
colorable planar graphs”, J.Combin. Theory 6
(1969) 271-278.
[2] A.G.Thomason. “Hamiltonian cycles and

uniquely edge colourable graphs”, Ann.Discrete
Math. 3 (1978) 259-268.
[3] Akbari S. “Two conjectures on uniquely

totally colorable graphs”. Discrete Math.1-3
(2003). 41-45.
42
ISSN:0974-8652

MATHEMATICAL IDEAS
PERCEIVING PLAGIARISM USING WEIGHTED WINDOW APPROACH-

PERFORMANCE ANALYSIS
Bobba Veeramallu@, T. Pavan Kumar#, Prof.V.Srikanth$, Prof.K.Rajasekhara Rao^
@
Dept. of CSE, KLEF University, bvmallu@rediffmail.com
#
Dept. of IST, KLEF University, pawantummala@yahoo.co.in
$
Dept. of IST, KLEF University, srikanth_ist@klce.ac.in
^
Dept. of CSE, KLEF University, rajasekhar.kurra@klce.ac.in
ABSTRACT
Plagiarism of data from the internet is a rapidly growing problem in this competitive
world. Most of the students get accustomed to sometimes get their work done with a “cut and
paste” approach in assembling a paper in part. It perverts learning and assessment of subject.
Detection of cut and paste plagiarism is a time consuming and cumbersome task when it is
done manually, and can be greatly aided by automated software tools. This paper presents an
approach on the implementation of a software tool called SNITCH that uses a fast and
accurate plagiarism detection algorithm using the Google Web API. Several issues related to
plagiarism detection software are discussed and in addition to it the performance and
accuracy study are also dealt with.
Keywords: Plagiarism, algorithm, design
1. INTRODUCTION
Plagiarism is a pervasive form of academic plagiarism detection and an associated software
dishonesty in collegiate settings. Since it distorts tool called SNITCH.
learning and assessment, deterring and detecting it
are crucial to maintaining academic integrity. 2. DEFINITION OF PLAGIARISM
Plagiarism fundamentally warps two essential
aspects of education, learning and assessment. “Plagiarize” derives from a Latin root meaning
Students who submit plagiarized work deprive “to kidnap”[4]. It is a form of dishonesty that
themselves of the learning opportunities afforded misrepresents intellectual property and that
by authentic academic productions and by deprives the creator of intellectual property due
assessments of those productions by educators [2, recognition. In academia, it is an unpardonable
3]. Large class sizes and an increase in writing mistake. Examples of plagiarism include failure to
assignments that result from writing across the give appropriate acknowledgement when using
curriculum combine to make detection of other’s words and presenting other’s line of
plagiarism burdensome. The rapid increase of thinking[5].
written material on the Internet and its ease of Plagiarism substitutes the physical labour of
appropriation contribute to the problem. theft and misrepresentation for the labour and
Detection of cut and paste plagiarism is time growth of learning. Moreover, plagiarism erodes
consuming when done by hand, so plagiarism the sense of community that is essential to free
detection software has emerged in response[13]. academic inquiry. Misrepresentation masks the true
This paper describes the design of an algorithm for identity of members of the community.
automated
Perceiving Plagiarism Using Weighted Window Approach- Performance Analysis
43
ISSN:0974-8652
3. DETECTING PLAGIARISM continues to prove an invaluable tool both for the

detection of cheating and for grading assistance in
Detecting plagiarism can be a time-consuming task, several courses.
all characteristics that make the problem ideally
suited to a software solution. In this, a brief 3.2.1 Existing Softwares
discussion of the general, non-software-based
approach to plagiarism detection is provided, The Eve2 software performs adequately for
followed by an overview of existing software shorter papers, but its report generation feature can
solutions and issues to be considered in the design be inaccurate. Although Eve2 is really designed to
of such software solutions. determine how closely a student paper matches a
single online source, for simply detecting the
3.1 Manual Approach simple presence of plagiarism it is acceptable.
TurnItIn is a well-respected service,
In a time with a seemingly limitless cache of data where student papers are submitted via the Internet
from which to “borrow” from the Internet, an for analysis. Reports are generated and returned to
approach commonly used by educators to detect the the instructor, normally within four to six hours of
cut and paste approach to plagiarism is to highlight submission. This service is expensive.
suspicious excerpts in a paper, and then enter them The MyDropBox service is a recent and
into an online search engine. If identical excerpts able competitor to TurnItIn, with a similar pricing
are found in an online source, it is likely that the strategy and turn-around time. Reports are
excerpt was plagiarized [6]. Certainly, if multiple generated within 24 hours of submission. But the
such instances of identical excerpts are discovered cost varies according to the plans that are chosen.
in a single paper, a strong decision can be made
that intentional plagiarism is present. 4. TECHNICAL ISSUES
The problem with this manual approach is Plagiarism detection software that uses the
that it is labour intensive, requiring detailed, on- Internet for its corpus is subject to effective
screen reading and re-reading of each paper, countermeasures. One is that Web sites associated
coupled with the repeated use a search engine with the sale of term papers are not openly
including copying and pasting selected passages connected to the World Wide Web. Materials
from each paper using a mouse and keyboard. acquired through these sites are likely to escape
Although this tedious approach is perhaps less detection. In addition, Web sites can deploy
exhausting than referring textbooks looking for software that repels Web crawlers such as those
potential matches, or being intimately familiar with used by TurnItIn[8,9].
enough such textbooks and other sources to Another issue is that an instructor can recognize
recognize stolen excerpts. where some copied material has been slightly
revised by replacing, adding or deleting one or
3.2 Software Approach more words to avoid detection. Software can
duplicate this approach, although it is a difficult
Software has been developed that reduces a lot of problem to solve. Because search engines allow for
the labour intensive aspects of manual approach of “wildcards” a simple approach of replacing any
detecting plagiarism. There are a number of short or common words in the passage with a
commercially available software tools and services wildcard may be an effective technique to detect
that perform automated checking. But the cost is cut and paste plagiarism.
relatively high, the turn-around time is sufficiently Other technical issue is that of false positives
long, or both, reducing the availability to those and false negatives. Corpus-based programs, such
educators with budget constraints. These available as TurnItIt.com, do an excellent job of finding
automated approaches often assume that large matches between student submissions and items in
sections of a paper, or even entire papers, would be their database. These programs, however, do not
copied verbatim[7]. Yet, for technical oriented distinguish between matches that are properly cited
research papers, such as in computer science and and those that do not, contain a high index of
engineering disciplines, a cut and paste approach plagiarism.
where paragraphs, sentences or even phrases can be Software tools that are available for use by
gathered into a report is easier to get away with. instructors could also be used by students, such that
Software tools have been successfully used students may try to defeat automated plagiarism
for detection of plagiarism in student programming detection by using such a program while submitting
assignments for many years. Two program files are their work [10]. Because of this the student submits
compared after some compiler like pre-processing his work without understanding the original
to try to find similarities in the files that could content.
indicate plagiarism. This form of software
44
ISSN:0974-8652
The issues and approaches raised here, and the will increase or decrease the thoroughness, and
features and techniques used in previous tools and lengthen or shorten the time taken to analyze a
manual methods, provide the motivation for the paper, since the time to perform each search is
design and implementation of the SNITCH determined by load on the Internet, and Google
software. specifically, rather than the capabilities of the
user’s own computer.
5. DESIGN OF SNITCH
5.3 Evaluation
Here in the design of SNITCH the design of the
plagiarism analysis and detection algorithm used in An initial evaluation of the SNITCH software was
SNITCH is discussed in detail. performed to measure its effectiveness at detecting
instances of plagiarism in custom-designed
5.1 Algorithm plagiarism benchmarks and a sampling of typical
computer science student term papers. Results are
The algorithm developed for SNITCH uses a compared with results for the same papers using
sliding window technique and average length per the only other available practical and cost-effective
word metric to identify potential instances of software tool, Eve2. No formal comparison was
plagiarism. In general, the algorithm uses the done with online subscription services due to cost
following steps: constraints.
Open a document
Analyze the document 5.3.1 Comparison of Performance
o Read a window containing the first/next W
words Experiments using four synthetic benchmark term
o Measure the number of characters for each papers and a sampling of 10 actual student term
word papers were performed. The synthetic benchmarks
o Calculate the Weight of the window, the consisted of carefully crafted documents containing
average number of characters per word for the known amounts and instances of cut and paste
words in the window plagiarism representing hypothetical papers
o Associate this Weight with this particular containing high, moderate, minimal, and no
window for later use. plagiarism[4,13]. Actual student papers were
o Repeat for all such windows in the document, manually analyzed using careful online detective
shifting the window forward in the document work, and were divided into similar groupings
by 1 word based on the prevalence of plagiarism that was
Search for plagiarized passages found. These student papers were all rough drafts,
o Order windows in decreasing order, and and any plagiarism detected was later removed by
eliminate overlapping windows. the
o Rank all windows in decreasing order by Students. Three experiments were performed to
Weight. measure analysis speed and accuracy of SNITCH
o Select the top N weighted windows, and on the synthetic and real documents, and in
search the Internet for each, gathering the top comparison with the commercial plagiarism
search result (if any) for each detection program Eve2.
Generate a report – Create an HTML document
containing statistics of search time, number of Manual stats SNITCH stats
searches performed, percentage of document found
PET INSTANCES PET INSTANCES
to be plagiarized, and other
Pertinent statistics. Include the original document 0 0 0 0
with embedded HTML tags linking plagiarized
passages to their sources on the Internet [11]. 15 5 40 2
The algorithm is parameterized to allow
variation of the size of the sliding window (W) and 40 10 50 5
number of searches performed (N), to enable fine-
tuning on a per-user basis. Decreasing W will lead 75 19 63 12
to more potential candidates, but may increase false
TABLE 1: ANALYSIS RESULTS
positive results because the fewer words there are
in a search phrase, the more likely they could occur
by chance. Increasing W can improve the The graphical representation that shows the
confidence in individual search results, but if set performance of SNITCH is better than that of the
manual approach.
too high, it may reduce that likelihood that any
matches will be found if the window is larger that
the plagiarized passage. Increasing or decreasing N
45
ISSN:0974-8652
6. CONCLUSION
In this paper we have noticed that SNITCH

program provides an efficient and accurate
alternative to commercial tools and services,
producing good accuracy and faster analysis at
affordable cost.
The availability of SNITCH will increase the
threat of detection and prevents individuals from
using cut and paste plagiarism.
FIG 2 : MANUAL STATS
7. FUTURE WORK
This SNITCH tool can compare only twenty

documents at a time. In future we enhance the
performance of SNITCH tool so that it can be used
to look into more files with a better accuracy and
speed even for pdf files.
ACKONOWLEDGEMENT
We would like to express our thanks to referees for

valuable comments that improved the paper.
REFERENCES:
FIG
3: SNITCH STATS
[1]. Freedman, M. (2004, March). A tale of
Plagiarism and a new paradigm. Phi Delta Kappan
5.3.4 Comparison of performance of
85 (7), 545-549. Retrieved May 6, 2004 from
SNITCH and Eve2
Academic Search Elite.
Eve2 SNITCH
[2]. Kansas College Give First ‘XF’ Grade to
6:45 0.15 Plagiarist. (2003, December 8).Community College
Week. Retrieved May 6, 2004 from Academic
6:45 0.18 Search Elite.
7:00 0.38 [3]. Scanlan, P. (2003, Fall). Student online

plagiarism: how do we respond. College Teaching,
54 (4) 161-164. Retrieved May 6, 2004 from
7:30 0.44
Academic Search Elite
TABLE 2: ANALYSIS TIME (RESULTS) [4]. Kellogg, A. (2002, February 15).Students

plagiarize online less than many think, a new study
The graphical representation is as follows finds.Chronicle of Higher Education, 48, A44.
Retrieved May 5, 2004 from Academic Search
average analysis time
Elite
8
7 [5]. Howard, R. (2002, January). Don’t police
6 plagiarism. Just teach! Education Digest, 67 (5),
5 46-50. Retrieved May 5, 2004 from Academic
Series1
4
Series2 Search Elite.
3
2 [6]. Carroll, J. A Handbook for deterring
1
Plagiarism in Higher Education. Oxford, Oxford
0
1 2 3 4 5
Centre for Staff and Learning Development, 2002.
[7]. Whitley, B. Academic Dishonesty: an

FIG 4 : ANALYSIS TIME COMPARISION Educator’s Guide. Mahwah, N.J., Erlbaum, 2002.
RESULTS (Hesburgh LB 3609.W45 2002)
46
ISSN:0974-8652
[8]. Decoo, W. Crisis on Campus: Confronting

Academic Misconduct. Cambridge, MA., MIT
Press, 2002. (Hesburgh LB 2344 .D43 2002).
[9]. Harris, R. A. The Plagiarism Handbook.Los

Angeles, Pyrczak Publishing, 2001
www.AntiPlagiarism.com
[10]. Lathrop A. Student Cheating and Plagiarism

in the Internet Era.Englewood, CO., Libraries
Unlimited, 2000. (Hesburgh LB 3609 .L28 2000)
[11]. C. Humes, J. Stiffler and M. Malsed.

Examining Anti-Plagiarism Software: Choosing the
Right Tool. Claremont-McKenna College technical
report. 2003.
[12]. Brian Martin. Plagiarism: policy against

cheating or policy for learning? Nexus: Newsletter
of the Australian Sociological Association, 16:2,
pp. 1-12, 2004.
[13]. L. Renard. Cut and paste 101: Plagiarism and

the Net. Educational Leadership, 57:4, pp. 38-42,
2000
[14]. Arwin, C. and S. M. M. Tahaghogh (2006).

Plagiarism Detection Across Programming
Languages. ACM International Conference
Proceeding Series, vol. 171. Proceedings of the
29th Australasian Computer Science Conference,
Hobart, Australia, vol. 48 ,pp. 277 – 286.
[15]. Niezgoda, S. and T. Way. (2006) SNITCH: A

Software Tool for detecting Cut and Paste
th
Plagiarism. Proceedings of the 37 Special Interest
Group on Computer Science Education. Pp. 51 –
55. New York: Association for Computing
Machinery.
47
ISSN:0974-8652

MATHEMATICAL IDEAS
SYSTEM REPRESENTATION FOR SOFTWARE ARCHITECTURE

RECOVERY
Shaheda Akthar@, Sk.MD.Rafi#

@
Assoc.Professor, Sri Mittapalli College of engineering, Affiliated to JNTU, Kakinada,
Email:shaheda.akthar@yahoo.com
#
Asst.Professor, Sri Mittapalli Institute of Technology for women, Affiliated to JNTU, Kakinada.
Email:rafi_527@rediffmail.com
ABSTRACT
The source code of the system needs to be analyzed in every step. For analyzing, system should be
represented in the form of a tree using any of the models that suits for our system because the source code
of the soft ware system will be highly detailed. So we can’t analyze the source code without using any
model. So, for analyzing a model is used to describe the entities and relationships .so, to represent that
model a graph is used. In this chapter we are considering a graph known as Attributed Relational graph
[ARG].
Software systems uses graphical representation because it will be very easy for them to trace out
the errors ,if any the source code of the system should be represented in the form of graph and that graph
should be divided in to smaller sub graphs based on their properties.
After representing the system as a graph, we have to find out the entities that were related, To
partition the graph into smaller modules. Related entities can be found out using association properly.
After finding out the similar entities [1], file level and function level measurements are defined.
Keywords:ARG,XML,DTD,Domainmodel,datamining,association.
1.1. Graph representation of a software includes file, function, statement,

system expression, variable
A Graph is a collection of nodes (vertices) and 2. Finding Relationship between source
links (edges). So, when a software system is to code entities as an association between
be designed as an attributed relational graph, model classes.
there should be a specific domain model to XML notation can also be used
represent the nodes and links of the graphs. to define a domain model using
1.1.1. Specified Domain Model Document Type Definition (DTD). The
A Specific domain model [2] is used to represent major advantage in using these type of
the software system which consists of entities, in representations is, XML can easily
the form of graphs, charts etc. For any domain validate data.
software system a domain model should be After representing a software
proposed. system using a domain model it should
For example, for every programming language be analyzed.
also domain model should be proposed. A software system should be
1.1.2. Domain Model for Programming analyzed at two levels
Language i) File level
The basic theme is to obtain the ER ii) Function level
graph of a system at the source code level. The files and directories are to be used to
domain model should be represented in terms of make analysis at the file level.
classes and associations. This can be done by Global variables, aggregate data types
1. Considering source code constructs as and functions are to be considered to
the domain model classes which make analysis at the function level.
System Representation For Software Architecture Recovery
48
ISSN:0974-8652
Entities and relations of the specified more relations in the software system’s
domain model will be analyzed in source code. The advantage of this
directly if both the file level and domain model is that it is simpler than
function level analysis is carried out in a the detailed source-code domain model.
proper way. In Abstract domain model It is language independent for
the different types of entities are a procedural programming paradigm; and
subset of the types of entities in the yet it is adequate for architecture level
software system’s source code, and each Analysis.
relation in the abstract domain model is
an aggregation of one are
Here the file entities with other type of of these were of different
entities are separated using separate granularities.
entity-abs. this separation is because of iv) Each relation contains the
these were of the same granularity. attributes from and to, to
From the above figure, we can observe denote the source and
the following properties. destination entity for that
i) Entity-abs in a class relation.
ii) Relation-abs inherits all the Two types of relations exists
properties of that are identified i)file level and
by every entity in the abstract ii)function level
domain model.
iii) File entity with other type of File level relations
entities are separated using File level relations are of 4 types.
simple entity-abs (simpEnt- i) import-resource
abs).this separation is because ii) export-resource
iii) use-resource
49
ISSN:0974-8652
iv)contains-resource and function f1 calls function f, and

function f is the implementation of
function abstraction F.
Relation imp-R (import-resource): is
defined between a file-abstraction L and
Entity-abs entity-abstraction R, denoting that L
uses R but does not contained in R1.
attribute Example Description Relation exp-R (export-resource): is
defined between a file-abstraction L and
Name “des” Name of the entity in the
entity-abstraction R, denoting that L
source code
contains R and another file-abstraction
File# 4 File number of the source L1 uses R.
code file where the entity 3.1.3. Source Graph
is defined To analyze the software system, a
line# 45 Line number of the entity source graph should be modeled. The
in the source code file notation for Attributed Relational Graph
(ARG) that is presented in is adopted to
Implement- 13 Unique identifier of the define all graphs.
id object in the source-level The attributed relational graph
domain model representation of the source-graph is a
six-tuple Gs={Ns, Rs, Es, µ s, €s)2 that is
defined as:
Function level relations Ns: {n1,n2,……..,nn} is the set of
Relation use-F: this relation is defined attributed nodes, obtained from the
between two function-abstractions F abstract domain model.
and F1, denoting that the source code Rs: {r1,r2,…..rm} is the set of attributed
function f calls the source code function edges, obtained from the abstract
of f1 domain model.
Relation use-T: this is defined between As: alphabet for node attribute values
a function abstraction F and a data type such as node labels, node types, and
abstraction T, denoting that function f their values.
reads the value of a variable v and the Es: alphabet for edge attribute values
variable v is of type t, and t is the such as edge labels, edge types, and
implementation of data type abstraction their values.
T. µs: Ns->(As × As)p: a function for
Relation use-V: this is defined between returning the “node attribute, node
a function-abstraction F and a variable attribute value” pairs where p is a
abstraction V, denoting that the function constant and denotes the number of
f reads the value of global variable v. node attributes.
File level relationships: €s: Rs-> (Es × Es)q: a function for
Relation cont-R(contain-resource): A returning the “ edge attribute, edge
file-abstraction is called a composite attribute value” pairs where p is a
entity and a function-abstraction, a type- constant and denotes the number of
abstraction, or a variable-abstraction is edge attributes.
called a simple-entity such that a
composite entity contains a set of
simple entities.
this relation is defined between a file-
abstraction L and either a function-
abstraction F, or a data type-abstraction
T, or a variable-abstraction V, denoting
that the source-file l defines the
implementation of function f.
Relation use-R(use resource): this
relation is defined between a file-
abstraction L and either a function-
abstraction F, or a data type-abstraction Figure 1.1: an attributed relational
T, or a variable-abstraction V, denoting graph representation of a source graph
that source-file l defines a function f1 , Gs={Ns, Rs, Es, µ s, €s).
50
ISSN:0974-8652
Ni(Gs) = ((type, function Data mining [3] refers to a collection of

abs),(des,”/a/..mgr”),(id,13),(line45),(fil algorithms for discovering interesting relations
e4))indicating that node I of the source among data in large databases [4].
graph G is an entity of type Function- Frequent items can be found out by applying
abs with des “/a/../mgr” and idF10 and it association rules [5], which is an
has been defined in the line 45 of the implication of the form x ≈> y, where x and y are
source file 4; and disjunctive item sets which are subsets of set of
E j (Gs)=((from,n2),(to,n8),(type, use- items I={i1,i2,i3……iN},N ≥ 2.The association
F),(#line89),(file#4) indicating that edge rules [6]are generated by frequent-itemsets and
j of the source graph Gs is an object of the frequent itemsets can be grouped by the
type use-F that relates the function n2 to Apriori algorithm. A k-itemset whose elements
the function n8 with a function call are contained in every basket of a group of
relation in line89 of file 4. baskets. The cardinality of this group of baskets
3.2. Computing maximal association must be greater than a user-defined threshold
Maximal association can be extracted by data called minimum-support.
mining and is considered as an interesting In order to apply the Apriori algorithm on the
property for grouping the entities in to cohesive source-graph Gs, we define B(Gs) as the basket
modules. representation of the source-graph Gs=(Ns,Rs):
Maximal association is defined in a B (Gs) = {b: Function-abs; I: set (Entity-abs)
group of entities in the form of a maximal set of Figure 3.2(a),(b),(c) Application of data mining
entities that all share the same relation to every in extracting frequent itemsets.(d) Representation
member of another maximal set of entities. of the frequent itemsets for system analysis.
For every set of functions, denoted as F, we The above figure illustrates the process of
can determine a set of shared entities, denoted as generating frequent itemsets from the source-
E, where every function f in F has a relation r to graph Gs. In figures (a),(b), the entities and
an entity e in E. for example, two functions f and relationships in source-graph Gs are represented
g can share the data type t and variable v by the as a data base of baskets and items B(Gs). in
relations use-T and use-V, respectively. The figure (c), two frequent-itemsets generated by the
operation sh-ents(F) returns the set of shared apriori algorithm on B(Gs) are shown. Each
entities E for the set F as follows: frequent-itemset is presented as a tuple ({baskets,
sh-ents(F)={e | ∀ f ∈ F; ∃ rel :X {items}), where {baskets} is the set of functions
• X ∈ {use-F,use-T,use-V} ∧ (f,e) ∈ rel } and {items} is the set of “relation and entity”
Similarly, for every set E of entities we can pairs, such that:
determine a set of functions F, where every {baskets}=sh- funcs({items}) {items}=sh-
function f in F has a relation r to an entity e in E. ents({baskets})
the operation sh-funcs(E) returns the set of Hence the set of functions in {baskets} and the
sharing functions F for the set E as follows: set of entities in {items} are related by maximal
sh-funcs (E)={f | ∀ e ∈ E; ∃ rel :X association.
• X ∈ {use-F,use-T,use-V} ∧ (f,e) ∈ rel } Finally figure (d) represents a small portion of
frequent 5-itemsets extracted from a software
system’s source-graph. The first line is
interpreted as: all the functions F774, F800, F807
use functions F209, F811, F812, and use
aggregate type T5 and global variable V259.
Each frequent 5-itemset is equivalent to a
concept with intended size 5.
3.1 Similarity measure between two entities
A similarity measure[7] is defined so that two
entities that are alike possess higher similarity
value than two entities that are not alike. the
clustering research literature provides a rich
collection of techniques for extracting groups of
related software entities using different similarity
metrics namely association metrics, correlation
metrics, and probabilistic metrics but the jaccard
A of functions F and a set of entities E are related metric produces better clusters than the others.
by maximal association, iff:
F = sh-funcs (E) ∧ E = sh-ents (F)
2.1. Data Mining
51
ISSN:0974-8652
3.2Entity association similarity measure Therefore, the association property is defined

The entity association measure is an between either: sharing entities (clustering), or
extension to the notion of association in the shared entities (data mining).
clustering and data mining domains that are To apply the association rules to a graph an
briefly compared below: associated graph of graph nodes is defined, when
Clustering: The association similarity is defined two more source nodes share one or more sink
between two entities are the proportion of the nodes. A source node is a node where an edge
numbers of shared and total attribute-values, emanates from it. A sink node is a node where an
figure3.3 (a). [8] edge points to it. In this sense, the whole group
Data mining: The association rule is defined of source nodes and sink nodes are denoted as an
between two sets of items as the proportion of associated group. By considering the source
the numbers of the shared and total baskets, nodes as the “basketset” and the sink nodes as
figure 3.3(b). [4] the “itemset” “
Fig.,3.3 Association Rules

3.3.2. Source region
A source region
Gsreg=(Nsreg,Rsreg,Asreg,Esreg,Usreg,Esreg) of
a source graph Gs=(Ns,Rs,As,Es,Us,Es) is a sub
graph of Gs.In the source region Gsreg(j) each
node ni!=nj satisfies the association property
entAssoc(nj,ni)>0 with respect to node nj.We
call ni the main seed of the source region
Gsreg(j) and use it as the identity of this source
region.
N j
sr
= { ni | ni ∈ Ns ∧
∃ nj ∈ Ns
• entAssoc (nj , ni) > 0 } ∪ { nj }
R jsr = { ns ; nt | ns , nt ∈ N jsr ∧ ∃ rk
• rk = (ns , nt) ∧r k ∈ Rs }
For a given source graph Gs=(Ns,Rs) we

generate | Ns | source regions. A source graph Gs=(Ns,Rs) at functional level
represented as an
ARG
52
ISSN:0974-8652
determine the maximum association value of

each node with respect to the source regions
main seed.
Applying Apriori:
In the source region Gsreg(1) with
node 1 as main seed in fig 3.3.2(d) We consider
two associated groups with nodes 1,7,10,2,13
with entAssoc of 4;and 1,6,10,7,2 with entAssoc
of 3,5. The similarity value of node 10 to the
main seed node 1 is 4 and is obtained from the
first associated group
Fig 3.3.2(a) Source region with 1 as main seed
3.3.2(c) Source region Gsreg(1) after applying

Apriori
Fig 3.3.2(b) Source region with 6 as main seed
In these figures fig 3.3.2(a) and fig 3.3.2(b)

represent 2 source regions Gsreg(1) and Gsreg(6)
of the source graph Gs. Each node of Gsreg (1) is
a member of an associated group with respect to
main seed n1. However it is not clear what is the
highest association value of each node with
regard to main seed n1 since each node can be a
member of several associated groups, thus
different association values in each group. The
Apriori algorithm computes all the associated
groups in a source region and allows to
3.3.2 (d) Source region Gsreg(6) after applying Three methods of similarity measures
APriori between two groups of entities are commonly used
At phase I of the incremental graph matching in clustering. They are
process the user may select a main seed nj that 1. Single linkage
corresponds to the source region Gsreg(j) for the 2. complete linkage
current matching phase i. 3. group average similarity
3.4 Similarity measure between two groups of In single linkage method, the maximum or
entities minimum similarity between every pair of entities,
Group association can be defined as a one in each group, is considered as the similarity
similarity measure between two groups of system value between two groups. The single linkage
entities gi, gj based on the similarity between two computes higher similarity value for the groups that
entities in a graph. are non-compact and isolated, where as, complete
53
ISSN:0974-8652
linkage computes higher similarity values for REFERENCES

cohesive and compact groups. To avoid the [1] Douglas B. West. Introduction to Graph
extremes, the group average similarity method Theory. Prentice Hall, 1996. Page 19.
defines the similarity between two groups as the [2] Allan Terry et al. An annotated repository
average of similarities between all pairs of entities schema, domain-specific software
that are made up of one entity from each group. architecture. Technical report, TFS and
In this thesis, the group average similarity ARDEC, October 1993.
method is adopted to compute the proposed group [3] Baeza-Yates, R., & Ribeiro-Neto, B. (1990).
association metric group assoc, as follows: Modern Information retrieval, ACM Press,
In this equation, the first summation New York.
iterates over every entity in group g i and the [4] Bayardo Jr., R.j (1997). Brute-force mining of
second summation iterates over every entity in high-confidence classification rules. In
group g k in order to add the similarity values Heckerman, D., Manila, H., & Pregibon, D.
(Eds.), Proceedings of the Third International
Conference on Knowledge discovery and Date
Mining (KDD-97), pp. 123-126 Newport
Beach, CA. AAAI Press.
[5] Agrawal, R., & Srikant, R. (1994). Fast
algorithms for mining association rules. In
Proceedings of the 20th International
conference on very Large Databases (VLDB-
94), pp. 487-499 santiago, Chile.
sim j, m between every pair of entities, one entity [6] Ahonen-Myka, H., Heinonen, O., Klemettinen,
in each group. sim j, m refers to the similarity M., & verkamo, A.I. (1999). Finding co-
value between node n j in group g i and n m in occurring text phrases by combining sequence
group g k. and frequent set discovery. In Feldman, R.
for every entity n m € g k that does not exist in the (Ed.), Proceedings of the Sixteenth
domain of n j the similarity value sim j, m between n International Joint Conference on Artificial
j and n m is zero. Therefore, only those entities in g Intelligence (IJCAI-99) workshop on Text
n
k that exist in the domain D j are considered for Mining: Foundations, Techniques and
similarity calculation between two groups. The applications, pp.1-9 Stockholm, Sweden.
terms |g i| and |g k| denote to the cardinality of each [7] Arun Lakhotia. A unified framework for
group. expressing software subsystem classification
techniques. Journal of Systems and Software,
36(3):211–231, 1997.
4. System representation [8] Brian S. Everitt. Cluster Analysis. John Wiley,
A soft ware system can be 1993.
represented at a higher-level of abstraction in the [9] Agrawal, R… Imielinsky, T., & Swami, A.
form of a source graph Gs along with the collection (1993). Mining association rules between sets
of domains, which is defined as two-tuple; of items in large databases. In Proceedings of
System= (Gs, D(Ns)) the 1993 ACM SIGMOD International
Where G = ( Ns, Rs) ∧ D(Ns) = [Dnj | j ∈ [1
s Conference on Management of Data
..| Ns| ] ] (SIGMOD-93), pp. 207-216.
, D(Ns) is an ordered sequence of entity domains D [10] Angell, R.C., Rreund, G. E., & Willet,
n
j by the average similarity of each domain, where
P.(1983). Automatic spelling correction
each domain is a search space for a module using a trigram similarity measure.
recovery. In this model the matching process Information Processing and Management.
searches only with the appropriate domains not the 19(4), 255-261.
whole source graph. [11] Anil K. Jain. Algorithms for Clustering Data.
Prentice Hall, Englewood Cliffs,
ACKNOWLEDGEMENT N.J.,1988.
[12] Architectural level. In Proceedings of the
17th ICSE, pages 186–195, 1995.
We would like to express our thanks to
[13] Bayardo Jr., R. J., & Agrawal, R.(1999).
referees for valuable comments that improved Mining the most interesting rules. In
the paper. Proceedings of the Fifth International
Conference on Knowledge Discovery and
Data Mining (KDD-99), pp.145-154 San
Diego, CA.
54
ISSN:0974-8652

MATHEMATICAL IDEAS
ISSN: 0974-8652 / www.ijcmi.webs.com / VOL 2-NO1-PP 55-59 (2010)
EXPERIMENTAL AND THEORETICAL EVALUATION OF ULTRASONIC VELOCITIES IN

BINARY LIQUID MIXTURE CYCLOHEXANE + O-XYLENE AT 303.15, 308.15, 313.15 AND
318.15K.
Narendra K@, Narayanamurthy P# & Srinivasu Ch$

@
Department of Physics, V.R.Siddhartha Engg.College, Vijayawada, Andhrapradesh, 520007, Email:-
kollanarendra@rediff.com
#
Department of Physics, Acharya Nagarjuna University, Nagarjuna nagar, Guntur, Andhrapradesh
$
Department of Physics, Andhra Loyala College, Vijayawada, Andhrapradesh, 520008
ABSTRACT
A comparison of ultrasonic velocity evaluated from Nomoto’s relation, Vandael ideal mixing
relation, impedence relation, Rao’s specific velocity relation and Junjie’s theory has been made in the
binary mixture cyclohexnae with o-xylene at 303.15, 308.15, 313.15 and 318.15 K. Ultrasonic velocity
and density of these mixtures have also been measured as a function of temperature and the experimental
values are compared with the theoretical values. A good agreement is found between experimental and
Vandael ideal mixing relation ultrasonic velocities. U2exp/U2imx has also been calculated for non-ideality in
the mixtures. The relative applicability of these theories to the present system discussed. The results are
explained in terms of intermolecular interactions occurring in these binary liquid mixtures.
Keywords - Ultrasonic velocity, Binary liquid mixtures, O-xylene, Theories of ultrasonic velocity.
AMS_82D15
2. I. INTRODUCTION 3. II. EXPERIMENTAL DETAILS

Ultrasonic study of liquid and liquid mixtures has gained The chemicals were redistilled and purified by the standard
much importance during the last two decades in assessing the methods described 18,19. Liquid mixtures of different known
nature of molecular interactions and investigating the physico- compositions were prepared by mixing measured amounts of
chemical behaviour of such systems. A survey of literature1-5 the pure liquids in cleaned and dried flasks. Ultrasonic
indicates that excess values of ultrasonic velocity, adiabatic velocity was measured by a single crystal variable path
compressibility and molar volume in liquid mixtures are interferometer (Mittal enterprises, Model F-80X) at a
useful in understanding the interactions between the frequency of 3 MHz. The working principle used in the
molecules. Several reseachers6-10 carried out ultrasonic measurement of speed of sound through medium was based
investigations on liquid mixtures and correlated the on the accurate determination of the wavelength of ultrasonic
experimental results of ultrasonic velocity with the theoretical waves of known frequency produced by quartz crystal in the
relations11-15 and interpreted the results in terms of molecular measuring cell20,21. The apparatus is standardized first with
interactions. Velocities in the binary liquid mixture distilled water then with benzene at various temperatures, the
cyclohexane with o-xylene using the above theoretical results obtained are found to be in good agreement with
relations are compared with the experimental values of reported values in the literature. An electronically digital
ultrasonic velocities at four temperatures 303.15, 308.15, operated constant temperature bath has been used to circulate
313.15 and 318.15K. An attempt has been made to study the water through the double walled measuring cell made up of
molecular interaction from the deviation in the value of U2exp/ steel containing the experimental solution at the desired
U2 imx from unity based on the earlier studies16,17. temperature. The accuracy of the velocity measurements is
±5ms-1. The densities of pure liquids and liquid mixtures were
measured by employing a 25ml specific gravity bottle at all
the temperatures and weights were taken to an accuracy of
±0.1mg.
Experimental and theoretical evaluation of ultrasonic velocities in binary liquid mixture cyclohexane + o-xylene at 303.15, 308.15, 313.15
and 318.15k.
55
ISSN:0974-8652
The measurements were made at all the temperatures with the Collision Factor Theory,
help of thermostat with an accuracy of ±0.1K. UCFT = Uα x Smix x Rf mix --- (7)
III. THEORY
Vandael Theory:
The adiabatic compressibility has been determined by using
the experimentally measured ultrasonic velocity (U) and
1
U Vandael = 1/2
-- (8)
density (ρ) by the following formula   x1 x 
1 ( xM 2 2) 
1 1 +xM 2
+ 2 2 
βad = --- (1)  MU
1 1 MU 
2 2 
ρU 2 Where, x1, x2 – Mole fractions
The molar volumes of the binary mixtures were calculated M1, M2 – Molecular weights
using the equation U1, U2 – Ultrasonic velocities
V = (X1M1+X2M2)/ρ --- (2)
Junjie Theory:
Nomoto Theory:
3
x1Vm1 +x2 Vm2
 x R + x2 R2  UJunjie = 1/2
-- (9)
UNomoto =  1 1  --- (3)  x1Vm1 x2Vm2 
 x1V1 + x2V2  ( x1 M1 +x2 M2 )  2 + 2 
Where R1 and R2 are the radiuses of 1st and 2nd liquid   U1 U2 
V1 and V2 are the molar volumes for pure liquids Where, x1, x2 – Mole fractions
R1 = molar volume of 1st liquid x (velocity)1/3 M1, M2 – Molecular weights
R2 = molar volume of 2nd liquid x (velocity)1/3 Vm1, Vm2 – Molar volumes
U1, U2 – Ultrasonic velocities
Free Length Theory (FLT):
k IV. RESULTS AND DISCUSSION
UFLT = --- (4) The values of density, viscosity, adiabatic compressibility and
L f x (density )1/ 2 molar volume for different mole fractions of o-xylene with
Where k is a constant and its values are 631, 636.5, 642 and cyclohexane at different temperatures are given in Table 1.
647 at 300C, 350C, 400C and 450C respectively.
Lf – Free length TABLE I - VALUES OF DENSITY (ρ), ADIABATIC COMPRESSIBILITY (β) AND
MOLAR VOLUME (Vm) FOR DIFFERENT MOLEFRACTIONS OF O-XYLENE WITH
CYCLOHEXANE AT DIFFERENT TEMPERATURES
Collision Factor Theory (CFT):
3Vm  γ RT   MU2  

γGopal rao = 3 1−  1+  −1 --- (5)
16ΠN  MU2   γ RT  
 
Where, N = Avagadro’s Number = 6.023 x 10 23

γ = Cp/Cv = Ratio of principle Specific heats
R = gas constant = 8.3144 x 107
M = Molecular weight of the liquid
U = Ultrasonic velocity
T = Absolute temperature
Vm= Molar volume
3Vm  γ RT   MU2  

γShaff’s = 3 1−  1+  −1 --- (6)
16ΠN  MU2   3γ RT  
  
γavg = (γGopal rao + γShaff’s) / 2
4
Bi = x Π x (γavg i)3 x N
3
Space filling factor, Rf i = Bi / Vmi
Collision Factor, Si = Uexp / (1600 x Rf i)
and 318.15k.
56
ISSN:0974-8652
X1 ρ x 103 (kg/m3) β x 1012 (m2 N-1) Vm (cm3 mol-1)
at 303.15 K
0.0000 0.7678 85.3058 109.6119
0.0908 0.7768 84.0621 110.9140
0.1835 0.7865 82.2063 112.1388
0.2781 0.7975 80.1706 113.2023
0.3747 0.8096 77.9252 114.1358
0.4734 0.8222 75.7054 115.0268
0.5742 0.8304 74.1076 116.5612
0.6772 0.8368 72.3779 118.3771
0.7824 0.8486 69.6749 119.4594
0.8900 0.8614 67.0067 120.4319
1.0000 0.8707 64.0815 121.9249
at 308.15 K
0.0000 0.7625 90.1520 110.3738
0.0908 0.7731 88.3109 111.4448
0.1835 0.7831 85.9713 112.6257
0.2781 0.7946 83.5235 113.6155
0.3747 0.8070 80.9327 114.5035
0.4734 0.8194 78.7345 115.4199
0.5742 0.8280 76.7718 116.8990
0.6772 0.8349 75.1008 118.6465
0.7824 0.8466 72.2496 119.7416
0.8900 0.8600 69.3633 120.6280
1.0000 0.8694 66.5164 122.1072
at 313.15 K
0.0000 0.7587 93.2715 110.9266
0.0908 0.7711 91.1762 111.7339
0.1835 0.7811 88.6288 112.9141
0.2781 0.7930 86.2194 113.8447
0.3747 0.8054 83.4881 114.7310
0.4734 0.8181 80.9494 115.5891
0.5742 0.8270 78.9435 117.0404
0.6772 0.8337 77.3219 118.8173
0.7824 0.8453 74.2945 119.8258
0.8900 0.8585 71.2342 120.8387
1.0000 0.8677 68.4567 122.3464
at 318.15 K
0.0000 0.7531 97.3125 111.7514
0.0908 0.7668 94.8148 112.3605
0.1835 0.7787 92.0530 113.2621
0.2781 0.7907 89.4967 114.1759
0.3747 0.8034 86.7270 115.0166
0.4734 0.8158 84.0702 115.9292
0.5742 0.8254 81.9252 117.2672
0.6772 0.8329 80.0900 118.9314
0.7824 0.8440 77.0418 120.1105
0.8900 0.8567 73.5534 121.0926
1.0000 0.8659 70.6254 122.6008
and 318.15k.
57
ISSN:0974-8652
The experimental values along with the values calculated theoretically using the relations of Nomoto’s, Free length
theory, Collision factor theory, Vandael ideal mixing, Junjie relation for cyclohexane+ o-xylene at the temperatures 303.15,
308.15, 313.15 and 318.15 K are given in Table 2.
TABLE II – EXPERIMENTAL AND THEORETICAL VALUES IN CYCLOHEXANE +O-XYLENE SYSTEM AT DIFFERENT TEMPERATURES
X1 Uexp UNomoto UFLT UCFT Uvandael UJunjie
303.15K
0.0000 1235.63 1235.62 948.71 1235.63 1235.63 1410.15
0.0908 1237.50 1245.69 961.29 1242.93 1239.63 1411.28
0.1835 1243.65 1255.81 978.13 1251.56 1244.67 1412.70
0.2781 1250.63 1265.98 997.37 1262.46 1250.83 1414.41
0.3747 1259.00 1276.21 1019.29 1275.29 1258.24 1416.41
0.4734 1267.50 1286.50 1042.14 1289.08 1267.04 1418.69
0.5742 1274.75 1296.84 1058.55 1296.19 1277.38 1421.28
0.6772 1284.95 1307.23 1075.25 1300.76 1289.46 1424.17
0.7824 1300.50 1317.68 1103.60 1313.96 1303.52 1427.37
0.8900 1316.25 1328.19 1133.82 1328.94 1319.83 1430.88
1.0000 1338.75 1338.75 1165.65 1338.75 1338.75 1434.71
308.15K
0.0000 1206.13 1206.12 919.67 1206.13 1206.13 1381.26
0.0908 1210.25 1216.63 935.64 1215.08 1210.49 1382.68
0.1835 1218.75 1227.22 654.40 1223.26 1215.90 1384.41
0.2781 1227.50 1237.89 975.37 1233.93 1222.46 1386.46
0.3747 1237.38 1248.64 998.56 1246.16 1230.30 1388.83
0.4734 1245.00 1259.48 1020.15 1258.53 1239.57 1391.54
0.5742 1254.25 1270.41 1038.52 1265.26 1250.45 1394.58
0.6772 1261.88 1281.43 1054.37 1269.63 1263.14 1397.97
0.7824 1278.63 1292.53 1082.48 1281.52 1277.91 1401.71
0.8900 1294.75 1303.72 1113.49 1296.17 1295.06 1405.82
1.0000 1315.00 1315.00 1143.26 1305.00 1315.00 1410.31
313.15K
0.0000 1188.75 1188.75 901.90 1188.75 1188.75 1364.76
0.0908 1192.63 1199.21 919.63 1205.33 1193.13 1366.05
0.1835 1201.88 1209.76 938.78 1217.48 1198.55 1367.67
0.2781 1209.38 1220.40 959.03 1231.88 1205.12 1369.61
0.3747 1219.50 1231.13 982.19 1246.33 1212.96 1371.88
0.4734 1228.75 1241.96 1005.36 1260.61 1222.22 1374.48
0.5742 1237.63 1252.88 1023.52 1267.95 1233.08 1377.43
0.6772 1245.50 1263.89 1038.37 1271.31 1245.75 1380.74
0.7824 1261.88 1275.00 1066.66 1281.33 1260.48 1384.42
0.8900 1278.75 1286.20 1097.81 1292.88 1277.60 1388.47
1.0000 1297.50 1297.50 1125.84 1297.50 1297.50 1392.91
318.15K
0.0000 1168.13 1168.13 879.71 1168.13 1168.13 1346.06
0.0908 1171.88 1178.70 900.00 1188.45 1172.65 1347.27
0.1835 1181.13 1189.39 919.74 1201.52 1178.21 1348.80
0.2781 1188.75 1200.18 939.94 1215.84 1184.92 1350.68
0.3747 1198.00 1211.07 962.47 1230.42 1192.91 1352.91
0.4734 1207.50 1222.07 985.08 1243.71 1202.34 1355.50
0.5742 1216.07 1233.19 1003.75 1251.88 1213.37 1358.46
0.6772 1224.38 1244.41 1019.78 1256.02 1226.23 1361.79
0.7824 1240.13 1255.74 1046.67 1264.70 1241.19 1365.52
0.8900 1259.75 1267.19 1079.23 1274.82 1258.56 1369.65
1.0000 1278.15 1278.75 1107.27 1278.75 1278.75 1374.21
and 318.15k.
58
ISSN:0974-8652
2 2
TABLE III - U exp/ U imx VALUES WITH MOLEFRACTION FOR CYCLOHEXANE
WITH O-XYLENE AT DIFFERENT TEMPERATURES [6] Shipra Baluja & Swathi Oza, J Pure & Appl Ultrason, 24 (2002) 850
[7] Ali A, Yasmin A & Nain A K, Indian J Pure & Appl Phys. 40 (2002)
X1 U2exp/ U2 imx 315
[8] Amalendu Pal, Gurcharan Dass & Harsh Kumar, J Pure & Appl
303.15 K 308.15 K 313.15 K 318.15 K Ultrason, 23 (2001) 10
0.0000 1.0000 1.0002 1.0000 1.0000 [9] Rastogi et al, Indian J Pure & Appl Phys, 40 (2002) 256
0.0908 0.9966 1.0008 0.9992 0.9987 [10] Anwar Ali A, Anil Kumar Nain & Soghra Hyder, J Pure & Appl
Ultrason, 23 (2001) 73
0.1835 0.9984 1.0069 1.0056 1.0050 [11] Nomoto O, J Phys Soc Japan, 4 (1949) 280 and 13 (1958) 1528
0.2781 0.9998 1.0116 1.0071 1.0065 [12] Van Dael W & Vangeel E, Proc int conf on calorimetry and
0.3747 1.0014 1.0161 1.0108 1.0085 thermodynamics, Warasa (1955) 555
0.4734 1.0009 1.0147 1.0107 1.0086 [13] Shipra Baluja & Parsania P H, Asian J Chem, 7 (1995) 417
[14] Gokhale V D & Bhagavat N N, J Pure & Appl Ultrason, 11 (1989) 21
0.5742 0.9961 1.0135 1.0074 1.0045 [15] Junjie Z, J China Univ Sci Techn, 14 (1984) 298
0.6772 0.9932 1.0070 0.9996 0.9970 [16] Prakash A, prakash S & Prakash Q, Proc Nat Acd Sci, 55(A) 11 (1985)
0.7824 0.9956 1.0120 1.0022 0.9983 114.
[17] Sabeson R, Natarajan & Varadha Rajan R, Indian J Pure & Appl Phys,
0.8900 0.9947 1.0125 1.0018 1.0019 25 (1987) 489
1.0000 1.0000 1.0154 1.0000 1.0000 [18] Vogel A I, A text book of practical organic chemistry, 5th Edn (John
Wiley, New York)1989
[19] Riddick J A, Bunger W B & SokanoT K, techniques in chemistry, Vol
The ratio U2exp/ U2 imx is used as an important tool to measure
2, organic solvents, 4th Edition (John Wiley, New York) 1986
the non-ideality in the mixtures, especially in those cases [20] Satyanarayana N, Satyanarayana B & Savitha jyostna T, J Chem Eng
where the properties other than sound velocity are not Data, 52 (2007) 405.
known22. A perusal of Table 2 indicate large deviations from [21] Satyanarayana B, Savitha Jyostna T & Satyanarayana N, Indian J Pure
& Appl Phys, 44 (2006) 587.
ideality, which may be due to the existence of strong tendency [22] Viswanatha Sarma A & Viswanatha Sastry J, J Acous Soc Ind, Vol
for the formation of association in liquid mixtures through XXVII (1999) 309
hydrogen bonding as reported by Shukla et al 23. The [23] Shukla B D, Jha L K & Dubey G P, J Pure & Appl Ultrason, 13 (1991)
deviations between theoretical and experimental value of 72 Nikkam P S, Jagdale B S, Sawant A B & Mehdi hasa, J Pure & Appl
Ultrason, 22 (2000) 115.
ultrasonic velocities decrease with increase of temperature due
to breaking of hetero and homo molecular clusters at higher
temperatures24. On increasing the temperature, the ultrasonic
velocity values decreases in the binary liquid mixture. This is
probably due to the fact that the thermal energy activates the
molecule, which would increase the rate of association of
unlike molecules. Hence the complex formation through
hydrogen bonding will occur9. In the present work it is evident
to say experimental values of ultrasonic velocities are nearer
to ultrasonic velocities as calculated from Vandael ideal
mixing relation followed by CFT theory and Nomoto’s
relation. Further from Table.3 U2exp/U2imx indicates small
deviations from ideality, that means no strong interactions in
liquid mixtures through hydrogen bonding. This is probably
due to weak interactions in the liquid mixture.
CONCLUSIONS
An estimation of ultrasonic velocities of the binary mixtures at
four temperatures reveals that it agrees well with Vandael
ideal mixing relation followed by CFT and Nomoto’s relation.
ACKNOWLEDGEMENT
We would like to express our thanks to referees for valuable
comments that improved the paper.
REFERENCES
[1] Prausnitz J M, Lichenthalar & Azevedo, Molecular Thermodynamics of
fluid phase equilibria, second edition, (Prentice-Hall Inc, Englewood
Cliffs, New Jersey) (1986)
[2] Acree W E (Jr), thermodynamics properties of non electrolyte solutions
(Academic Press, New York)(1984)
[3] Rodriguez S, Lafuente C, Artigas H, Royo F M & Urieta J S, J Chem
Thermodynamics, 31 (1999) 139
[4] Naidu P S & Ravindra Prasad K, J Pure Appl Ultrason, 24 (2002) 18.
[5] Rao T S, Veeraiah N & Rambabu C, Indian J Pure & Appl Phys,
40(2002) 850
and 318.15k.
59
ISSN:0974-8652

MATHEMATICAL IDEAS
ISSN:0974-8652 / www.ijcmi.webs.com / VOL 2-NO 2-PP 60-64 (2010)
OVER VIEW TO IMPLEMENTATION OF ROBOTICS WITH VOICE RECOGNITION
Ande Stanly Kumar@, Dr.K.Mallikarjuna Rao#, Dr.A.Bala Krishna$, B.Venkatesh^
@
Asst.Professr, Sri Venkateswara Institute of Science Information Technology,Tadepalligudem. Stanlykumar.ande@gmail.com,
#
Professor, JNTU College of Engineering, Kakinada.
$
Professor,SRKR Engineering College,Bhimavaram.
^
Asst.Professr, Sri Venkateswara Institute of Science Information Technology,Tadepalligudem. venkatesh.visit@gmail.com
ABSTRACT
Automatic speech recognition by machine has been a goal of a research for a long time. Speech recognition is the process of
converting an acoustic signal, captured by a microphone or a telephone, to a set of words. The recognized words can be the final
results, as for applications such as commands & control, data entry, and document preparation. They can also serve as the input to
further linguistic processing in order to achieve speech understanding. There are many works carried out in this area. The speech
recognition system has also been implemented on some particular devices. Some of them are personal computer (PC), digital signal
processor, and another kind of single chip integrated circuit. In this paper we propose voice recognition to control robot.
Key words: Euclidean square distance, LPC, Voice recognition, Finger print.
Introduction: They h introduced a novel method for isolating the rove of

higher order polynomials in Linear predictive systems. Y.M.
The term "voice recognition" is sometimes used to refer to Lam et al. implemented [3] fixed point implementations for
speech recognition where the recognition system is trained to a speech recognition, they achieved recognition rate of 81.33%.
particular speaker - as is the case for most desktop recognition SoshiIba et al. proposed [4] the framework takes a three-step
software, hence there is an element of speaker recognition, approach to the robot programming i.e multi-modal
which attempts to identify the person speaking, to better recognition, intention interpretation, and prioritized task
recognize what is being said. Speech recognition is a broad execution.
term which means it can recognize almost anybody's speech -
such as a call-centre system designed to recognize many In previous works, speech recognition system was
voices. Voice recognition is a system trained to a particular implemented [5] on ATMEL 89C51RC microcontroller to
user, where it recognizes their speech based on their unique control the movement of Wheelchair. They used the LPC
vocal sound. model for speech recognition and achieved recognition rate of
78.57%. Thiang implemented [6] the speech recognition for
The first speech recognizer appeared in 1952 and controlling movement of Mobile Robot ATmega162
consisted of a device for the recognition. Microcontroller. Used Techniques were Linear Predictive
Coding (LPC) combined with Euclidean Squared Distance and
Literature:
Hidden Markov Mode (HMM). In this project, highest
Treeumnuk & Dusadee, implemented [1] the Speech recognition rate achieved was 87%. Stanly & Ande
Recognition on FPGA with segmentation technique. implemented [7] Voice Recognition Robotic Car with filters
Sriharuksa & Janwit implemented [2] a complete design and and finger print conversion method. Coming to this project, it
layout of an ASIC Design of Real Time Speech Recognition. describes continuation work to the previous works.
Over View To Implementation Of Robotics With Voice Recognition
60
ISSN:0974-8652
Speech is a natural source of interface for human– In order to analyze speech, we needed to look at the
machine communication, as well as being one of the most frequency content of the detected word. To do this we used
natural interfaces for human–human communication [8]. several 4th order Chebyshev band pass filters. To create 4th
However, environmental robustness is still one of the main order filters, we cascaded two second order filters using the
barriers to the wide use of speech recognition. Speech following "Direct Form II Transposed" implementation of a
recognition performance degrades significantly under varying difference equations.
environmental conditions for many application areas.
In this paper, speech recognition system is

implemented to recognize the word used as the command for
controlling the movement of robot. The proposed novel
method will increase the recognition rate. Especially monitor
the need of Embedded Systems in Industrial applications to
control the movement of either simple or Bulky devices. There
are two approaches used to recognize the speech signal. The
first approach is Linear Predictive Coding combined with
Euclidean Squared Distance (ESD). In this approach LPC is
used as the feature extraction method and Euclidean Squared
Where the coefficient a’s and b’s were obtained through
Distance is used as the recognition method. The second
Matlab using the following commands.
approach is Hidden Markov Model, which is used to build
[B,A] = cheby2(2,40,[Freq1, Freq2]);
reference model of the words and also used as the recognition
(Where 2 defines a 4th order filter, 40 defines the stop band
method. Feature extraction method used in the second
ripple in decibels, and Freq1 and Freq2 are the normalized
approach is a simple segmentation and centroid value. Both
cutoff frequencies).
approaches work on time domain. Experiments have to do in
[sos2, g2] = tf2sos (B2, A2,'up','inf');
several variations of observation symbol number and number
of samples. The hoist & crane can move in accordance with
the voice command. Maximum recognition rate will be Fingerprint Calculation:
expected here by introducing a novel method.
Due to the limited memory space on the Mega32, we
Design: needed a way to encode the relevant information of the spoken
word. The relevant information for each word was encoded
in a “fingerprint”. To compare fingerprints we used the
Euclidean distance formula between sampled word fingerprint
and the stored fingerprints to find correct word.
Euclidean distance formula is:
Fig. 1. Layout robotic system
The design had been done in the field of robotics and there P=( ) and Q = ( )
exists a line follower robots, sensor robots and used speech to
control a robot. It would make a robot which obeys human Where P is a dictionary fingerprint and Q is the sampled word
speech commands and performs errands. fingerprint and p i and q i are the data points that make up the
fingerprint. To see if two words are the same we compute the
Mathematics for speech analysis: Speech Analysis: Euclidean distance between them and the words with the
minimum distance are considered to be the same. The
61
ISSN:0974-8652
formula above requires squaring the difference between the

two points, but since we are using fixed point arithmetic, we
found that squaring the difference produced too large of a
number causing our variables to overflow. Thus we
implemented a "pseudo Euclidean distance calculation" by
moving the sum out of the square root reducing the equation to
D=
PWM (Pulse Width Modulation) duty cycle calculation:
The motors in the robot were measured to have a 50 Hz PWM

frequency and movement was controlled by varying the duty
cycle from 5% to 10%. To generate the PWM signals we used
timer/counter1 in phase correct mode. The top value of
timer/counter 1 was set to be 20000 and using a /8 pre scalar Fig.2.flow chart for voice recognition
the PWM signal was set to have a frequency of 50Hz =
16MHz/(8*2*20000). To calculate the duty cycle the The Basic algorithm of our code was to check the
following equation was used OCR1x = (20000 - 40000*duty ADC input at a rate of 4 KHz. If the value of the ADC is
cycle). Where OCR1x is the value in the output compare greater than the threshold value it is interpreted as the
register 1 A or B. beginning of a half a second long word. The sample word
passes through 8 band pass filters and is converted into a
Hardware/Software tradeoffs: fingerprint. The words to be matched are stored as fingerprints
in a dictionary so that sampled word fingerprints can be
The signal coming out of the microphone needed to compared against them later. Once a fingerprint is generated
be amplified. We had two different versions of operational from a sample word it is compared against the dictionary
amplifier, LMC 711 and LM 358. The LMC711 has a slew fingerprints and using the modified Euclidean distance
rate of 0.015 V/µ s, on the other hand LM 358 has 0.3V/µ s. calculation finds the fingerprint in the dictionary that is the
The LM358 has a better slew rate and it gave us better closest match. Based on the word that matched the best the
response to input signals so we used it when we designed our program sends a PWM signal to the car to perform basic
amplification circuit. operations like left, right, go, stop, or reverse.
The signal processing of speech requires lot of Initial-Threshold Calculation:

computations, which implies we need fast processor, but we
had to operate at 16 M Hz. In order to minimize the number of At start up as part of the initialization the program reads the
cycles we used filtering the audio signal we had to write most ADC input using timercounter0 and accumulates its value 256
of the code in assembly. We wrote all of 10 digital filters in times. By interpreting the read in ADC value as a number
assembly which made them very efficient and significantly between 1 to 1/256, in fixed point, and accumulating 256
improved our performance over a C code implementation. times. The average value of ADC was calculated without
doing a multiply or divide. Three average values are taken
each with a 16.4msec delay between the samples. After
receiving three average values, the threshold value is to be
four times the value of the median number. The threshold
value is useful to detect when a word has been spoken or not.
62
ISSN:0974-8652
Fingerprint Generation: first 2 KHz since this usually contains the first and second
speech formants, (resonant frequencies). This also allowed us
The program considers a word detected if a sample value from to sample at 4 KHz and gave us almost enough time to
the ADC is greater than the threshold value. Every sample of implement 10 filters. We thought we needed ten filters each
ADC is typecast to an int and stored in a dummy variable A with approximately a 200 Hz bandwidth so that we would
in. The A in value passes through 8 4th order Chebyshev band have enough frequency resolution to properly identify words.
pass filters with a 40 dB stop band for 2000 samples (half a Originally we had 5 filters that spanned from 0 – 4 KHz and
second) once a word has been detected. When a filter is used were sampling at 8 KHz, but this scheme did not produce
its output is squared and that value is accumulated with the consistent word recognition.
previous squares of the filter output. After 125 samples the
accumulated value is stored as a data point in the fingerprint of Fingerprint Comparison:
that word. The accumulator is then cleared and the process is
begun again. After 2000 samples 16 points have been Once the fingerprints are created and stored in the
generated from each filter, thus every sampled word is divided dictionary when a word was spoken, it was compared against
up into 16 parts. Our code is based around using 10 filters the dictionary fingerprints. In order to do the comparison, we
and since each one outputs 16 data points every fingerprint is called a lookup() function. The lookup() function did a pseudo
made up of 160 data points. Euclidean distance formula by calculating the sum of the
absolute value of the difference between each sample finger
Implementation of Filter: print a finger print from the dictionary. The dictionary has
multiple words in it and the lookup went through all of them
and picked the word with the smallest calculated number.
We had originally used the square of the correct Euclidean
distance calculation, d = Σ(pi – qi) 2. The words we finally
used in our dictionary were Let's Go, (sound of a finger
snapping), daiya [right – in Hindi], rukh [stop – in Hindi],
peiche [back – in Hindi]. We had originally used English
words, go, left, right, stop, and back, but many of these words
seemed to be very similar in frequency as far as our algorithm
was concerned. We then went to vowels and had better
success, but we still wanted to use words that were directions
and so we went to Hindi The set of words that we used were
mostly orthogonal, but in Hindi left is baiya, which sound very
similar to daiya and so that could not be used. We had
previously had success with snapping so we used that for left.
Fig.3. finger print implementation
PWM signal to move ROBOT:
Filter Implementation:
Once a word is recognized, its time to perform an
We chose a 4th order Chebyshev filter with 40 dB action based on the recognized word. To perform an action we
stop band since it had very sharp transitions after the cutoff generated a PWM signal using timercounter1. Control of the
frequency. We designed 10 filters a low pass with a cutoff of PWM signal generation is done by the car control() function.
200 Hz, a high pass with a cutoff of 1.8 KHz, and eight band For our robot, we needed to generate two different PWM
passes that each had a 200 Hz bandwidth and were evenly signals, one for moving the car front/back and another one to
distributed from 200Hz to 1.8 KHz. Thus we had band pass steer left or right. We also need to send a default PWM signal
filters that went from 200-400 Hz, 400-600, 600 – 800 and so to pause a robot. We chose timercounter1 because it has two
on all the way to the filter that covered 1.6 Khz – 1.8 Khz. different compare registers, OCR1A and OCR1B and can
We designed our filters in this way because we felt that most output two unique PWM signals. We used Phase correct mode
of the important frequency content in words was within the
63
ISSN:0974-8652
to generate PWM signals because it is is glitch free, which is [2]. Thiang, “Implementation of Speech Recognition on
better for the motor. MCS51Microcontroller for Controlling Wheelchair”
proceedings of International conference on Intelligent and
To find out a frequency and a duty cycles at which advanced systems. Kuala Lumpur, Malaysia, 2007
car turns forward/backward and left/right, we attached an [3]. Y.M. Lam, M.W. Mak, and P.H.W. Leong , “Fixed point
oscilloscope probe to a car’s receiver. We sent different implementations of Speech Recognition Systems”.
signals to the receiver using the car’s remote control an d Proceedings of the International Signal Processing
measured the frequency and duty cycle for different motions. conference.Dallas. 2003.
From the measurements, we found that car PWM frequency [4]. Treeumnuk, Dusadee. (2001). Implementation of Speech
was 50Hz (period of 20ms) and had the following properties. Recognition on FPGA.Masters research study, Asian Institute
of Technology, 2001).Bangkok: Asian Instituteof Technology.
Conclusion: [5]. Soshi Iba, Christiaan J. J. Paredis, and Pradeep K. Khosla.
“Interactive MultimodalRobot Programming”. The
The Embedded systems design covers a very wide International Journal of Robotics Research (24), pp 83 –104,
range of microprocessor designs Our task is to design a 2005.
control module for a robot. The robot is a simple two wheel [6]. Sriharuksa, Janwit. (2002). An ASIC Design of Real Time
robot that uses two stepper motors for driving. The robot can Speech Recognition.(Masters research study, Asian Institute of
be programmed to drive autonomously a certain path. A list of Technology, 2002). Bangkok: AsianInstitute of Technology.
driving commands are first downloaded from a PC to the
robot, after which the robot will drive automatically through [7]. Lawrence Rabiner, and Biing Hwang Juang,
the program and to provides a framework to specify a system. Fundamentals of Speech Recognition.Prentice Hall, New
At the beginning of our project, we set a goal to recognize five Jersey, 1993.Speech recognition by machine. By William
words, at the end of project we got ive words to be recognized. Anthony Ainsworth, Institution of Electrical Engineers.
However our five words needed to be orthogonal to each other
[8] Andre Harison and Chirag Shah Voice recognition by
because our filters were not giving a high enough resolution
robot.
and inaccuracy in fingerprint calculations due to using fix
point arithmetic made the lookup function to be error prone. [9]. www.speechrecognition.com -/ united states.
As a result, we had to pick various different words that sound
apart. If we had to do this again instead of trying to use the [10]. Frank Vahid and Tony Givargis, Embedded System
Euclidean distance formula to match words we would like to Design: A Unified Hardware/Software Approach.
try do perform a correlation of the two fingerprints. A
correlation is less sensitive to amplitude differences and is a
better way of identifying patterns between two objects. If we
had faster process chip, we could modified our algorithm to
add more filters, perform Fourier transform, or floating point
arithmetic in order to improve our results.
ACKNOWLEDGEMENT
We would like to express our thanks to referees for valuable

comments that improved the paper.
REFERENCES:
[1]. Thiang, “Limited speech recognition for controlling

movement of Mobile Robot Implemented on ATmega162
Microcontroller” proceedings on International conference on
Computer and Automation Engineering.2009.
64
ISSN:0974-8652

MATHEMATICAL IDEAS
NOVELTY OF EXTREME PROGRAMMING

Ch.V.Phani Krishna@, S.Satyanarayana#, K.Rajasekhara Rao$
@
Sana Engg. College, Kodad, Phanik16@yahoo.co.in
#
Dept of Mathematics, Sri Venkateswara Institute of Science & Information Technology, Tadepalligudem,
s.satyans1@gmail.com
$
K.L. College of Engineering, Vijayawada, rajasekhar.kurra@klce.ac.in
ABSTRACT
Extreme Programming is one of the most discussed to topics in the software development community. In this
paper, we discussed the fundamentals of Extreme Programming, how Extreme Programming distinguished from other
methodologies, how Extreme Programming addresses the risks encountered in software development, the values and basic
principles of Extreme Programming, advantages and disadvantages of Extreme Programming. Then we will see how
Extreme Programming uses a set of practices to build an effective software development team that produces quality
software in a predictable and repeatable manner.
Introduction: Simple design that constantly evolves to add

needed flexibility and remove unneeded
complexity.
Extreme Programming was visualized and
developed to address the specific needs of software Putting a token (nominal) system into
development conducted by small teams in spite of vague production quickly and growing it in whatever
rapidly changing requirements. directions prove most valuable.
This new light weight methodology challenges Reasons why XP is controversial: -
many conventional principles & opinions, including the
Does not force team members to specialize
long held assumption that the cost of changing a piece of
because - every XP programmer participates
software necessarily rises dramatically over the courses
in (all of these practices all the critical
of time. XP recognizes that projects have to work to
activities everyday.
achieve cost reduction and make use of savings once
they have been earned. Do not conduct complete – up – front analysis
and design because – XP project make
Fundamentals of XP:
analysis and design decisions through
Distinguishing between the decisions to be development.
made by business interests and those to be
Delivering business value is the heart beat that
made by project stake holders.
drives XP projects.
Writing unit tests before programming and
Do not write and maintain implementation
keeping them running at all times.
documentation because – communication in
Integrating and testing the whole system several XP projects occurs face to face or through
times. efficient tests or carefully written coding.
Producing all software in pairs (pair XP makes two sets of promises:
programming)
To Programmers:
Novelty Of Extreme Programming
65
ISSN:0974-8652
XP promises that they will be able to work on

things that really matter, everyday. They
Novelty of XP:
won’t have to face scary situations alone.
They will be able to do every thing in their Putting all the practices under one umbrella
power to make their system successful. They
Making sure they are practiced as thoroughly as
will make decisions that they can make best
possible
and they won’t make decisions that they can
make best and they won’t make decisions they Making sure the practices support each other to
are not best qualified to make. the greatest possible degree.
To Customers and Managers: XP addressing the risks encountered in software
XP promises that they will get the most development:
possible value out of every programming The basic problem of software development is
week. Every few weeks they will able to see risk.
concrete progress on goals they care about.
They will be able to change the direction of (1) Schedule slips: - short releases
project in the middle of development without Within an iteration, XP plans with 1-
incurring exorbitant costs. 3 days tasks, so the team can solve problems
XP promises to reduce project Risk, even during iteration. XP calls for
improve responsiveness to business changes, implementing the highest priority features
improve productivity throughout the entire first, so any features that slip past the release
life of a system & add fun to building will be of lower value
software in teams – all at the same time. (2) Project cancelled: - same as above
XP asks the customer to choose the
XP is distinguished from other methodologies by: smallest release that makes the most business
sense, so there is a less chance to go wrong &
Its early, concrete, continuing feed back from the value of the software is greatest.
short cycles
(3) System goes sour: - Testing.
Its incremental planning approach, which
comes up with an overall plan that in expected Repeated testing in XP ensures a
to evolve through the life of the project quality base line.
Its ability to flexibly schedule the (4) Defect Rate: Testing (both programmer as
implementation of functionality, responding well as customer perspectives) Programmer
to changing business needs. (Testing function – b y – function)
Its dependence, on automated tests written by Customer (program feature – by – Program

programmers and customers to monitor the feature)
progress of development and to catch defects (5) Business Misunderstood: - On – site
early. customer
Its dependence on oral communication, tests Specification of project is
and source code to communicate system continuously refined during development, so
structure and goal learning by the customer & the team can be
Its dependence on an evolutionary design reflected in the software.
process that lasts as long as the system lasts (6) Business changes: - Short Releases:
Its dependence on the close collaboration of XP shortens release cycle, so there is
programs with ordinary skills. less change during the development of a
Its dependence on practices that work with both single release. During the release the
the short-term instincts of programmers and customer is welcome to substitute a new
the long term interests of the project. functionality for functionality not yet
completed.
(7) False feature Rich:
66
ISSN:0974-8652
XP insists that only the highest Scope: - Less scope makes it possible to deliver
priority tasks are addressed. better quality. It also let us delivers sooner or
cheaper.
(8) Staff turnover: - Pair programming
XP asks programmers to accept
responsibility for estimating and completing Four values:
their own work, gives them feedback about
Project will be successful when the team
the actual time taken, so that their estimates
follows a style that celebrates a consistent set
can improve.
of values that serve human and commercial
Thus there is less chance of or a needs:
programmer to get frustrated by being asked
Communication
to do the obviously impossible.
Simplicity
XP development cycle:
Feedback
Pair of programmers program together.
Courage
Development is driven by tests. Until, all the
tests run, the process of adding functionality Communication:
is not succeeded. Then coding activity begins.
XP aims to keep the right
Pairs don’t just make the test cases run. They communications flowing by employing many
also evolve the design of the system. Changes practices that can’t be done without
are not restricted to any particular area. Pairs communicating. They are practices that make
add value to the analysis, design, short term sense, like unit, testing, pair
implementation, testing of the system. They programming and task estimator. The effect of
add that value wherever the system needs it. testing, pairing and estimating is that
programmers and customers and managers
Integration immediately follows development,
have to communicate.
including integration testing.
This doesn’t mean that communications
Four Variables:
don’t sometimes get logged in an XP project.
There are four control variables in software People get scared, make mistakes, get
development model distracted XP employs a coach whose job is to
notice when people are not communicating
Cost
and reintroduce them.
Time
Simplicity:
Quality
The second XP value is simplicity. XP is
Scope making a bet. It is betting that it is better to do
a simple thing today and pay a little more
Interactions Between the variables: tomorrow to change it if it needs it than to do
Cost: - Giving a project too little money and it a more a complicated thing today that may
won’t be able to solve the customer’s business never be used any way.
problem, on the other band too much money
Simplicity and communication have a
too soon creates more problems than it solves.
wonderful mutually supporting relationship.
Time: - More time to deliver can improve
Feed back:
quality and increase scope. Give a project too
little time and quality suffers, with scope, time The third value in XP is feedback.
and cost not far behind. Concrete feed back about the current state of
the system is absolutely priceless. Optimism
Quality: - Quality is an important control
is an occupational hazard of programming.
variable we can make very short – term gains Feedback is the treatment. The programmers
by deliberately sacrificing quality, but the cost have minute – by – minute feed back about
– human, business, and technical is enormous.
the state of their system. When customers
write new ‘stories” (descriptions of feature),
67
ISSN:0974-8652
the programmers immediately estimate them, problem today, and trust your ability to add
so the customers have concrete feedback complexity in the future where you need it.
about the quality of their stories.
Incremental change:
The person who tracks progress watches
Any problem is solved with a series of the
the completion of the tasks to give the whole
smallest changes that make a difference.
team feedback about whether they are likely
Hence the adoption of changes in XP must be
to finish everything they set out to do in a
taken in little steps.
span of time. The customers and testers write
functional tests for all the stories implemented Embracing change:
by the system. They have concrete feed back
about the current state of their system. The The best strategy is the one that preserves the
most obtains while actually solving your most
customers review the schedule every
pressing problem.
frequently to see if the terms over all velocity
matches the plan and to adjust the plan. The
system is put into production as soon as it
makes sense, to the business can begin to Quality work:
“feel” what the system is like in action & Of all the four Project variables – Quality is
discover how it can best be exploited. not really a free variable. The only possible
Concrete feedback works together with values are “excellent” and insanely excellent
communication and simplicity. depending on whether lives are at stake or not
Courage: otherwise we don’t enjoy our work & the
project goes down the drain.
When combined with communication,
simplicity and concrete feedback, courage Some less central principles:
becomes extremely valuable. Communication Teach learning
supports courage because it opens the
possibility for more high – risk, high – reward Small initial investment
experiments. Simplicity supports courage. Play to win
Concrete feedback supports courage because
of feeling much safer trying radical surgery on Concrete experiments
the code. Open, honest communication
Basic principles: Work with people’s Instincts, not against them
The fundamental principles are Accepted Responsibility
Rapid feed back Local Adaptation
Assume simplicity Travel light
Incremental change Honest Measurement
Embracing change Teach learning:
Quality work We will focus on teaching strategies for
Rapid feed back: learning how much testing you should do.
Also how much design refactoring and
Learning psychology teaches that the time everything else you should do.
between an action and its feedback is critical
to learning. So, our principle to get feed back, Small investment (Initial):
interpret it and put what is learned back into Too many resources too early in a project
the system as quickly as possible. are a recipe for disaster. Tight budgets force
Assume simplicity: programmers & customers to pare
requirements and approaches. Resources can
Treat every problem as if it can be solving be too tight. If you don’t have the resources
with ridiculous simplicity. This is the hardest to solve even one interesting problem, then
principle for programmers to swallow. XP the system you create is guaranteed not to be
says to do a good job of solving today’s
68
ISSN:0974-8652
interesting. If you have some one dictating going along. Along the way a person told
scope, dates, quality & cost, there you are what to do will find a thousand of expressing
unlikely to be able to navigate to a successful their frustration, most of them to the detriment
conclusion. of the team & many of them to the detriment
of the person
The alternate is that responsibility be
accepted, not given you are past of a team,
Play to win: and if the team comes to the conclusion that a
certain task need doing, someone will choose
The difference between playing to win to do it, no matter how odious.
and playing not to lose, most software
development I see is played not to lose. Lots Local adaptation:
of paper gets written. Lots of meetings are
This is an application of accepted
held. Everyone is trying to develop “by the
responsibility to your development process.
book”, not because it makes any particular
Adopting XP means that you get to decide
sense, but because they want to be able to say
how to develop i.e. deciding on something
at the end’ that it wasn’t their fault, they were
today and being aware of whether it still
following the process. Software development
works tomorrow. You have to change and
played to win does every thing that helps the
adapt.
team to win and doesn’t do anything that
doesn’t help to win. Travel light:
Concrete experiments: The artifacts to be maintained are
Every time you make a decision and you Few
don’t test it, there is some probability that the
Simple
decision is wrong, the more decisions you
make the more these risks compound. The Valuable
result of a discussion of requirements should
XP team becomes intellectual nomads, always
also be a series of experiments. Every
abstract decision should be tested. prepared to quickly pack up the tents and
follow the herd. XP team gets used to
Open, Honest Communication: traveling light. They don’t carry much in the
way of baggage except what they must have
Programmers have to be able to tell each
to keep producing value for the customer –
other where there are problems in the code.
tests and code.
They have to be free to express their fears,
and get support. They have to be free to Honest measurement:
deliver bad news to customers and
management to deliver it early, and not be Our quest for control over software
development has led us to measure, which is
punished.
fine, but it has led us to measure at a level of
Works with people’s Instincts, not against detail that is not supported by our instruments.
them:
Practices of XP:
People like winning. People like
interacting with other people. People like The values of XP are implemented by
learning. People like being past of a team. employing 12 practices as elucidated by Kent
Beck.
People like being in control – people like
being trusted. XP celebrates what Planning game:
programmers seen to do when left to their
own devices, with just enough to keep the Determining the scope of project and releases
whole process on track, XP matches by combining business priorities with the
observations of programmers in the wild. technical estimates according to the changing
requirements.
Accepted Responsibility:
Primate dominance displays work only so
long in getting people to act like they are
69
ISSN:0974-8652
Small releases: The customer if located at the same site as a

domain expert to help the programmer’s team
Enforce a simple system into production
in the production of system.
quickly, and then release new versions on
short cycles. Coding standards:
System Metaphor: Programmers write all code in accordance
with the standards agreed upon by the
A shared story which helps the programmers
development team to ensure that
as well as customers understand the basic
communication is made through code.
elements on which the system works and their
relationships. Jeffries developed them further and has 13
practices.
Simple Design:
Whole Team:
Keeping the system design as simple as
possible and remove excess complexity as All people who take part in the project gather
soon as possible. at one place to develop the system as a team.
Testing: In addition to the above said practices there
are certain implicit practices.
Continuously writing and running required
tests, each time a new code is written are Caves and commons
changing the existing code including unit
Fixed iterations and engineering tasks
testing and customer written functionality
testing. Write it on a card (RDP Technique)
Refactoring: Spike Solutions
Improving the design of project without All tests all the time
changing the functionality of there existing
code by removing the duplication of the code Promiscuous Pairing
and by improving communication, Yesterday’s weather
simplification and flexibility.
Track velocity and track progress
Pair Programming:
Regression test
Writing code with two programmers at one
machine . There are certain misconceptions regarding
XP. But the real truth is ….
Collective Code Ownership:
No written design documentation
All programmers accepting responsibility for
all code therefore being able to make changes • Truth: no prescribed standards for how much
to any piece of code at any time when or what kinds of docs are needed.
necessary. No design
Continuous Integration: • Truth: minimal explicit, upfront design:
Soon after completing a task, the system design is an explicit part of every activity
should be integrated and run several times through every day.
continuously. XP is easy
40-Hour week: • Truth: although XP does try to work with the
To keep programmers active, creative and natural tendencies of developers, it requires
fresh, no programmer should work more than great discipline and consistency.
40-hours per week. No programmer should do XP is just legitimized hacking
more than a week’s overtime in a row.
• Truth: XP has extremely high quality
On-site Customer: standards throughout the process.
XP is the one, true way to build software
70
ISSN:0974-8652
• Truth: It seems to be a sweet spot for certain Keep the customer in charge of what is
kinds of projects. developed and when.
Advantages: To quickly demonstrate the benefits of XP,
implement it first on a new project with no
XP is indeed flexible. Changes in the legacy code.
priorities can be done repeatedly with very
little notice and customers will be served with Conclusion:
what they requested.
Tacit knowledge and communication among all team
Results and output can be forecasted to members are highlighted in XP. XP practices such as
customers at every schedule. Pair Programming and extensive testing further reinforce
this insight, as well as minimizing documentation. XP
XP team can inculcate satisfaction in puts a high premium on customer satisfaction. Taking
customer by revealing short releases and the customer within the team and receiving feedback
adapting changes requested by the customer. frequently are ways to accomplish it. This way
The unit tests written by the developer customer’s suggestions can be taken into account
team and acceptance tests written by the throughout the development project. The customer also
customer increases the degree of confidence participates in testing. Thus, XP is developed to provide
in the product. a favorable setting for programmers to be able to respond
rapidly to changing customer requirements.
Disadvantages:
Extreme flexibility exerts heavy responsibility
on the customers to produce strategic plans. ACKONOWLEDGEMENT
Responding to sudden changes is part and
parcel of XP, but adopting changes for every We would like to express our thanks to referees for valuable
iteration is confusing and may not be a sound comments that improved the paper.
business practice.
The lack of emphasis on documentation within REFERENCES:
XP does not take into account end user needs
for documentation, including user guides, [1]. Lindstrom, L., & Jeffries, R. (2004). Extreme
integration kits, reference texts and fact programming and agile software Methodologies.
sheets. Information Systems Management, 21(3), 41.
[2]. Beck, Kent and Martin Fowler. Planning Extreme
Programming, Addison-Wesley, Boston MA,
Problems encountered in the implementation of XP: October 2000.
Overly engineering [3]. Armitage, J. (2004). Are agile methods good for
design?`. Interactions, 11(1), n/a. Retrieved
Overly complex integration September 9, 2006, from Proquest database.
Unrepresentative acceptance testing [4]. Jeffries,R.What is extreme programming?
http://www.xprogramming.com
Coding assistant [5]. Osamu Kobayashi., Mitsuyoshi Kawabata., Makoto
Hard to test software Sakai., Eddy Parkinson., Analysis of the Interaction
between Practices for introducing XP effectively,
Obtuse specification ACM, 2006.
Recommendations: [6]. Grenning.,J., Launching Extreme Programming at a
Process Intensive Company, IEEE Software, Vol 18,
Have a contingency plan to manage resistant No.6, pp. 27-33, 2001.
participants who are not won over to XP. [7]. William.A.Wood., William.L.K., Exploring XP for
Make the necessary physical changes to the Scientific Research, IEEE Software, Vol.20, No. 3,
work place to foster too key tenets of XP: Pair pp30-36, 2003.
Programming and constant customer - [8]. Martin Lippert., Stefan Roock., Adopting XP to
developer communication. Complex Application Domains, ACM, 2001.
71
ISSN:0974-8652
[9]. Glenn Vanderburg., A Simple Model of Agile

Software Process –or- Extreme Programming
Annealed, ACM, 2005.
[10]. Kuppuswami, S., Vivekanandan K., and Paul Rodrigues
(2003): A Sys-tem Dynamics Simulation Model to Find the
Effects of XP on Cost of Change Curve. In proceedings of
Fourth International Conference on Extreme Pro-
gramming and Agile process in Software Engineering,
(XP2003), May 25 – 29, 2003, Genova, Italy.
72
ISSN:0974-8652

MATHEMATICAL IDEAS
RADIATION EFFECTS ON MHD FREE CONVECTION FLOW PAST A SEMI-INFINITE MOVING VERTICAL
POROUS PLATE WITH SORET AND DUFOUR EFFECT
G.Venkata Ramana Reddy@ and Dr. A.Rami Reddy#

@
Assistant Professor, LBR College of Engineering, Mylavaram, Krishna,A.P. Email: gvrr2020@rediff.com
#
Associate Professor, LBR College of Engineering, Mylavaram, Krishna,A.P
ABSTRACT
In this paper, we deal with the interaction of Soret and Dufour effects on steady MHD free convection flow in a
porous medium with dissipative fluid has received little attention. Hence, the object of the present chapter is to analyze the
Soret and Dufour effects on steady MHD free convection flow past a semi-infinite moving vertical plate in a porous
medium with viscous dissipation. The governing equations are transformed by using similarity transformation and the
resultant dimensionless equations are solved numerically using the Runge-Kutta method with Shooting technique. The
effects of various governing physical parameters on the fluid velocity, temperature, concentration, skin-friction
coefficient, Nusselt number and Sherwood number are shown in figures and tables and analyzed in detail.
Key words: MHD free convection, porous medium, vertical plate, and Nusselt number
1. INTRODUCTION two dimensional natural convective flow of an incompressible,

Combined heat and mass transfer (or double- electrically conducting fluid along an infinite vertical porous
diffusion) in fluid-saturated porous media finds applications in plate embedded in a porous medium. Helmy [7] studied MHD
a variety of engineering processes such as heat exchanger unsteady free convection flow past a vertical porous plate
devices, petroleum reservoirs, chemical catalytic reactors and embedded in a porous medium. Elabashbeshy [8] studied heat
processes, geothermal and geophysical engineering such and mass transfer along a vertical plate in the presence of
moisters migration in a fibers insulation and nuclear waste magnetic field. Chamkha and Khaled [9] investigated the
disposal and others. Double diffusive flow is driven by problem of coupled heat and mass transfer by
buoyancy due to temperature and concentration gradients. magnetohydrodynamic free convection from an inclined plate
Bejan and Khair [1] investigated the vertical free convection in the presence of internal heat generation or absorption.
layer flow in a porous media owing to combained heat and In the above all studies, the level of concentration of
mass transfer. Lai and Kulacki [2] used the series expansion foreign mass assumed very low, so that the Soret and Dufour
method to investigate coupled heat and mass transfer in effects can be neglected. However, expectations are observed
natural convection from a sphere in a porous medium. The therein. The Soret effect, for instance, has been utilized for
suction and blowing effects on free convection coupled heat isotropic separation, and in mixture between gases with very
and mass transfer over a verrtical plate in a saturated porous
medium was studied by Raptis et al. [3] and Lai and Kulacki
light molecular weight ( H2 , He ) and of medium molecular
[4], respectively. weight ( N 2 , air ) . The Dufour effect was found to be of order
Magnetohydrodynamic flows have applications in
of considerable magnitude such that it cannot be ignored [10].
meteorology, solar physics, cosmic fluid dynamics,
The Soret effect arises when the mass flux contains a term that
astrophysics, geophysics and in the motion of earths core. In
depends on the temperature gradient. The analogous effect that
addition from the technological point of view, MHD free
arises from a concentration gradient dependent term in the
convection flows have significant applications in the field of
heat flux is called the Dufour effect. Dursunkaya and Worek
stellar and planetary magnetospheres, aeronautical plasma
[11] studied diffusion-thermo and thermal-diffusion effects in
flows, chemical engineering and electronics. An excellent
transient and steady natural convection from a vertical surface.
summary of applications is to be found in Huges and Young
In view of the importance of above mentioned effects,
[5]. Raptis [6] studied mathematically the case of time varying
Kafoussias and Williams [12] studied the Soret and Dufour
Radiation Effects On Mhd Free Convection Flow Past A Semi-Infinite Moving Vertical Porous Plate With Soret And Dufour Effect
73
ISSN:0974-8652
effects on free convective and mass transfer boundary layer ∂u ∂v

flow with temperature dependent viscosity. Anghel et al. [13] + =0
investigated the Dufour and Soret effects on free concentration ∂x ∂y
boundary layer flow over a vertical surface embedded in a (2.1)
porous medium. Postelnicu [14] studied numerically the Momentum equation
influence of a magnetic field on heat and mass transfer by ∂u ∂u ∂2u σB2
natural convection from vertical surfaces in porous media u +v =ν 2 + gβ (T −T∞ ) + gβ* ( C −C∞ ) − 0 u
considering Soret and Dufour effects. Recently, Alam and ∂x ∂y ∂y ρ
Rahman [15] investigated the Dufour and Soret effects on (2.2)
mixed convection flow past a vertical porous flat plate with Energy equation
variable suction. Alam et al. [16] studied Dufour and Soret ∂T ∂T ∂ 2T 1 ∂ qr Dm kT ∂ 2C
effects on steady free convection and mass transfer flow past a u +v =α 2 − +
semi-infinite vertical plate in a porous medium. ∂x ∂y ∂y ρ c p ∂ y cs c p ∂y 2
In most of the studies mentioned above, viscous (2.3)
dissipation is neglected. Gebhart [17] has shown the
importance of viscous dissipative heat in free convection flow Species equation
in the case of isothermal and constant heat flux at the plate.
∂C ∂C ∂ 2C D k ∂ 2T
Gebhart and Mollendorf [18] considered the effects of viscous u +v = Dm 2 + m T
dissipation for external natural convection flow over a surface. ∂x ∂y ∂y Tm ∂y 2
Soundalgekar [19] analyzed viscous dissipative heat on the (2.4)
two-dimensional unsteady free convective flow past an infinite The boundary conditions for the velocity, temperature and
vertical porous plate when the temperature oscillates in time concentration fields are
and there is constant suction at the plate. Israel Cookey et al.
[20] investigated the influence of viscous dissipation and u = U 0 , v = v0 ( x ) , T = Tw , C = Cw at y = 0
radiation on unsteady MHD free convection flow past an u → 0, v → 0, T → T∞ , C → C∞ as y→∞
infinite heated vertical plate in a porous medium with time
dependent suction. where U 0 is the uniform velocity and v0 ( x ) is the velocity
of suction at the plate and u , v are the velocity components
2. MATHEMATICAL ANALYSIS in x, y directions respectively, ρ - the fluid density, g-
A steady two-dimensional hydromagnetic flow of a
viscous incompressible, electrically conducting and viscous the acceleration due to gravity, β and β * - the thermal and
dissipating fluid past a semi-infinite moving vertical porous concentration expansion coefficients respectively, K ′ - the
plate embedded in a porous medium is considered. The flow is permeability of the porous medium, T - the temperature of
assumed to be in the x - direction, which is taken along the the fluid in the boundary layer, ν - the kinematic viscosity,
semi-infinite plate and y - axis normal to it. Initially, it is σ- T∞ - the
the electrical conductivity of the fluid,
assumed that the plate and the fluid are at the same temperature of the fluid far away from the plate, α - the
temperature T and the concentration C . The surface is thermal diffusivity, C - the species concentration in the
maintained at a constant temperature Tw , which is higher than boundary layer, C∞ - the species concentration in the fluid
the constant temperature T∞ of the surrounding fluid and the B0 - the magnetic induction, k - the
far away from the plate,
concentration Cw is greater than the constant thermal conductivity, c p - the specific heat at constant
concentration C∞ . It is assumed that the interaction of the pressure, kT - the thermal diffusion ratio, cs - the
induced axial magnetic field with the flow is considered to be
concentration susceptibility, Tm - the mean fluid temperature,
negligible compared to the interaction of the applied field B0 ,
with the flow. It is also assumed that all the fluid properties Dm - the mass diffusivity.
are constant except that of the influence of the density Thermal radiation is assumed to be present in the
variation with temperature and concentration in the body force form of a unidirectional flux in the y-direction i.e.
term (Boussinesq’s approximation). Also, there is no chemical q r (transverse to the vertical surface). By using the Rosseland
reaction between the diffusing species and the fluid. Then,
under the boundary layer approximations, the governing approximation (Brewster [29]), the radiative heat flux q r is
equations are given by
Continuity equation
74
ISSN:0974-8652
4σ s ∂ T ′ 4 number, M - the magnetic field parameter, , Pr - the Prandtl

qr = − number, Du - the Dufour number, Sc - the Schmidt number ,
3k e ∂ y Sr - the Soret number.
(6) The mass conservation equation (2.1) is satisfied by the
Cauchy-Riemann Equations
where σ s is the Stefan-Boltzmann constant and k e - the mean ∂ψ ∂ψ
u= and v = − .
absorption coefficient. It should be noted that by using the ∂y ∂x
Rosseland approximation, the present analysis is limited to In view of the Equation (2.6) , and following the analysis of
optically thick fluids. If temperature differences within the Chamkha and Issa [21], the equations (2.2), (2.3) and (2.4)
flow are sufficiently small, then Equation (6) can be linearized reduce to the following non-dimensional form
by expanding T ′ into the Taylor series about T∞′ , which
4
f ′′′ + ff ′′ + Gr θ + Gm φ − Mf ′ = 0
after neglecting higher order terms takes the form (2.7)
3 4
θ ′′ + Pr f θ ′ + radiationterm + Pr Du φ ′′ = 0
T ′4 ≅ 4T∞′ T ′ − 3T∞′ (2.8)
(7) φ ′′ + Sc f φ ′ + Sc Sr θ ′′ = 0
(2.9)
In view of Equations (6) and (7), Equation (3) reduces to
The corresponding boundary conditions are
∂T′ ∂T′ ∂ 2 T ′ 16σ s T∞′3 ∂ 2T ′ Dm kT ∂ 2C f = f w , f ′ = 1, θ = 1, φ = 1 at η = 0

u +v =α + + f ′ → 0, θ → 0, φ → 0 as η → ∞
∂x ∂y ∂ y 2 3ke ρ c p ∂ y 2 cs c p ∂y 2
(8) 2x
where f w = −v0 is the dimensionless suction velocity
The Equations (2.2) to (2.4) are strongly coupled, ν U0
parabolic and nonlinear partial differential equations. An
analytical solution cannot be obtained and therefore we seek and primes denote partial differentiation with respect to the
numerical solutions. Numerical computations are greatly variable.
facilitated by non- dimensionalization of the equations. The skin-friction coefficient, Nusselt number and
Proceeding with the analysis, we introduce the following Sherwood number are important physical parameters for this
similarity transformations and dimensionless variables which type of boundary layer flow.
will convert the partial differential equations from two The skin-friction coefficient in non-dimensional form is
1
independent variables ( x, y ) to a system of coupled, non- C f = 2 ( Re )
−
2 f ′′ ( 0 )
linear ordinary differential equations in a single variable ( η ) The Nusselt number in non-dimensional form is
i.e. coordinate normal to the plate. 1
In order to write the governing equations and the boundary Nu = −(Re) 2 θ ′ ( 0 )
conditions in dimensionless form, the following non-
The Sherwood number in non-dimensional form is
dimensional quantities are introduced. 1
U T −T C−C∞ Sh = − ( Re ) 2 φ ′ ( 0 )
η = y 0 , ψ = νxU0 f (η) , θ(η) = ∞ , φ(η) = ,
2νx Tw −T∞ Cw −C∞ U0 x
where Re = is the Reynolds number
∂ψ ∂ψ gβ(Tw −T∞) 2x gβ ( Cw −C∞) 2x *
ν
u= , v=− , Gr = , Gm= ,
∂y ∂x U02 U02 3. NUMERICAL SOLUTION
(2.
2
νρcp The set of coupled non-linear governing boundary
2σB x0 ν
M= , R= , Sc= , Pr = , layer Equations (2.7) - (2.9) together with the boundary
ρU0 Dm k conditions (2.10) are solved numerically by using Runge-
Kutta fourth order technique along with Shooting method.
Dm kT ( Cw − C∞ ) D k (T − T )
6) Du = , Sr = m T w ∞ First of all, higher order non-linear differential Equations (2.7)
cs c p (Tw − T∞ ) ν Tm ( Cw − C∞ ) - (2.9) are converted into simultaneous linear differential
equations of first order and they are further transformed into
where ψ is the stream function, θ - the non-dimensional initial value problem by applying the Shooting technique (Jain
temperature function, φ - the non-dimensional concentration, et al. [22]). The resultant initial value problem is solved by
Gr - the thermal Grashof number, Gm - the solutal Grashof employing Runge-Kutta fourth order technique.
75
ISSN:0974-8652
the heated surface more rapidly than for higher values of Pr.
4. RESULTS AND DISCUSSION Hence in the case of smaller Prandtl numbers as the boundary
As a result of the numerical calculations, the layer is thicker and the rate of heat transfer is reduced.
dimensionless velocity, temperature and concentration The effect of the viscous dissipation parameter i.e.,
distributions for the flow under consideration are obtained and the Eckert number Ec on the velocity and temperature are
their behaviour have been discussed for variations in the shown in Figs. 6(a) and 6(b) respectively. The Eckert number
governing parameters viz., the thermal Grashof number Gr, Ec expresses the relationship between the kinetic energy in the
solutal Grashof number Gm, magnetic field parameter M, flow and the enthalpy. It embodies the conversion of kinetic
permeability parameter K, Prandtl number Pr, Eckert number energy into internal energy by work done against the viscous
Ec, Dufour number Du, Schmidt number Sc, Soret number Sr fluid stresses. The positive Eckert number implies cooling of
and the suction parameter f w . the surface i.e., loss of heat from the plate to the fluid. Hence,
greater viscous dissipative heat causes a rise in the
The influence of the thermal Grashof number Gr on temperature as well as the velocity, which is evident from
the velocity is presented in Fig.1. The thermal Grashof Figs. 6 (a) and 6 (b).
number Gr signifies the relative effect of the thermal For different values of the Dufour number Du, the
buoyancy force to the viscous hydrodynamic force in the velocity and temperature profiles are plotted in Figs. 7(a) and
boundary layer. As expected, it is observed that there is a rise 7(b) respectively. The Dufour number Du signifies the
in the velocity due to the enhancement of thermal buoyancy contribution of the concentration gradients to the thermal
force. Here the positive values of Gr correspond to cooling of energy flux in the flow. It is found that an increase in the
the surface. Also, as Gr increases, the peak values of the Dufour number causes a rise in the velocity and temperature
velocity increases rapidly near the wall of the porous plate and throught the boundary layer. For Du ≤ 1 , the temperature
then decays smoothly to the free stream velocity. profiles decay smoothly from the surface to the free stream
Fig.2 presents typical velocity profiles in the value.
boundary layer for various values of the solutal Grashof The influence of Schmidt number Sc on the velocity
number Gm, while all other parameters are kept at some fixed and concentration profiles are plotted in Figs. 8(a) and 8(b)
values. The solutal Grashof number Gm defines the ratio of respectively. The Schmidt number embodies the ratio of the
the species buoyancy force to the viscous hydrodynamic force. momentum to the mass diffusivity. The Schmidt number
The velocity distribution attains a distinctive maximum value therefore quantifies the relative effectiveness of momentum
in the vicinity of the surface and then decreases properly to and mass transport by diffusion in the hydrodynamic
approach the free stream value. As expected, the fluid velocity (velocity) and concentration (species) boundary layers. As the
increases and the peak value is more distinctive due to Schmidt number increases the concentration decreases. This
increase in the species buoyancy force. causes the concentration buoyancy effects to decrease yielding
For various values of the magnetic parameter M, the a reduction in the fluid velocity. The reductions in the velocity
velocity profiles are plotted in Fig.3. It can be seen that as M and concentration profiles are accompanied by simultaneous
increases, the velocity decreases. This result qualitatively reductions in the velocity and concentration boundary layers.
agrees with the expectations, since the magnetic field exerts a These behaviors are clear from Figs. 8(a) and 8(b).
retarding force on the free convection flow. Figs. 9(a) and 9(b) depict the velocity and
The effect of the permeability parameter K on the concentration profiles for different values of Soret number Sr.
velocity field is shown in Fig. 4. The parameter K as The Soret number Sr defines the effect of the temperature
defined in equation (2.6) is inversely proportional to the actual gradients inducing significant mass diffusion effects. It is
permeability K ′ of the porous medium. An increase in K obvious that an increase in the Soret number Sr results in an
will therefore increase the resistance of the porous medium (as increase in the velocity and concentration with in the boundary
the permeability physically becomes less with increasing K ′ ) layer.
which will tend to decelerate the flow and reduce the velocity. Figs.10(a), 10(b) and 10(c) illustrate the influence of
Figs.5(a) and 5(b) illustrate the velocity and suction parameter f w on the velocity, temperature and
temperature profiles for different values of Prandtl number Pr. concentration respectively. It is observed that an increase in
The Prandtl number defines the ratio of momentum diffusivity the suction parameter results in a decrease in the velocity,
to thermal diffusivity. The numerical results show that the temperature and concentration.
effect of increasing values of Prandtl number results in a The effects of various governing parameters on the
decreasing velocity. From Fig.5 (b), it is observed that an
skin friction coefficient C f , Nusselt number Nu and the
increase in the Prandtl number results a decrease of the
thermal boundary layer thickness and in general lower average Sherwood number Sh are shown in Tables 1 and 2. From
temperature with in the boundary layer. The reason is that Table 1, it is observed that as Gr or Gm increases, there is a
smaller values of Pr are equivalent to increasing the thermal rise in the local skin-friction coefficient, Nusselt number and
conductivities, and therefore heat is able to diffuse away from
76
ISSN:0974-8652
the Sherwood number. It is seen that, as M or K increases, 1.4

there is a fall in the skin-friction coefficient, the Nusselt
M = 0.0, 0.5, 1.0, 2.0
number and the Sherwood number. Also, it is noticed that as 1.2
the suction parameter f w increases, the local skin-friction 1

Gr = 2.0 Gm = 2.0 K = 0.5
coefficient decreases, while the Nusselt number and Sherwood 0.8 Pr = 0.71 Ec = 0.01 Du = 0.2
Sc = 0.6 Sr = 1.0 fw = 0.5
number increase. From Table 2, it is observed that an increase
f'
0.6
in Pr leads to a decrease in the skin-friction and Sherwood
number and an increase in the Nusselt number. It is also 0.4
noticed that an increase in Ec or Du leads to an increase in the
0.2
skin-friction and Sherwood number and a decrease in the
Nusselt number. It is observed that an increase in the Schmidt 0
0 0.5 1 1.5 2 2.5 3 3.5 4
number Sc reduces the skin-friction coefficient and Nusselt η
number and increases the Sherwood number. It is also seen Fig.3. Velocity profiles for different values of M
that an increase in Soret number Sr leads to an increase in the
1.4
skin-friction and Nusselt number and a decrease in the
Sherwood number. 1.2
K = 0.5, 1.0, 1.5, 2.0
1.4
1
Gr = 1.0, 2.0, 3.0, 4.0
Gr = 2.0 Gm = 2.0 M = 0.5
1.2
0.8 Pr = 0.71 Ec = 0.01 Du = 0.2
Sc = 0.6 Sr = 1.0 fw = 0.5
f'
1
Gm = 2.0 M = 0.5 K = 0.5 0.6
Pr = 0.71 Ec = 0.01 Du = 0.2
0.8
Sc = 0.6 Sr = 1.0 f w = 0.5
f'
0.4
0.6
0.2
0.4
0.2 0
0 0.5 1 1.5 2 2.5 3 3.5 4
η
0
0 0.5 1 1.5 2 2.5 3 3.5 4
η Fig.4. Velocity profiles for different values of K
Fig.1. Velocity profiles for different values of Gr
1.4
1.4
1.2
1.2 Gm = 1.0, 2.0, 3.0, 4.0 Pr = 0.71, 1.0, 1.25, 1.5
1
1 Gr = 2.0 Gm = 2.0 K = 0.5
Gr = 2.0 M = 0.5 K = 0.5 Sc = 0.6 Ec = 0.01 Du = 0.2
0.8
0.8 Pr = 0.71 Ec = 0.01 Du = 0.2 M = 0.5 Sr = 1.0 f w = 0.5
f'
Sc = 0.6 Sr = 1.0 fw = 0.5

f'
0.6
0.6
0.4
0.4
0.2
0.2
0
0 0.5 1 1.5 2 2.5 3 3.5 4
0
0 0.5 1 1.5 2 2.5 3 3.5 4 η
η
Fig.5(a). Velocity profiles for different values of Pr
Fig.2. Velocity profiles for different values of Gm
77
ISSN:0974-8652
1 Fig.7(a). Velocity profiles for different values of Du

1
0.8
0.8
0.6 Pr = 0.71, 1.0, 1.25, 1.5
Du = 0.0, 0.2, 0.6, 1.0
θ
Gr = 2.0 Gm = 2.0 K = 0.5 0.6

0.4 Sc = 0.6 Ec = 0.01 Du = 0.2
M = 0.5 Sr = 1.0 fw = 0.5
θ
0.4
0.2 Gr = 2.0 Gm = 2.0 K = 0.5
Pr = 0.71 Ec = 0.01 M = 0.5
Sc = 0.6 Sr = 1.0 f w = 0.5
0 0.2
0 0.5 1 1.5 2 2.5 3 3.5 4
η
Fig.5(b). Temperature profiles for different values of Pr 0

0 0.5 1 1.5 2 2.5 3 3.5 4
η
1.4
Fig.7(b). Temperature profiles for different values of Du
Ec = 0.0, 0.01, 0.02, 0.03
1.2
1.4
1 Sc = 0.3, 0.6, 0.78, 0.94
1.2
0.8 Gr = 2.0 Gm = 2.0 K = 0.5
Pr = 0.71 Sc=0.6 Du = 0.2 1
f'
M = 0.5 Sr = 1.0 fw = 0.5 Gr = 2.0 Gm = 2.0 K = 0.5

0.6
0.8 Pr = 0.71 Ec = 0.01 Du = 0.2
M = 0.5 Sr = 1.0 fw = 0.5
0.4 f'
0.6
0.2
0.4
0
0 0.5 1 1.5 2 2.5 3 3.5 4 0.2
η
Fig.6(a). Velocity profiles for different values of Ec 0

0 0.5 1 1.5 2 2.5 3 3.5 4
η
1
Gr = 2.0 Gm = 2.0 K = 0.5
Fig.8(a). Velocity profiles for different values of Sc
Pr = 0.71 Sc=0.6 Du = 0.2
0.8 M = 0.5 Sr = 1.0 fw = 0.5 1
0.6 0.8 Sc = 0.3, 0.6, 0.78, 0.94

θ
Ec = 0.0, 0.01, 0.02, 0.03

0.4 0.6 Gr = 2.0 Gm = 2.0 K = 0.5
Pr = 0.71 Ec = 0.01 Du = 0.2
φ
M = 0.5 Sr = 1.0 f w = 0.5

0.2 0.4
0 0.2
0 0.5 1 1.5 2 2.5 3 3.5 4
η
Fig.6(b). Temperature profiles for different values of Ec 0

0 0.5 1 1.5 2 2.5 3 3.5 4
η
1.4
Fig.8(b). Concentration profiles for different values of Sc
1.2
1.4
Du = 0.0, 0.2, 0.6, 1.0
1 1.2 Sr = 1.0, 1.5, 2.0, 2.5
0.8 Gr = 2.0 Gm = 2.0 K = 0.5 1

Pr = 0.71 Ec = 0.01 M = 0.5 Gr = 2.0 Gm = 2.0 K = 0.5
f'
Sc = 0.6 Sr = 1.0 fw = 0.5 0.8 Pr = 0.71 Ec = 0.01 Du = 0.2

0.6 Sc = 0.6 M = 0.5 f w = 0.5
f'
0.6
0.4
0.4
0.2
0.2
0
0 0.5 1 1.5 2 2.5 3 3.5 4 0
η 0 0.5 1 1.5 2 2.5 3 3.5 4
η
78
ISSN:0974-8652
Fig.9(a). Velocity profiles for different values of Sr 1
1
Sr = 0.0, 1.0, 1.5, 2.0 0.8
fw = 0.5, 1.0, 1.5, 2.0
0.8
0.6
Gr = 2.0 Gm = 2.0 K = 0.5
Pr = 0.71 Ec = 0.01 Du = 0.2 Gr = 2.0 Gm = 2.0 K = 0.5
φ
0.6
Sc = 0.6 M = 0.5 fw = 0.5 Pr = 0.71 Ec = 0.01 Du = 0.2
0.4 M = 0.5 Sr = 1.0 Sc= 0.6
φ
0.4
0.2
0.2
0
0 0.5 1 1.5 2 2.5 3 3.5 4
0 η
0 0.5 1 1.5 2 2.5 3 3.5 4
η Fig.10(c). Concentration profiles for different values of fw
Fig.9(b). Concentration profiles for different values of Sr
1.4 Table 1 Numerical values of the skin-friction coefficient C f ,
1.2
fw = 0.5, 1.0, 1.5, 2.0 Nusselt number Nu and
1
Sherwood number Sh for
0.8
Gr = 2.0 Gm = 2.0 K = 0.5 Pr = 0.71 , Ec = 0.01 , Du = 0.2 , Sc = 0.6 , Sr = 1.0
f'
Pr = 0.71 Ec = 0.01 Du = 0.2

0.6 M = 0.5 Sr = 1.0 Sc= 0.6
0.4
0.2
0
0 0.5 1 1.5 2 2.5 3 3.5 4
η
Gr Gm M K fw Cf Nu Sh
Fig.10(a). Velocity profiles for different values of fw
1
2.0 2.0 0.5 0.5 0.5 0.8230 0.8618 0.4362
4.0 2.0 0.5 0.5 0.5 2 6 2
0.8
2.0 4.0 0.5 0.5 0.5 1.6865 0.9019 0.4647
fw = 0.5, 1.0, 1.5, 2.0
2.0 2.0 1.0 0.5 0.5 0 3 9
2.0 2.0 0.5 1.0 0.5 1.8853 0.9188 0.4794
0.6
2.0 2.0 0.5 0.5 1.0 3 3 3
θ
0.4906 0.8400 0.4213

0.4
Gr = 2.0 Gm = 2.0 K = 0.5 8 5 6
Pr = 0.71 Ec = 0.01 Du = 0.2
M = 0.5 Sr = 1.0 Sc= 0.6
0.4878 0.8395 0.4198
0.2 1 6 4
0.5115 1.0936 0.4730
0
0 0.5 1 1.5 2 2.5 3 3.5 4
4 8 1
η
Fig.10(b). Temperature profiles for different values of fw
79
ISSN:0974-8652
Table 2 Numerical values of the skin-friction coefficient C f , [8]. Elabashbeshy E.M.A. (1997), Heat and mass transfer
along a vertical plate with variable temoerature and
Nusselt number Nu and concentration in the presence of magnetic field, Int. J.
Sherwood number Sh for Eng. Sci., Vol.34, pp. 515-522.
[9]. Chamkha A.J. and Khaled A.R.A. (2001), Similarity
Gr = 2.0 , Gm = 2.0 , M = 0.5 , K = 0.5 , f w = 0.5
solutions for hydrodynami simultaneous heat and
mass transfer by natural convection from an inclined
Pr Ec Du Sc Sr Cf Nu Sh plate with internal heat generation or absorption,
Heat Mass Transfer, Vol.37, pp.117-123.
0.7 0.0 0.2 0.6 1.0 0.8230 0.8618 0.4362 [10]. Eckert E.R.G. and Drake R. M. (1972), Analysis of
1 1 0.2 0.6 1.0 2 6 2 Heat and Mass Transfer,
1.0 0.0 0.2 0.6 1.0 0.7537 1.1087 0.2908 [11]. McGraw-Hill Book Co., New York. pp.217-230.
0.7 1 0.4 0.6 1.0 1 2 4 Dursunkaya Z. and Worek W. M. (1992), Diffusion-
1 0.0 0.2 0.7 1.0 0.8240 0.8590 0.4378 thermo and thermal diffusion effects in transient and
0.7 2 0.2 8 2.0 6 6 8 steady natural convection from a vertical surface, Int.
1 0.0 0.6 0.8405 0.8292 0.4573 J. Heat Mass Transfer, Vol.35, pp.2060-2065.
0.7 1 3 4 4 [12]. Kafoussias N.G. and Williams E.M. (1995),
1 0.0 0.7731 0.8469 0.4994 Thermal-diffusion and Diffusion-thermo effects on
0.7 1 5 4 9 free convective and mass transfer boundary layer
1 0.0 1.0084 0.9330 0.2931 flow with temperature dependent viscosity, Int. J.
1 4 8 3 Engg. Science, Vol. 33, pp.1369-1376.
[13]. Anghel M., Takhar H.S. and Pop I. (2000), Dufour
and Soret effects on free convection boundary layer
ACKNOWLEDGEMENT over a vertical surface embedded in a porous
[14]. medium, J. heat and Mass Transfer. Vol.43,
We would like to express our thanks to referees for valuable pp.1265-1274.
comments that improved the paper. [15]. Postelnicu A. (2004), Influence of a magnetic field
on heat and mass transfer by natural convection
REFERENCES from vertical surfaces in porous media considering
Soret and Dufour effects, Int. J. Heat Mass Transfer,
[1]. Bejan A. and Khair K.R. (1985), Heat and mass Vol.47, pp.1467-1472.
transfer by natural convection in a porous medium, [16]. Alam M.S. and Rahman M.M. (2006), Dufour and
Int. J. Heat Mass Transfer, Vol. 28, pp.909-918. Soret effects on mixed convection flow past a
[2]. Lai F.C. and Kulacki F.A. (1990), Coupled heat and vertical porous flat plate with variable suction,
mass transfer from a sphere buried in an infinite Nonlinear Analysis: Modelling and Control, Vol.11,
porous medium, Int. J. Heat Mass Transfer, Vol. 33, pp.435-442.
pp.209-215. [17]. Alam M.S., Ferdows M. and Maleque M.A. (2006),
[3]. Raptis A., Tzivanidis G. and Kafousias N. (1981), Dufour and Soret effects on steady free convection
Free convection and mass transfer flow through a and mass transfer flow past a semi-infinite vertical
porous medium bounded by an infinite vertical plate in a porous medium, Int. J. of Applied
limiting surface with constant suction, Lett. Heat Mechanics& Eng., Vol. 11, No.3, pp.535-545.
Mass Transfer, Vol. 8, pp. 417-424. [18]. Gebharat B.(1962), Effects of viscous dissipation in
[4]. Lai F.C. and Kulacki F.A. (1991), Coupled heat and natural convection , J. Fluid Mech., Vol. 14, pp.225-
mass transfer by natural convection from vertical 232.
surfaces in a porous medium, Int. J. Heat Mass [19]. Gebharat B. and Mollendorf J.(1969), Viscous
Transfer, Vol. 34, pp.1189-1194. dissipation in external natural
[5]. Huges W.F. and Young F.J. (1966), The Electro- [20]. convection flows, J. Fluid. Mech.,Vol.38, pp.97-
Magneto Dynamics of fluids, John Wiley and Sons, 107.
NewYork. [21]. Soundalgekar V.M.(1972), Viscous dissipation
[6]. Raptis A. (1986), Flow through a porous medium in effects on unsteady free convective flow past an
the presence of magnetic field, Int. J. Energy Res., infinite, vertical porous plate with constant suction,
Vol.10, pp. 97-101. Int. J. Heat Mass Transfer, Vol. 15, pp.1253-1261.
[7]. Helmy K.A. (1998), MHD unsteady free convection
flow past a vertical porous plate, ZAMM, Vol. 78,
pp. 255-270.
80
ISSN:0974-8652
[22]. Israel-Cookey C., Ogulu A. and Omubo-Pepple

V.B.(2003), Influence of viscous dissipation on
unsteady MHD free-convection floe past an infinite
heated vertical plate in porous medium with time-
dependent suction, Int. J. Heat Mass transfer, Vol.46,
pp.2305-2311.
[23]. Chamkha A.J. and Camille I. (2000), Effects of heat
generation/absorption and thermophoresis on
hydromagnetic flow with heat and mass transfer over
a flat surface, Int. J. Numerical Methods in Heat and
Fluid Flow, Vol.10, pp.432-448.
[24]. Jain M.K., Iyengar S.R. K. and Jain R.K. (1985),
Numerical Methods for Scientific and Engineering
Computation, Wiley Eastern Ltd., New Delhi, India.
81
ISSN:0974-8652

MATHEMATICAL IDEAS
PARETO DISTRIBUTION - SOME METHODS OF ESTIMATION
R. Subba Rao@, R.R.L. Kantam#, G.Srinivasa Rao$

@
Shri Vishnu Engineering College for Women, Bhimavaram-534 202, Andhra Pradesh, INDIA, E-mail:
rsr_vishnu@rediffmail.com
#
Department of Statistics, Acharya Nagarjuna University, Nagarjuna Nagar – 522 510. Andhra Pradesh,
INDIA, E-mail: kantam-rrl@rediffmail.com
$
Assistant professor, College of Agriculture, KEREN, ERITREA, E-mail: gadde-srao@Yahoo.co.in
ABSTRACT
Pareto distribution of type IV is considered with a known location and shape parameters.
Estimation of its scale parameter by the well known maximum likelihood method is modified by two
different approaches in order to yield linear estimators. Estimation based on a single optimum quantile is
also presented. The proposed methods are compared with respect to simulated sampling characteristics.
Key words: Order statistics, M.L Estimation, Quantiles, Asymptotic Variance.

1. Introduction given random sample in order to satisfy the
The Probability distribution function (P.d.f.) requirement that X ≥ µ. Any other estimator of µ
different from the first order statistic may not be that
of Pareto (IV) distribution is given by efficient, because it contains the maximum
information about µ. Here without loss of generality
α   x − µ −( α + 1)  we assume that µ is zero. Accordingly the density
f (x ;σ , α ) = 1 +   , x f µ, α f 1, σconsidered
f0 for estimation, is
σ   σ  
α  x  − ( α + 1)

(1.1) f ( x ;σ , α ) = 1 +    , X >0
σ  σ  
We start with general M.L. estimation of
‘α ‘and ‘σ’ taking µ as zero. As the estimating (2.1)
equations are to be solved by numerical iterative Let X1 < X 2 < X3 < X4 < -------- < Xn be an ordered
techniques we suggest some modifications to M.L. sample of size ‘n’ from a Pareto distribution (2.1).
method from complete as well as censored samples. The log likelihood equations to estimate α, σ from the
Discussion of complete sample situation is given in given complete sample are given by
section 2, whereas section 3 deals with the situation
of censored samples. Quantile estimation based on
optimally selected sample quantiles is presented in ∂ logL n  x   x   x 
section 4. Whenever the results are based on
= − log1+ 1  . 1+ 2 . ......1+ n  
∂α α  σ   σ   σ 
numerical computations, all such results are
presented in the form of numerical tables towards the
end with appropriate identification labels. (2.2)
∂ log L n zi
= (1 + α ) ∑ −n
2. Estimation from Complete Sample ∂σ i =1 1 + z
i
The parameter µ in the p d f given by (2.3)

equation (1.1) is the threshold parameter and is
generally estimated by the first order statistic in a
Pareto Distribution - Some Methods Of Estimation
82
ISSN:0974-8652
xi Method I
where zi =
σ i
Let p = , i = 1, 2, 3, − − − − − n,
For M.L.Es of α and σ , i n +1
∂ logL ∂ logL Let Zi, Zi’ be the solutions of the following
= 0, =0 equations
∂α ∂σ
' ' '' ''
After simplification these equations become F(zi ) = pi and F(zi ) = pi
∧
n where
α =
n  x  pi qi pi qi
∑ log  1 + i  '
pi = pi − , pi
' '
= pi +
i =1
 σ  n n
(2.4) The solutions of zi ' and zi ' ' in our Pareto

n z distribution are
(1 + α ) i∑= 1 i − n = 0 1
1 + zi
(2.5)
'
zi = 1 − pi ( '
)− α − 1
xi 1
where z i =
σ
and
''
zi = 1 − pi ( ''
)− α − 1
It can be seen that equation, (2.5) can be The intercept γ i and slope δ i of the linear
solved only by iterative method for σ. The MLE approximation in the equation (2.7) are respectively
of α is an analytical expression involving σ. In order given by
to overcome the iterative techniques that may some
times lead to convergence problems we approximate δi =
''
h ( zi ) − h zi ( ) '
'' '
the expression zi − zi
zi (2.9)
h(z i ) =
1 + zi and γ i = h zi ( )
− δi zi
(2.6) (2.10)
of the log likelihood equation (2.5) for estimating σ The values of γi and δi in this method for n = 5, 10,
by a linear expression say 15 and 20 and for α = 2, 3and 4 are given in table (1)
h( z i ) ≅ γ i + δ i z i
(2.7) Method II
in certain admissible ranges of Zi. Such
approximations are not feasible for the log likelihood Consider the Taylor’s expansion of
equation of α. Hence we develop our approximate zi
h ( zi ) = in the neighbourhood of ith
ML method for estimation of σ with a known α. As 1 + zi
per the parametric specifications we take α = 2, 3
quantile of our standard Pareto population. We get
and 4. After using the linear approximation given by
another linear approximation for h(z), with δi = h '(zi
equation (2.7) in the equation (2.5) and solving it for
),
σ we get
1
−
(1 + α )∑δ
n
xi
z = (1 − p
i i
α ) , −1
∧ i
σ = i =1
n i
n − (α + 1 ) ∑ γ i pi =
i = 1
n +1
(2.8)
as an approximate MLE of σ , which is a linear γ i = h (z i ) − δ i z i
estimator. We suggest two methods of finding Substituting these approximations in the equation
γi, δi of equation (2.7 ). Similar methods are (2.8) we get another linear estimator of σ with
given in Srinivasa Rao and Kantam (2002) different values of γ i, δ i. The values of γi and δ i in
and Kantam and Sri Ram (2003) this method for n = 5, 10, 15, and 20 and α = 2, 3 and
4 are given in Table ( 2)
83
ISSN:0974-8652
These two methods are asymptotically as efficient as 3 Estimation From Right Censored Samples
exact maximum likelihood estimators as can be seen
As described in section 2, we develop
from the following narration (Tiku et al , 1986;
estimation of the parameter σ from a censored
Balakrishnan, 1990).
sample with a known α.
In the suggested two methods of
Let X1 < X2 < X3 < X4 < -------- < Xn be
approximation, the function h(z) is linearised in the
an ordered sample of size ‘n’ from a Pareto
neighbourhood of the population quantile in two
different senses. Because sample quantiles are distribution with unknown scale parameter σ and a
consistent estimators of the corresponding population known shape parameter α. Let the largest ‘r’
quantiles and sample moments are consistent observations be deleted so that X1 < X2 < X3 <
estimators of the population moments, for large X4 < ------- < Xn –r is Type II right censored sample
values of ‘n’, the neighbourhood of the population (also called failure censored sample) from a Pareto
quantile becomes narrower, thereby giving more distribution (2.1). The log likelihood function to
linearity of h(z) in that neighbourhood. That is the estimate σ from the given censored sample is given
closeness of h (z) to γ + δ z is stronger, the larger from
the sample size. Hence the approximate and exact
∂ log L
expressions of the log likelihood equation
∂σ  n −r
 α   xi 
−(α + 1)
 −α r
differ by little values. Also exact and approximate L ∝  ∏   1+    ( 1+ xn−r ) 
i =1  σ   σ  
2
∂ log L  
values of 2
would differ little. Hence the
∂σ n−r   x 
α
exact MLE, Modified MLEs by the two methods for tant + ∑  log  −(α +1) log 1+ i  
logL=Cons
the parameter σ shall have the same asymptotic bias i =1
 σ  σ 
and asymptotic variance (Bhattacharya, 1985).
 xn −r 
However the same can not be said in small samples.
−α. r log1+ 
Since the exact MLE of σ is an iterative solution of  σ
equation (2.5), its sampling variance can not be
mathematically tractable. Hence, we have resorted where the constant is independent of the parameters
to Monte Carlo simulation to get the empirical to be estimated. The log likelihood equation for
sampling characteristics of the exact M L E. We estimating σ is given by
have computed the simulated bias, variance and M S
E of exact M L E solving equation (2.5) iteratively
∂logL
for σ in 10,000 samples of size 5, 10, 15 and 20 =0
each generated from standard Pareto distribution ∂σ
with α = 2, 3 and 4. These are given in table (3).
⇒
n −r (α +1) n− r x
− 2 ∑ i − 2.
αr x n − r
= 0
The simulated bias, MSE, and variance of MML Es
of σ are also given in table (3).
σ σ i =1 1+xi σ xn − r
1+
σ σ
A comparison of the sampling
characteristics namely the bias, variance, and MSE
⇒
n−r (α +1) n− r zi αr zn − r
∧ ∧
of σ 1 , σ 2 - the two MMLEs together with those of
− ∑ − . =0
σ σ i =1 1+ zi σ 1+ zn − r
the corresponding exact MLE, reveal that MMLE of
Method I is preferable to that of Method II as well as n− r zi zn − r
exact MLE in small samples as Method I recorded ⇒(n − r) − (α + 1) ∑ − αr . =0
minimum values for bias variance and MSE. Coming i =1 1 + zi 1 + zn − r
to the actual magnitudes of these sample (3.1)
characteristics it is MMLE of method I that is closer It can be seen that equation (3.1) cannot be solved
to exact M L method rather than MMLE of Method analytically for σ .The M L E of σ has to be
II. obtained as an iterative solution of (3.1). We
zi
approximate the expression of the
1 + zi
84
ISSN:0974-8652
likelihood equation (3.1) for estimating σ by a linear which certain function is linearised depends on the
expression as size of the sample also. The larger the size, the
closer the approximation. That is, the exactness of
h (z i ) ≅ γi + δ i zi the approximation becomes finer and finer for large
(3.2) values of n. Hence, the approximate log likelihood
equation and the exact log likelihood equation differ
where γi, δi are to be suitably found, to get a by little quantities for large n. Therefore, the
modified MLE of σ. Proceeding on the same lines as solutions of exact and approximate log likelihood
mentioned in section 2.2 approximate likelihood equations tend to each other as n → ∞. Hence the
equation for σ is given by exact and modified M L Es are asymptotically
identical (Tiku et al 1986). However, the same
∂ log L ∂ log L cannot be said in small samples. At the same time
≅ = 0 the small sample variance of exact M L E is not
∂σ ∂σ mathematically tractable. We therefore compared
these estimates in small samples through Monte
Carlo simulation.
n− r
⇒(n−r) − (α+1) ∑ (γi +δi zi ) − αr (γn −r + δn −r zn −r ) = 0 The bias, the variance, the MSE of the
i =1
estimates by the two methods of modification and
that of the exact M L E obtained through simulation
for n = 5, 10, 15 and 20 and α = 2, 3 and 4 with all
n− r n −r xi possible considerations of right censored samples are
⇒ (n − r) − (α + 1) ∑ γ i − ∑ δi − α r γn − r
i =1 i =1 σ given in table 4for α = 2only.
xn − r
− α r δn − r =0 Conclusions:
σ
In most of the situations it is the MMLE of
Method – I that is rated as the most preferable
method; the second preference going to exact MLE.
⇒ σ (n−r) − (α+1) ∑ γi − α r γn − r 
n− r
The same trend is observed for other values of α also.
 i =1  Thus whether complete or censored sample, one can
n− r
go for MLE with iterative solution or MMLE – I
− (α+1) ∑ δi xi − α r δn − r xn − r = 0 with linear analytical estimator. In large samples, as
i =1 mentioned earlier all the three methods are equally
efficient.
n− r
∧
(α + 1) ∑ δi xi + α r δn − r xn − r
i =1
4 Estimation Based On Sample Quantiles
⇒σ = n− r
The concept of failure censored sample and
(n − r) − (α + 1) ∑ γi − α r γ n − r estimation therefrom as described in section 3 can be
i =1 modified slightly, with the notion of estimating
(3.3) unknown scale parameter σ based on selected order
and the resulting MMLEs of σ from the censored statistics in an optimum way. That is if ‘n’ is the
sample by the two methods are similar to those given given sample size and k is a positive integer less than
in section 2. The relevant values of slope and ‘n’ best linear unbiased estimation based on a subset
intercept are calculated for α = 2, 3, 4; n = 5, 10, 15, of k order statistics in the sample can be thought of
20, and for all possible combinations of r, which run using the theory of Lloyd (1952) if we have the
into 10 pages. Owing to the problem of space we are moments of order statistics in a sample of size ‘n’.
not including those tables here. n
 
Accordingly we can get  c  BLUES for σ each with
In the two methods referred above, the basic k 
zi  
principle is that the expression is its own variance given by the formulae of Lloyd
1 + zi
(1952). Among them, the BLUE with smallest
approximated by a linear function in some
variance is called estimator based on k – optimally
neighbourhood of the population quantile. It can be
selected order statistics. A revision of this procedure
seen that the construction of the neighbourhood over
85
ISSN:0974-8652
may be as fallows. Suppose a population contains a xp
single unknown parameter that requires estimation σˆ = −1
from a given sample of size ‘n’. We therefore select [1 − P] −1 α
one order statistic from the sample optimally, th

where x p is the p quantile in the sample; which can
corresponding to a population quantile. The be obtained as an ordered statistic in the sample
optimality here is specified as the minimum whose suffix is [n.p ] + 1.
asymptotic variance of the resulting estimator. If The above choice of p has to be made in
such an estimator is obtained, it is called quantile
estimator or optimum estimation based on a single such a way that the variance of the σˆ is the
quantile. In the case of multi parameter populations minimum with respect to p. But the exact variance of
the optimally selected sample quantiles will be as σˆ is not analytically available. However the
many as the number of independent unknown asymptotic variance of σˆ can be obtained as
parameters that are to be estimated. The optimality fallows.
criterion would be minimization of asymptotic
A S VA R ( x p )
generalized variance, taken as the trace of the Asymptotic variance of σˆ = 2
asymptotic dispersion matrix or the determinant of  −1

[1 − p ]
α
the asymptotic dispersion matrix. − 1
 
In our present investigation, we consider only a From the asymptotic theory of order statistics, we
single unknown parameter σ with other parameters know that the asymptotic variance of the Pth quantile
assumed to be known. Let P be a real number in the sample is
between 0 and 1. Let ξ be the Pth quantile of standard p (1− p )
Pareto population with known value of α. (i.e)ξ ∴ A S V A R (x p ) = − 2 (α + 1)
satisfies the equation F(ξ) = P  −1

α  [1 − p] α 
2
− α
 
 x (4.1)
( ie ) F ( x) = 1 −  1 + 
 σ  For an optimum choice of a sample quantile we have
to minimize the asymptotic variance given by (4.1)
with respect to P
=1 − [1 + x ] − α
(if σ = 1 ) d ∧
[ A S VA R (σ )] = 0
(i. e.) dp
− 1
⇒ ξ = [1 − p ] α
− 1
  −1
 
2
 P ( 1 − P )  [1 − P ] α − 1 
d    
⇒  − 2 (α + 1)  = 0
x x dp   −1
 
If is a standard Pareto variate, considering
 α  [1 − P ] 
2 α
σ σ 
as ξ, we get    
2
   
−1 1 1
x
ξ = = [1 − p ] α − 1 α
1 − (1 − p) 
α
2 − ( 1 − p) 
σ   2p  
 −1

⇒ 4
+ 4
= 0
⇒ ξ = σ  [1 − p ] α − 1 α
+ 2 α α
+ 2
( 1 − p) (1 − p )
 
d2
[ASVAR(xp )]=0
dp2
for a given sample of size n, if σ is to be estimated,
on the basis of a single sample quantile the above 1
 1 1

equation can be used as ⇒ (1− p) α  −6 pα − 6 p + 2α (1− p) α − 2α2 + 2α2 (1− p) α  =
 
ξ
σ = −1
[1 − p ] −1 α 1 1
2α2 (1− p) α + 4α (1− p) α − 12 pα −16p −2α2
where ξ is the population pth quantile. The above
equation suggests a possible estimator for σ as
86
ISSN:0974-8652
This equation has to be solved iteratively for P to get equation at α = 2, 3 and 4 gives that P = 0.000151,
its optimum value corresponding to a minimum of 0.000251, 0.000332. This shows that the sample size
the asymptotic variance of σˆ should be above thousand to get an asymptotic
We have found that Newton optimum sample quantile what ever may be the
Raphson method when applied to solve the above specified α.
87
ISSN:0974-8652
TABLE 1
Intercept and slope of the approximation h(Zi) = γi+ δi Zi Method-I

α =2 α =3 α =4
n I
γ δ γ δ γ δ
5 1 0.00000 0.81650 0.00000 0.87358 0.00000 0.90360
5 2 0.02055 0.63246 0.00982 0.73681 0.00573 0.79527
5 3 0.07083 0.44721 0.03564 0.58480 0.02136 0.66874
5 4 0.17051 0.25820 0.09241 0.40548 0.05763 0.50813
5 5 0.64998 0.00000 0.50334 0.00000 0.40837 0.00000
10 1 0.00000 0.90453 0.00000 0.93530 0.00000 0.95107
10 2 0.00503 0.80904 0.00232 0.86825 0.00133 0.89947
10 3 0.01599 0.71351 0.00751 0.79848 0.00434 0.84469
10 4 0.03407 0.61791 0.01635 0.72547 0.00956 0.78608
10 5 0.06094 0.52223 0.03000 0.64850 0.01778 0.72266
10 6 0.09904 0.42640 0.05024 0.56652 0.03023 0.65299
10 7 0.15222 0.33029 0.08005 0.47782 0.04910 0.57471
10 8 0.22729 0.23355 0.12518 0.37924 0.07871 0.48327
10 9 0.33903 0.13484 0.19955 0.26295 0.13018 0.36721
10 10 0.75536 0.00000 0.60884 0.00000 0.50538 0.00000
15 1 0.00000 0.93541 0.00000 0.95647 0.00000 0.96717
15 2 0.00223 0.87082 0.00101 0.91191 0.00058 0.93318
15 3 0.00693 0.80623 0.00319 0.86624 0.00183 0.89790
15 4 0.01441 0.74162 0.00672 0.81932 0.00388 0.86117
15 5 0.02503 0.67700 0.01185 0.77101 0.00688 0.82280
15 6 0.03924 0.61237 0.01886 0.72112 0.01104 0.78254
15 7 0.05761 0.54772 0.02817 0.66943 0.01663 0.74008
15 8 0.08087 0.48305 0.04030 0.61564 0.02403 0.69502
15 9 0.11003 0.41833 0.05599 0.55934 0.03375 0.64678
15 10 0.14645 0.35355 0.07634 0.50000 0.04660 0.59460
15 11 0.19216 0.28868 0.10301 0.43679 0.06382 0.53728
15 12 0.25036 0.22361 0.13883 0.36840 0.08759 0.47287
15 13 0.32671 0.15811 0.18917 0.29240 0.12222 0.39764
15 14 0.43358 0.09129 0.26710 0.20274 0.17876 0.30214
15 15 0.80098 0.00000 0.65912 0.00000 0.55388 0.00000
20 1 0.00000 0.95119 0.00000 0.96719 0.00000 0.97529
20 2 0.00125 0.90238 0.00057 0.93381 0.00032 0.94994
20 3 0.00386 0.85356 0.00176 0.89982 0.00100 0.92389
20 4 0.00793 0.80475 0.00365 0.86518 0.00209 0.89708
20 5 0.01361 0.75593 0.00633 0.82983 0.00364 0.86944
20 6 0.02105 0.70711 0.00990 0.79370 0.00573 0.84090
20 7 0.03043 0.65828 0.01447 0.75673 0.00842 0.81134
20 8 0.04197 0.60945 0.02019 0.71883 0.01182 0.78067
20 9 0.05593 0.56061 0.02725 0.67989 0.01606 0.74874
20 10 0.07262 0.51177 0.03588 0.63981 0.02130 0.71538
20 11 0.09244 0.46291 0.04636 0.59841 0.02773 0.68037
20 12 0.11588 0.41404 0.05907 0.55551 0.03564 0.64346
20 13 0.14359 0.36515 0.07453 0.51087 0.04539 0.60428
20 14 0.17644 0.31623 0.09345 0.46416 0.05752 0.56234
20 15 0.21562 0.26726 0.11685 0.41491 0.07282 0.51697
20 16 0.26290 0.21822 0.14634 0.36246 0.09251 0.46714
20 17 0.32108 0.16903 0.18458 0.30571 0.11877 0.41113
20 18 0.39512 0.11952 0.23669 0.24264 0.15583 0.34572
20 19 0.49587 0.06901 0.31503 0.16824 0.21459 0.26269
20 20 0.82795 0.00000 0.69066 0.00000 0.58522 0.00000
88
ISSN:0974-8652
TABLE 2
Intercept and slope of the approximation F(t i) = γ i + δ I Method-I I
α =2 α =3 α =4
n I γ δ γ δ γ δ
5 1 0.00759 0.83333 0.00348 0.88555 0.00199 0.91287
5 2 0.03367 0.66667 0.01598 0.76314 0.00929 0.81650
5 3 0.08579 0.50000 0.04256 0.62996 0.02531 0.70711
5 4 0.17863 0.33333 0.09403 0.48075 0.05768 0.57735
5 5 0.35017 0.16667 0.20221 0.30285 0.13036 0.40825
10 1 0.00217 0.90909 0.00098 0.93844 0.00055 0.95346
10 2 0.00911 0.81818 0.00419 0.87478 0.00239 0.90453
10 3 0.02167 0.72727 0.01014 0.80872 0.00586 0.85280
10 4 0.04092 0.63636 0.01956 0.73984 0.01142 0.79772
10 5 0.06836 0.54545 0.03347 0.66758 0.01977 0.73855
10 6 0.10615 0.45455 0.05342 0.59118 0.03201 0.67420
10 7 0.15759 0.36364 0.08193 0.50946 0.04993 0.60302
10 8 0.22826 0.27273 0.12355 0.42055 0.07692 0.52223
10 9 0.32902 0.18182 0.18791 0.32094 0.12041 0.42640
10 10 0.48789 0.09091 0.30289 0.20218 0.20331 0.30151
15 1 0.00101 0.93750 0.00045 0.95789 0.00026 0.96825
15 2 0.00417 0.87500 0.00190 0.91483 0.00108 0.93541
15 3 0.00972 0.81250 0.00447 0.87073 0.00256 0.90139
15 4 0.01795 0.75000 0.00836 0.82548 0.00482 0.86603
15 5 0.02919 0.68750 0.01379 0.77896 0.00800 0.82916
15 6 0.04386 0.62500 0.02103 0.73100 0.01229 0.79057
15 7 0.06250 0.56250 0.03046 0.68142 0.01795 0.75000
15 8 0.08579 0.50000 0.04256 0.62996 0.02531 0.70711
15 9 0.11462 0.43750 0.05801 0.57630 0.03486 0.66144
15 10 0.15026 0.37500 0.07777 0.52002 0.04729 0.61237
15 11 0.19447 0.31250 0.10330 0.46050 0.06367 0.55902
15 12 0.25000 0.25000 0.13693 0.39685 0.08579 0.50000
15 13 0.32147 0.18750 0.18288 0.32759 0.11694 0.43301
15 14 0.41789 0.12500 0.25000 0.25000 0.16435 0.35355
15 15 0.56250 0.06250 0.36379 0.15749 0.25000 0.25000
20 1 0.00058 0.95238 0.00026 0.96800 0.00015 0.97590
20 2 0.00238 0.90476 0.00108 0.93545 0.00061 0.95119
20 3 0.00550 0.85714 0.00251 0.90234 0.00143 0.92582
20 4 0.01005 0.80952 0.00463 0.86860 0.00265 0.89974
20 5 0.01616 0.76190 0.00751 0.83419 0.00432 0.87287
20 6 0.02398 0.71429 0.01126 0.79906 0.00651 0.84515
20 7 0.03367 0.66667 0.01598 0.76314 0.00929 0.81650
20 8 0.04546 0.61905 0.02183 0.72636 0.01277 0.78680
20 9 0.05957 0.57143 0.02896 0.68861 0.01705 0.75593
20 10 0.07632 0.52381 0.03760 0.64980 0.02228 0.72375
20 11 0.09606 0.47619 0.04801 0.60980 0.02866 0.69007
20 12 0.11926 0.42857 0.06054 0.56844 0.03644 0.65465
20 13 0.14653 0.38095 0.07567 0.52551 0.04595 0.61721
20 14 0.17863 0.33333 0.09403 0.48075 0.05768 0.57735
20 15 0.21667 0.28571 0.11653 0.43380 0.07230 0.53452
20 16 0.26220 0.23810 0.14455 0.38415 0.09088 0.48795
20 17 0.31760 0.19048 0.18031 0.33105 0.11517 0.43644
20 18 0.38693 0.14286 0.22776 0.27328 0.14839 0.37796
20 19 0.47802 0.09524 0.29521 0.20855 0.19756 0.30861
20 20 0.61118 0.04762 0.40646 0.13138 0.28394 0.21822
89
ISSN:0974-8652
TABLE 3
Sample Characteristics of MLE and MMLE of σ from complete Samples.
BIAS VARIANCE MSE

α n
MLE MMLE-I MMLE-II MLE MMLE-I MMLE-II MLE MMLE-I MMLE-II
2 5 0.17289 0.00784 0.35621 0.68399 0.52377 1.28796 0.71388 0.52383 1.41484
2 10 0.08545 0.01004 0.20054 0.26088 0.23200 0.40863 0.26818 0.23210 0.44884
2 15 0.05653 0.00637 0.13965 0.15975 0.14723 0.21541 0.16294 0.14728 0.23492
2 20 0.04262 0.00526 0.10723 0.11277 0.10649 0.14184 0.11458 0.10652 0.15333
3 5 0.11358 -0.00765 0.20724 0.46483 0.39623 0.60761 0.47773 0.39629 0.65055

3 10 0.05785 0.00617 0.12425 0.19810 0.18459 0.24237 0.20145 0.18463 0.25781
3 15 0.03881 0.00486 0.08977 0.12507 0.11938 0.14570 0.12657 0.11940 0.15376
3 20 0.02946 0.00440 0.07054 0.08965 0.08701 0.10095 0.09051 0.08703 0.10592
4 5 0.08450 -0.01788 0.14625 0.37894 0.34572 0.44431 0.38608 0.34604 0.46570

4 10 0.04394 0.00316 0.09012 0.17001 0.16379 0.19258 0.17194 0.16380 0.20070
4 15 0.02978 0.00342 0.06621 0.10915 0.10657 0.12061 0.11004 0.10658 0.12499
4 20 0.02271 0.00350 0.05266 0.07878 0.07784 0.08528 0.07930 0.07785 0.08805
90
ISSN:0974-8652
TABLE 4
Sample Characteristics of MLE and MMLE of σ from right censored samples (for α = 2)
BIAS VARIANCE MSE
n r MMLE- MMLE-
MLE MMLE-I II MLE MMLE-I II MLE MMLE-I MMLE-II
5 1 0.14677 -0.02913 0.20894 0.75861 0.58640 0.94666 0.78015 0.58725 0.99031
5 2 0.14934 -0.00262 0.17056 0.74603 0.56047 0.77531 0.76833 0.56047 0.80440
5 3 0.12808 -0.00075 0.13638 0.93202 0.73245 0.94696 0.94843 0.73245 0.96556
10 1 0.07112 -0.02030 0.12721 0.25736 0.22158 0.30009 0.26242 0.22200 0.31628

10 2 0.09819 0.01465 0.13395 0.30383 0.26172 0.32726 0.31347 0.26194 0.34521
10 3 0.08458 0.00637 0.10723 0.28847 0.24735 0.30036 0.29562 0.24739 0.31186
10 4 0.04250 -0.02704 0.05754 0.24044 0.20976 0.24795 0.24224 0.21049 0.25126
10 5 0.07066 0.00321 0.08056 0.33098 0.29022 0.33661 0.33597 0.29023 0.34310
10 6 0.05555 -0.00744 0.06133 0.35651 0.31527 0.36059 0.35959 0.31533 0.36435
10 7 0.03839 -0.02002 0.04159 0.45516 0.40512 0.45764 0.45663 0.40552 0.45937
10 8 0.03765 -0.01732 0.03917 0.65484 0.58754 0.65694 0.65626 0.58784 0.65848
15 1 0.04947 -0.01035 0.09825 0.16036 0.14506 0.18140 0.16280 0.14516 0.19105

15 2 0.06085 0.00433 0.09446 0.16020 0.14461 0.17209 0.16390 0.14463 0.18101
15 3 0.03111 -0.02191 0.05429 0.14769 0.13290 0.15483 0.14866 0.13338 0.15778
15 4 0.05043 -0.00044 0.06930 0.16553 0.15224 0.17473 0.16807 0.15224 0.17953
15 5 0.04336 -0.00582 0.05719 0.17373 0.15847 0.17941 0.17561 0.15850 0.18268
15 6 0.07767 0.02943 0.08910 0.19782 0.18005 0.20154 0.20385 0.18092 0.20948
15 7 0.04635 0.00088 0.05442 0.21385 0.19526 0.21685 0.21600 0.19526 0.21981
15 8 0.04085 -0.00244 0.04714 0.20875 0.19195 0.21156 0.21042 0.19196 0.21378
15 9 0.02159 -0.01924 0.02626 0.22239 0.20487 0.22437 0.22285 0.20524 0.22506
15 10 0.05605 0.01526 0.05940 0.28886 0.26696 0.29067 0.29200 0.26719 0.29420
15 11 0.02927 -0.00894 0.03156 0.31104 0.28833 0.31238 0.31190 0.28841 0.31337
15 12 0.04223 0.00482 0.04355 0.40826 0.37951 0.40933 0.41005 0.37953 0.41122
15 13 0.05583 0.01927 0.05644 0.63646 0.59306 0.63711 0.63958 0.59344 0.64030
20 1 0.04036 -0.00480 0.08016 0.11234 0.10486 0.12537 0.11397 0.10488 0.13179

20 2 0.03295 -0.00873 0.06336 0.10709 0.10014 0.11601 0.10818 0.10022 0.12003
20 3 0.03236 -0.00904 0.05447 0.10853 0.10025 0.11393 0.10958 0.10033 0.11690
20 4 0.04639 0.00733 0.06630 0.12636 0.11797 0.13240 0.12851 0.11802 0.13679
20 5 0.05041 0.01167 0.06595 0.12318 0.11505 0.12794 0.12572 0.11519 0.13229
20 6 0.04053 0.00350 0.05348 0.12842 0.11983 0.13225 0.13006 0.11984 0.13511
20 7 0.03492 -0.00041 0.04612 0.13426 0.12597 0.13818 0.13548 0.12597 0.14030
20 8 0.03444 -0.00094 0.04259 0.12860 0.11976 0.13043 0.12978 0.11976 0.13224
20 9 0.03900 0.00514 0.04632 0.14323 0.13466 0.14595 0.14475 0.13468 0.14810
20 10 0.04341 0.00962 0.04871 0.13905 0.13024 0.14055 0.14094 0.13033 0.14292
20 11 0.02592 -0.00598 0.03054 0.16622 0.15619 0.16788 0.16689 0.15623 0.16881
20 12 0.04569 0.01420 0.04957 0.18193 0.17112 0.18326 0.18402 0.17132 0.18572
20 13 0.02297 -0.00712 0.02585 0.17536 0.16514 0.17627 0.17589 0.16519 0.17694
20 14 0.03449 0.00490 0.03674 0.21107 0.19921 0.21204 0.21226 0.19923 0.21339
20 15 0.05060 0.02137 0.05232 0.26901 0.25418 0.26982 0.27157 0.25463 0.27255
20 16 0.01968 -0.00791 0.02086 0.29481 0.27904 0.29545 0.29519 0.27910 0.29589
20 17 0.03786 0.01050 0.03860 0.39870 0.37800 0.39931 0.40013 0.37811 0.40080
20 18 0.00903 -0.01691 0.00934 0.51541 0.48920 0.51567 0.51549 0.48948 0.51576
91
ISSN:0974-8652
ACKNOWLEDGEMENT:
We would like to express our thanks to referees for

valuable comments that improved the paper.
REFERENCES:
[1]. Balakrishnan, N. (1990). “Approximate

maximum likelihood estimation for a generalized
logistic distribution”, Journal of Statist. Plann. & inf.,
26, 221-236.
[2]. Bhattacharya.G..K.(1985). “The Asymptotics of

maximum likelihood and related estimators based on
Type II censored data”. Journal of American
Statistical Association. 80, 398-404.
[3]. Kantam R .R .L. & Sriram.B. (2003), “Maximum

Likelihood Estimation from censored samples –some
modifications in length biased version of exponential
Model”, Statistical methods, Vol. 5, 63-78.
[4]. Lloyd, E. H. (1952). “Least-squares estimation of

location and scale parameters using order
statistics”, Biometrika, 39, 88-95.
[5]. Tiku. M.L., Tan. W .Y. and Balakrishnan. N.

(1986). “Roboust Inference”, Marcel Decker, I. N.C.
New York.
[6]. Srinivasa rao. G. and Kantam. R. R .L. (2002) “A

note on point estimation of system reliability
exemplified for the Log-Logistic distribution”,
Economic quality control, 19(2), 197-204,
92
FROM
Dr.S.SATYANARAYANA
Editor-in-Chief –IJCMI
ISSN:0974-8652
D.NO:3-250.,PEDAKAKANI(POST)
GUNTUR(DIST).,A.P.,INDIA
E-Mail: ijcmi@in.com,ijcmi@yahoo.com,sijcmi@gmail.com
www.satyanspublications.weebly.com
http://www.ijcmi.webs.com/
To
HOD
Department of Mathematics & CSE
Sir,
Sub: “CALL FOR PAPERS TO INTERNATIONAL JOURNAL OF COMPUTATIONAL MATHEMATICAL IDEAS (IJCMI)”
About this Journal
The INTERNATIONAL JOURNAL OF COMPUTATIONAL MATHEMATICAL IDEAS (IJCMI) is a refereed Mathematics &Computer
science and Engineering journal devoted to publication of original research papers, research notes, and review articles, with emphasis on unsolved
problems and open questions in mathematics &Computer science and Engineering. All areas listed on the cover of Mathematical Reviews, such as
pure and applied mathematics.
The INTERNATIONAL JOURNAL OF COMPUTATIONAL MATHEMATICAL IDEAS (IJCMI) is an international research journal,
which publishes top-level work on computational aspects of mathematics interface between applied mathematics, numerical computation, and
applications of systems - oriented ideas to the physical, biological, social, and behavioral sciences. It includes but not limited to areas of mathematics
such as algebra (especially group theory), combinatorics (especially graph theory), geometry, number theory and numerical analysis; computational
complexity; cryptology; symbolic and algebraic computation; optimization; the mathematical aspects of: models of computation; automata theory;
categories and logic in computer science; proof theory; type theory; semantics of programming languages; process algebra and concurrent systems;
specification and verification; databases; rewriting; neural nets and genetic algorithms; computational learning theory; theorem proving ,Engineering
Physics,Applied Physics,Computational Physicsand more...
Journal publishes research articles and reviews within the whole field of Mathematical Sciences& Computer Science and Engineering, and
it will continue to provide information on the latest trends and developments in this ever-expanding subject.
Publication Date and Frequency: Three issues in a year.

Submission: Authors are requested to submit their papers electronically to: ijcmi@in.com with mention journal title (IJCMI) in subject line.
Annual Subscription Price:
INDIA Rs/- US $
1Year 2000 400
3Years 5800 760
5 Years 9500 1900
Editor-In-Chief-IJCMI

Ijcmi Vol2 No1 No2 2010

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ijcmi Vol2 No1 No2 2010

Uploaded by

Copyright:

Available Formats

INTERNATIONAL JOURNAL OF

COMPUTATIONAL MATHEMATICAL IDEAS

ISSN: 0974-8652 Volume 2 / Number 1 / January 2010

A Note on r - partitions of n in which The Least Part is k 6-12

On Completely Prime And Completely Semi-Prime Ideals In Γ-Near-Rings 22-27

Reducibility For The Fiorini-Wilson-Fisk Conjecture 33-42

Perceiving Plagiarism Using Weighted Window Approach- Performance Analysis 43-47

System Representation For Software Architecture Recovery 48-55

ISSN: 0974-8652 Volume 2 / Number 2 / August 2010

Over view to Implementation of robotics with Voice recognition 60-64

Novelty of Extreme Programming 65-72

Pareto Distribution - Some Methods Of Estimation 82-92

The INTERNATIONAL JOURNAL OF

INTERNATIONAL JOURNAL OF COMPUTATIONAL

A NOTE ON r − partitions OF n IN WHICH THE LEAST PART IS k

Key words: p ( n ) , r − partiions , p r ( n ) , pr ( e ; n ) , pr ( o ; n ) and pr ( S ; n )

1. Introduction: 1.7 pr ( S ; n ) : pr ( S ; n ) is the number of

The number λi is called the i th part of the S = {am + b | a ∈ N , b ∈ Z and m = 1,2,..., n} b

1.3 r-partition: A partition containing r parts is [ 2.1]

A Note On r - partitions Of n In Which The Least Part Is k

= pr ( S ; n − ar ) [4.1] n − {a ( 2 ) + b} of n can be expressed as the sum

equal to the number of ( r − 1) − partitions of

A Note On r - partitions Of n In Which The Least Part Is k

least part of the partition λ1 = a ( 2 ) + b is  n 

A Note On r - partitions Of n In Which The Least Part Is k

Proof: Put a = 1, b = 0 in [ 6.2] Proof: Case 1: Let n = 2m for m ∈ N

Theorem 8: Let n, i , j ∈ N , then

A Note On r - partitions Of n In Which The Least Part Is k

A Note On r - partitions Of n In Which The Least Part Is k

Since r + 1 ≤ m + 1 < 2r + 1 and 2r + 1 ≤ 2r + 2 Proof:

A Note On r - partitions Of n In Which The Least Part Is k

applications v.35, Cambridge, 1990.

A Note On r - partitions Of n In Which The Least Part Is k

INTERNATIONAL JOURNAL OF COMPUTATIONAL

MIXED CONVECTION IN A POROUS MEDIUM WITH MAGNETIC FIELD,

C.N.B.Rao$,V. Lakshmi Prasannam# , T.Raja rani*

NOMENCLATURE Rax – Rayleigh number,

Greek Symbols λ and ε where λ is the power of index of

A viscosity variation coefficient γµ is

Ra x values for opposing flow. When RP is zero, the

Heat transfer coefficient ‘ − θ ′(0) ’ takes

example λ = 0.05) , f ′′ ( 0 ) is observed to

results of the two works for C = 1, λ =0 and

Assisting flow (RP Positive)

1. For fixed values of λ , C , γ µ , as RP increases

2. For fixed values of γ µ , RP & λ ,

Opposing flow (RP Negative) 1.

2. f ′′ ( 0 ) takes positive as well as negative

3. Dual solutions exist for certain values of RP.

REFERENCES [10] White,F.M, Viscous Fluid Flow, McGraw-

[2] Aly,E.H, Elliot,L, Ingham,D.B, Mixed

[3] Barletta.A, Lazzari.S, Magyare.E and Pop.I,

[4] Chin,K,E, Nazar,R, Arifin,N.M, Pop,I,

[5]Gorsan.T, Revnic.C, Pop.I and Ingham.D.B,

[6] Lai,F.C, and Kulacki,F.A,(1990), The Effect

[7] Lakshmi Prasannam.V, Raja Rani,T and

[8] Nield.D.A, Bejan.A, Convection in Porous

[9] Sobha.V.V and Ramakrishna.K, Convective

INTERNATIONAL JOURNAL OF COMPUTATIONAL

ON COMPLETELY PRIME AND COMPLETELY SEMI-PRIME

AMS Subject Classification: 16 D 25, 16 Y 30, and 16 Y 99

1. Introduction 1.1 Definition: An algebraic system (N, +, .) is

Now, we collect some existing fundamental If I is an ideal of N then we denote it by I ⊴ N.