International

OPEN

Journal

ACCESS

Of Modern Engineering Research (IJMER)

Integration by Parts for DK Integral

T. K. Garai1, S. Ray2
1

Department of Mathematics, Raghunathpur College, Raghunathpur-723133, Purulia, West Bengal, India

2
Department of Mathematics, Siksha Bhavan, Visva-Bharati, Santiniketan West Bengal,India

ABSTRACT: In this paper we have defined Dk integral and proved the integration by parts formula.
Key Words and phrases: Absolutely Continuous function, Generalised absolutely continuous function,
Denjoy integration. 2000 Mathematics subject Classification: Primary 26A24 Secondary 26A21, 26A48,
44A10.

I. PRELIMINERIES

Let f be a real valued function defined on a set E. Let c, d E, c d , and k 2 the oscillation of f on

[c, d ] E of order k is defined to be

Ok ( f ,[c, d ] E ) sup (d c)[ f , c, x1 , x2 ,...xk 1 , d ]
where the sup is taken over all points x1,x2,...xk-1 on [c, d] E and [f,c,x1,...,xk-1,d]
kth order divided difference of f at the k+1 points c,x1,...,xk-1,d
The weak variation of f of order k is defined as follows,

represents the

Vk ( f , E ) sup Ok ( f ,[ci , d i ] E )
i

where the sup is taken over all sequences {(ci,di)} of non overlapping intervals with end points on E. Then f is
said to be of kth variation in the wide sense if Vk ( f , E ) and it is written as f BVk (E ) . The function
f is said to be kth absolutely continuous on E if for any

0 there is 0

non overlapping intervals (ci,di) with end points on E and with

O ( f , [c , d ] E )
k

such that for every sequence of

(d

ci )

we have

and we write it as f AC k (E ) .

th
th
The function F is said to be generalised absolutely k continuous (resp. of generalised bounded k
variation) on E if E Ei where each Ei is closed and f AC k ( Ei ) (resp. f BVk ( Ei ) for each i
and we write it as f AC k G(E ) (resp. f BVk G(E ))

II. AUXILIARY RESULT

Following result will be needed which are proved in [2]
Lemma 2.1 Let E be a closed set. Then f AC k G(E ) (resp. f BVk G(E )) if and only if every closed
subset of E has a portion on which f is ACk( resp.BVk.)
Lemma 2.2 The classes of functions AC (E),BV (E),AC G(E),BV G(E)
k
k
k
k
(k)
Theorem 2.3 If f BVk ( Ei ) then fap exists finitely a.e on E, where
of f.
Theorem 2.4 Let k2 and f : [a, b] R be such that

1
2

are all linear spaces.

(k)
fap is repeated approximate derivative

The research work of first author is financially supported by UGC, MRP.

Corresponding auther.
| IJMER | ISSN: 22496645 |

www.ijmer.com

| Vol. 5 | Iss.1| Jan. 2015 | 61|

Integration By Parts For D Integral

K
i) f AC k G([a, b]) D[a, b]
( k 2)

ii) f ap

exists in [a,b]

(r)
(r)
(r)
iii) if for r=0,1,...,k-2 and x (a,b) one of (fap )+'(x) and (fap )-'(x) exists then (fap )'(x) exists.(this condition is
(r)
weaker than the smoothness condition of f ).
ap
iv) f ap 0 a.e
(k )

( k 1)
(k-1)
Then f
exists and is nondecreasing in [a,b] and. f
AC1 ([a, b])
The following theorem is proved for k=2 in Theorem-4 of [2], and similarly can be proved any k.
Theorem 2.5 Let F and G be AC (resp.BV ) on E . Then FG is ACk (resp.BV ) on E. Then FG is
k
k
k
AC k (resp.BVk) on E .
Corollary 2.6 If F and G are ACkG (resp.BVkG) on [a,b] then FG is so in [a,b]

The proof is similar of the corollary of Theorem-4 of [2].

III. THE Dk INTEGRAL

A function f : [a, b] R is said to be Dk integrable on [a,b] if there exists a continuous function

: [a, b] R such that
i) f AC k G([a, b])
ii)

( k 2)
ap

exists in [a,b]

iii)

(r )
ap

iv)

( k 1),
ap

is smooth in (a,b) for r=0,1,...,k-2.

( k 1),
(k )
( k 1),
( k 1),
(a) , ap
(b) exists and ap
f a.e in [a,b]. Where ap
(a) and ap
(b) denote

right hand approximate derivative of

order k-1. (The existence of
The function

(k )
ap a.e

at a of order k-1 and left hand approximate derivative of

at b of

in [a,b] is guaranteed by Theorem-2.3)

( k 1)
if exists is called kth primitive of f on [a,b]. If F ap we call F to be an indefinite

Dk integral of f and
( k 1),
( k 1),
ap
(b) ap
(a) F (b) F (a)

is called the definite Dk integral of f over [a,b]and is denoted by

b
(Dk)
f(t)dt.
a
The definite Dk integral is unique by Theorem-2.4. The indefinite Dk integral is unique upto an additive
th
constant and the k primitive is unique upto an addition of polynomial of degree k-1.
Theorem 3.1 Let f be D integrable on [a,b] and on [b,c] and let and be kth primitives of f in [a,b]
k
and [b,c] respectively. If

( k 1), (b)
b

Proof: Since

and

( k 1), (b)
c

exists then f is D integrable in [a,c] and

k

(D )
f+(D )
f=(Dk)
f.
k
k
a
b
a
( k 1),
( k 1),
ap
(b) and ap
(b) exists, the previous derivative exists in some left and right

neighbourhood of

b . Let

| IJMER | ISSN: 22496645 |

www.ijmer.com

| Vol. 5 | Iss.1| Jan. 2015 | 62|

Integration By Parts For D Integral

K

Then H is continuous in [a,c]. We show that H is the kth primitive of f in [a,c]. Clearly H ACk G([a, c]) .
Also for 0 r k 2,

(r )
H ap
( x)

( k 2 )

And hence H ap

exists in [a,c]. We are to show that H ap is smooth at b for 0 r k 2. Let h 0 and

(r )

0 r k 2. Then
(r )
(r )
(r )
H ap
(b h) H ap
(b h) 2 H ap
(b)

(r )
(r )
H ap
(b h) H ap
(b)

(r )
(r )
H ap
(b h) H ap
(b)

h
h
h
(r )
( r ),
( r 1),
(r )
( r ),
( r 1),
ap (b h) (b) h
(b) ap (b h) (b) h
(b)

O ( h)
h
h

(b) ( r 1), (b) and the second term tends to ( r 1), (b) ( r 1), (b) as
(r )
(r )
is smooth at b for 0 r k 2. Hence H ap ( x) is smooth on (a,c) for 0 r k 2.
h 0. So H ap

The first term tends to

( r 1),

The proof of other properties of kth primitive are easy. Hence H is the kth primitive of f on [a,c]. So f is Dk
integrable on [a,c]. Also
c

( k 1),
( k 1),
( Dk ) f H ap
(c) H ap
(a)
a

{ ( k 1), (b) ( k 1), (c)} { ( k 1), (a) ( k 1), (b)}

{ ( k 1), (b) ( k 1), (a)} { ( k 1), (c) ( k 1), (b)}
b

( Dk ) f ( Dk ) f
Theorem 3.2 Let f be Dk integrable in [a,b] and let a<c<b. Let
exists then f is Dk integrable on [a,c] and on [c,b] and
b
c
(Dk)
f+(D
)

k f=(Dk)
a
b

th
the k primitive of f in [a,b]. If

( k 1)
ap
(c )

f g is Dk

integrable

f.

a
The proof is immediate.
Theorem 3.3 Let f and g be Dk integrable in [a,b] and
in [a,b] and

are constants then

( Dk ) (f g ) ( Dk ) f ( Dk ) g
| IJMER | ISSN: 22496645 |

www.ijmer.com

| Vol. 5 | Iss.1| Jan. 2015 | 63|

Integration By Parts For D Integral

K
The proof follows from the definition of the integral and Lemma 2.2.
Theorem 3.4 If f be D integrable on [a,b], then f is measurable and finite a.e on [a,b].
k
Proof: Since f is Dk integrable on [a,b], there is a continuous function ACk G([a, b])

(r )
ap is

smooth for r 1,2,...k 2 and

(r )
ap

such

that

f , a.e on [a,b]. Let [a,b]= En , where En is closed and is

AC k on En for each n. Let En Pn D n where Pn is perfect and Dn is countable. Since

ACk G( En ) , ACk G( Dn ) and so by Theorem-6(i) of [3], ( k 1) is AC1 on Pn where the

( k 1)
( k 1)
derivative
is taken with respect to Pn . Let n ( x)
( x) for x Pn and n be linear in the
closure of all intervals contiguous to Pn . Then n BV1[a, b] and so n exists a.e on [a,b]. Also
(k )
f ap
( x) n ( x) a.e in En . Since derivative of a function of bounded variation is finite a.e and
measurable, f is finite a.e and is measurable on En for each n. The rest is clear.
Theorem 3.5 If f is D integrable and f0 in [a,b], then f is Lebesgue integrable and the integrals are equal.
k
Proof: Let

(k )
f 0 a.e in [a,b]. Then from Theorem-2.4, ( k 1) exists and
be the kth primitive of f . So ap

is nondecreasing in [a,b] and

AC1 ([a, b]). Hence
integrable in [a,b]. Hence f is Lebesgue integrable in [a,b].
( k 1)

(k )

exists a.e in [a,b] and is Lebesgue

Theorem 3.6 (Integration by parts). Let f : [a, b] R is Dk integrable in[a,b], F be its indefinite Dk
integral and F is D* integrable. Let G(k-1) be absolutely continuous in [a,b], then fG is Dk integrable in [a,b]
and
b

( Dk ) fG [ FG ] ( D ) FG
b
a

Proof: Let

be the k

th

primitive of f in [a,b] .Let

( x) x Gx , x [a, b]. Then since ap

(r )

is smooth

(r )
ap
is smooth for r=0,1,,k-2. Since G(k-1) is
absolutely continuous ,G is ACk in [a,b] and since is ACkG in [a,b] . And a.e in [a,b]
(k )
ap( r ) G ( k ) k c1 G ( k 1) ....... k ( k 1) G ap
G

for r=0,1,,k-2 and G(k-1) exists finitely so

G ( k ) k c1 G ( k 1) ........ kFG fG
(k ) k
( k 1)
So G c1 G
....... kFG fG is Dk integrable in [a,b].
Let

H ( x) G ( k 1) k 1c1 G ( k 2) ...... (k 1) ( k 2)G

Then in a similar way as that of Theorem-14 of [1] it can be proved that H ACG ([a, b]) .
*

Again since

( k 1)

and G(k) exists a.e in [a,b],we have,

H ( x) G ( k ) ( k 1c0 k 1c1 ) G ( k 1) .....(k 1) ( k 1) G

G ( k ) k c1 G ( k 1) ........ (k 1) FG
G ( k ) k c1 G ( k 1) ... (k 1) FG is D* integrable in [a,b] and so Dk integrable and H is indefinite Dk
=

So

integral. Since F is D* integrable and G is absolutely continuous in [a,b], FG is D* integrable by

Theorem2.5(p.246) of [4] and hence FG is Dk integrable in [a,b]. Again since,

........ kFG fG ), (G c1 G
( G c1 G
Dk integrable, by Theorem3.3, fG is Dk integrable in [a,b] and
(k )

( k 1)

| IJMER | ISSN: 22496645 |

(k )

www.ijmer.com

( k 1)

........ (k 1) FG) and FG are

| Vol. 5 | Iss.1| Jan. 2015 | 64|

Integration By Parts For D Integral

K
b

( Dk ) fG ( Dk ) (G ( k ) k c1 G ( k 1) ........ kFG fG )
b

( D* ) (G ( k ) k c1 G ( k 1) ........ (k 1) FG) ( D * ) FG
a

a
b

] [ H ]ba ( D * ) FG

= [ ap

( k 1) b
a

= [G

( k 1)

c1 G
k 1

( k 2 )

........ (k 1) ( k 2) G FG ]ba
b

[G ( k 1) k 1c1 G ( k 2) ........ (k 1) ( k 2) G]ba ( D * ) FG

a
b

= [ FG]a ( D ) FG
b

REFERENCES
[1.]
[2.]
[3.]
[4.]

S.N.Mukhopadhyay and S.K.Mukhopadhyay,"A generalised integral with application to trignometric series".

Analysis Math., 22(1996) 125-146.
th
S. N. Mukhopadhyay and S. Ray, "Generalised Absolutely k Continuous Function". Indian Journal of
Mathematics, Vol 53, No 3(2011) 459-466.
th
th
S.K.Mukhopadhyay and S.N.Mukhopadhyay,"Functions of bounded k variation and absolutely k continuous
functions". Bull.Austral.Math.Soc 46(1992),91-106.
S.Saks,"Theory of the integral",Dover,Newyork(1937).

| IJMER | ISSN: 22496645 |

www.ijmer.com

| Vol. 5 | Iss.1| Jan. 2015 | 65|