You are on page 1of 4

Volume 15, Number 4 January 2011

Olympiad Corner Klamkin’s Inequality


Below are the problems of the 2011
Kin Y. Li
Chinese Math Olympiad, which was
held on January 2011.

Problem 1. Let a1,a2,…,an (n≥3) be In 1971 Professor Murray Klamkin Example 3. Let a,b,c be sides of a
real numbers. Prove that established the following triangle with area Δ. If r,s,t are any real
n n
n numbers, then prove that
 ai2   ai ai 1    ( M  m) 2 ,
2
Theorem. For any real numbers x,y,z, 2
i 1 i 1
integer n and angles α,β,γ of any  ar  bs  ct  st tr rs
where an1  a1, M  maxai , m  minai ,      .
1i n 1i n triangle, we have  4  bc ca ab
[x] denotes the greatest integer not x2  y 2  z 2 
exceeding x. Solution. Let α,β,γ be the angles of the
(1)n 1 2( yz cos n  zx cos n  xy cos n ). triangle. We first observe that
Problem 2. In the figure, D is the
midpoint of the arc BC on the Equality holds if and only if 42  a2b2 sin2   b2c2 sin2   c2a2 sin2 
circumcircle Γ of triangle ABC. Point X x y z
is on arc BD. E is the midpoint of arc   . and cos 2θ = 1−2sin2 θ. So we can try to
sin n sin n sin n set n = 2, x=ar, y=bs, z=ct. Indeed, after
AX. S is a point on arc AC. Lines SD
and BC intersect at point R. Lines SE applying Klamkin’s inequality, we get
The proof follows immediately from the result.
and AX intersect at point T. Prove that
expanding
if RT || DE, then the incenter of triangle
Example 4. Let a,b,c be sides of a
ABC is on line RT. x  (1) ( y cos n  z cos n )
n 2
triangle with area Δ. Prove that
C D
 ( y sin n  z sin n ) 2  0. 2
R  a 2  b2  c2  a 2 b2 c2
   2  2  2 .
There are many nice inequalities that we  4  b c a
S
X
can obtain from this inequality. The
T following are some examples (see Comment: It may seem that we can use
A B references [1] and [2] for more). example 3 by setting r=a, s=b, t=c, but
unfortunately
E Example 1. For angles α,β,γ of any a 2 b2 c2
triangle, if n is an odd integer, then 3 2  2  2
(continued on page 4) b c a
Editors: 張 百 康 (CHEUNG Pak-Hong), Munsang College, HK cos n  cos n  cos n  3 / 2 . holds only when a=b=c by the AM-GM
高 子 眉 (KO Tsz-Mei)
梁 達 榮 (LEUNG Tat-Wing) If n is an even integer, then inequality.
李 健 賢 (LI Kin-Yin), Dept. of Math., HKUST
吳 鏡 波 (NG Keng-Po Roger), ITC, HKPU cos n  cos n  cos n  3 / 2. Solution. To solve this one, we bring in
the circumradius R of the triangle. We
Artist: 楊 秀 英 (YEUNG Sau-Ying Camille), MFA, CU
(This is just the case x=y=z=1.) recall that 2Δ=bcsin α and by extended
Acknowledgment: Thanks to Elina Chiu, Math. Dept.,
HKUST for general assistance. sine law, 2R=a/(sin α). So 4ΔR=abc.
Example 2. For angles α,β,γ of any Now we set r=bcx, s=cay and t=abz.
On-line:
http://www.math.ust.hk/mathematical_excalibur/ triangle, Then the inequality in example 3
becomes
The editors welcome contributions from all teachers and
students. With your submission, please include your name,
3 cos   2 cos   2 3 cos   4.
address, school, email, telephone and fax numbers (if ( x  y  z ) 2 R 2  yza 2  zxb 2  xyc 2 . (*)
available). Electronic submissions, especially in MS Word, (This is just the case n = 1, x = sin 90°,
are encouraged. The deadline for receiving material for the y = sin 60°, z = sin 30°.)
next issue is February 28, 2011.
Next, we set yz=1/b2, zx=1/c2, xy=1/a2,
For individual subscription for the next five issues for the
from which we can solve for x,y,z to get
09-10 academic year, send us five stamped self-addressed There are many symmetric inequalities b b2 c a
envelopes. Send all correspondence to: in α,β,γ, which can be proved by x  y , z .
ac 4R, 4R 4R
Dr. Kin-Yin LI, Math Dept., Hong Kong Univ. of Science standard identities or methods.
and Technology, Clear Water Bay, Kowloon, Hong Kong However, if we encounter asymmetric Then (*) becomes
Fax: (852) 2358 1643 inequality like the one in example 2, it 2
Email: makyli@ust.hk  a 2  b2  c2  a 2 b2 c2
may be puzzling in coming up with a    2  2  2 .
© Department of Mathematics, The Hong Kong University
proof.  4  b c a
of Science and Technology
Mathematical Excalibur, Vol. 15, No. 4, Jan. 11 Page 2

Example 5. (1998 Korean Math 2


 a   b  2  c  abc
2
n
cos  i n

Olympiad) Postive real numbers a,b,c    


 yz    zx    xy   xyz  4.
Prove that
 sin    cot  . i
satisfy a+b+c=abc. Prove that     i 1 i i 1

1 1 1 3 By the fact, there exists an acute triangle Solution. For n = 2, we have equality
   with angles α,β,γ such that
1 a 2
1 b 2
1 c 2 2 cos 1 cos  2 cos 1 cos 1
a b c   
 2 cos ,  2 cos  ,  2 cos . sin 1 sin  2 sin 1 sin 1
and determine when equality holds. yz zx xy
 0  cot 1  cot  2 .
Solution. Let a = tan u, b = tan v and c = Then the first equation becomes
tan w, where u,v,w > 0. Then For n = 3, α1, α2, α3 are angles of a
x  y  z  2( yz cos  zx cos  xy cos ). triangle, say with opposite sides a,b,c.
tan u+tan v+tan w = tan u tan v tan w,
From the equality case of Klamkin’s Let Δ be the area of the triangle. Now
which can be written as 2Δ = bcsin α1 = casin α2 = absin α3.
inequality, we get
Combining with the cosine law, we get
tan v  tan w x y z
 tan u   tan(v  w).   . cos 1 b 2  c 2  a 2
1  tan v tan w sin  sin  sin  cot 1  
sin 1 4
This implies u+v+w=nπ for some odd
Squaring both sides and using and similarly for cot α2 and cot α3. By
positive integer n. Let α = u/n, β = v/n
Klamkin’s inequality,
and γ = w/n. Taking x = y = z = 1 in a b c
Klamkin’s inequality (as in example 1), cos  , cos   , cos  , n
4 cos i
2 yz 2 zx 2 xy  sini
 2(bc cos 1  ca cos 2  ab cos 3 )
we have i 1

we can solve for x,y,z as


cos n  cos n  cos n  3 / 2 , 3
 a 2  b 2  c 2   4 cot  i .
which is the desired inequality. bc ca ab
x , y , z . i 1

Equality holds if and only if a = b = c 2 2 2


Cancelling 4Δ, we will finish the case
= 3. n = 3. For the case n > 3, suppose the
Example 7. (2007 IMO Chinese Team
case n−1 is true. We have
For the next two examples, we will Training Test) Positive real numbers u,v,w
introduce the following satisfy the equation u  v  w  uvw  4.
n
cos i  cos 1 cos 2 cos(1  2 ) 
Prove that  sin    
Fact: Three positive real numbers x,y,z i 1 i  sin1 sin2 sin(1  2 ) 
satisfy the equation vw uw uv
   u  v  w.  n cos  i cos(1   2 ) 
2 2
x +y +z +xyz = 4 2
(**) u v w    
 i 3 sin i sin(1   2 ) 
if and only if there exists an acute Solution. By the fact, there exists an acute
triangle with angles α,β,γ such that  cos 1 cos  2 cos(  (1   2 )) 
triangle with angles α,β,γ such that    
 sin1 sin 2 sin(  (1   2 )) 
x = 2cos α, y = 2cos β, z = 2cos γ.
u  2 cos  , v  2 cos  , w  2 cos  .
 n cos  i cos(1   2 ) 
Proof. If x,y,z > 0 and x +y +z +xyz = 4, 2 2 2
The desired inequality becomes    
 i 3 sin i sin(1   2 ) 
then x2, y2, z2 < 4. So 0 < x, y, z < 2.
Hence, there are positive α,β,γ < π/2 2 cos cos 2 cos cos 2 cos cos  cot1  cot2  cot(  (1  2 ))
 
such that cos cos cos
 n 
x = 2cos α, y = 2cos β and z = 2cos γ.  4(cos 2   cos 2   cos 2  ).   cot  i  cot(1   2 )
 i 3 
Substituting these into (**) and Comparing with Klamkin’s inequality, all n
simplifying, we get cos γ = −cos (α+β), we have to do is to take n = 1 and   cot  i .
which implies α+β+ γ = π. We can get i 1

the converse by using trigonometric 2 cos  cos  2 cos  cos 


x , y This finishes the induction.
identities. cos  cos 
References
Example 6. (36th IMO Shortlisted 2 cos  cos 
Problem) Let a,b,c be positive real z . [1] M.S.Klamkin, “Asymetric Triangle
cos  Inequalities,” Publ.Elektrotehn. Fak.
numbers. Determine all positive real
numbers x,y,z satisfying the system of Ser. Mat. Fiz. Univ. Beograd, No.
Example 8. (29th IMO Shortlisted 357-380 (1971) pp. 33-44.
equations
Problem) Let n be an integer greater than
x+y+z = a+b+c, 1. For i=1,2,…,n, αi > 0, βi > 0 and [2] Zhu Hua-Wei, From Mathematical
4xyz−(a2x+b2y+c2z) = abc. n n Competitions to Competition Mathe-
   
i 1
i
i 1
i  . matics, Science Press, 2009 (in
Solution. We can rewrite the second Chinese).
equation as
Mathematical Excalibur, Vol. 15, No. 4, Jan. 11 Page 3

Problem Corner yet colored and paint them blue. When the a contradiction.
plane contains three red points that are the
Other commended solvers: King’s
We welcome readers to submit their vertices of an equilateral triangle, then A
College Problem Solving Team
solutions to the problems posed below wins. Following the rules of the game, can
(Angus CHUNG, Raymond LO,
for publication consideration. The B stop A from winning?
Benjamin LUI), Anna PUN Ying
solutions should be preceded by the (HKU Math), The 7B Mathematics
solver’s name, home (or email) address Solution. LI Pak Hin (PLK Vicwood K.
T. Chong Sixth Form College), Anna Group (Carmel Alison Lam Foundation
and school affiliation. Please send Secondary School) and WONG Sze
submissions to Dr. Kin Y. Li, PUN Ying (HKU Math) and The 7B
Mathematics Group (Carmel Alison Lam Nga (Diocesan Girls’ School).
Department of Mathematics, The Hong
Kong University of Science & Foundation Secondary School).
Problem 358. ABCD is a cyclic
Technology, Clear Water Bay, Kowloon, The answer is negative. In the first 2n quadrilateral with AC intersects BD at
Hong Kong. The deadline for sending moves, A can color n red points on a line, P. Let E, F, G, H be the feet of
solutions is February 28, 2011. while B can color 2010n blue points. For perpendiculars from P to sides AB, BC,
each pair of the n red points A colored, CD, DA respectively. Prove that lines
Problem 361. Among all real numbers there are two points (on the perpendicular EH, BD, FG are concurrent or are
a and b satisfying the property that the bisector of the pair) that can be chosen as parallel.
equation x4+ax3+bx2+ax+1=0 has a real vertices for making equilateral triangles
root, determine the minimum possible with the pair. When n > 2011, we have Solution. U. BATZORIG (National
value of a2+b2 with proof. University of Mongolia), King’s
n
2   n(n  1)  2010n. College Problem Solving Team
Problem 362. Determine all positive  2 (Angus CHUNG, Raymond LO,
rational numbers x,y,z such that Benjamin LUI), Abby LEE Shing
Then B cannot stop A from winning.
1 1 1 Chi (SKH Lam Woo Memorial
x  y  z , xyz,   Other commended solvers: King’s
x y z Secondary School), LI Pak Hin (PLK
College Problem Solving Team (Angus Vicwood K. T. Chong Sixth Form
are integers. CHUNG, Raymond LO, Benjamin College), Anna PUN Ying (HKU
LUI), Andy LOO (St. Paul’s Co-ed Math), Anderson TORRES (São Paulo,
Problem 363. Extend side CB of College),Emanuele NATALE (Università Brazil) and WONG Sze Nga (Diocesan
triangle ABC beyond B to a point D di Roma “Tor Vergata”, Roma, Italy) and Girls’ School).
such that DB=AB. Let M be the Lorenzo PASCALI (Università di Roma
midpoint of side AC. Let the bisector of “La Sapienza”, Roma, Italy), WONG Sze
∠ABC intersect line DM at P. Prove Nga (Diocesan Girls’ School). A
that ∠BAP =∠ACB.
E
Problem 357. Prove that for every B
Problem 364. Eleven robbers own a positive integer n, there do not exist four H
treasure box. What is the least number integers a, b, c, d such that ad=bc and n2 < P F
of locks they can put on the box so that a < b < c < d < (n+1)2.
there is a way to distribute the keys of
the locks to the eleven robbers with no Solution. U. BATZORIG (National D C
University of Mongolia) and LI Pak Hin G
five of them can open all the locks, but
every six of them can open all the locks? (PLK Vicwood K. T. Chong Sixth Form
The robbers agree to make enough College).
Since ABCD is cyclic, ∠BAC =∠CDB
duplicate keys of the locks for this plan and ∠ABD =∠DCA, which imply ΔAPB
We first prove a useful
to work. and ΔDPC are similar. As E and G are
Fact (Four Number Theorem): Let a,b,c,d feet of perpendiculars from P to these
Problem 365. For nonnegative real be positive integers with ad=bc, then there triangles (and similarity implies the
numbers a,b,c satisfying ab+bc+ca = 1, exists positive integers p,q,r,s such that corresponding segments of triangles are
prove that a=pq, b=qr, c=ps, d=rs. proportional), we get AE/EB=DG/GC.
1 1 1 1 Similarly, we get AH/HD=BF/FC.
    2. To see this, let p=gcd(a,c), then p|a and p|c.
ab bc ca abc
So q=a/p and s=c/p are positive integers. If EH || BD, then AE/EB = AH/HD,
Now p=gcd(a,c) implies gcd(q,s)=1. From which is equivalent to DG/GC=BF/FC,
ad=bc, we get qd=sb. Then s|d. So r=d/s is and hence FG || BD.
***************** a positive integer and a=pq, b=qr, c=ps,
Solutions d=rs. Otherwise, lines EH and BD intersect at
**************** some point I. By Menelaus theorem and
For the problem, assume a,b,c,d exist as
Problem 356. A and B alternately color required. Applying the fact, since d > b > its converse, we have
points on an initially colorless plane as a, we get s>q and r>p. Then s≥q+1, r≥p+1
follow. A plays first. When A takes his AE BI DH
and we get    1,
turn, he will choose a point not yet d  rs  ( p  1)(q  1)  ( pq  1)2 EB ID HA
colored and paint it red. When B takes which is equivalent to
his turn, he will choose 2010 points not  ( a  1) 2  (n  1) 2 ,
Mathematical Excalibur, Vol. 15, No. 4, Jan. 11 Page 4

not all zeros, simple algebra yields For i from 1 to n, let ji be the largest
BI DG CF
   1, x=y=z=1. This is the only solution. element in Bi. However, by the
ID GC FB definition of the Bi sets, {j1,j2,…,jn}
and lines BD and FG also intersect at I. Comments: Some solvers overlooked the contradicts the n-divisiblity of S. So
possibility that x or y or z may be negative there are at most n−1 Bi’s.
Other commended solvers: Lorenzo
in applying the Cauchy Schwarz
PASCALI (Università di Roma “La
Sapienza”, Roma, Italy). inequality! Since m ≥ (n−1)2+1, one of the Bi must
have at least n elements. Then for S, we
Problem 359. (Due to Michel Other commended solvers: U. can choose n elements from this Bi with
BATAILLE) Determine (with proof) all BATZORIG (National University of ki included so that ki is divisible by all
real numbers x,y,z such that x+y+z ≥ 3 Mongolia) and Shaarvdorj (11th High the remaining n−1 integers. Therefore,
and School of UB, Mongolia), King’s College the least m is (n−1)2+1.
Problem Solving Team (Angus CHUNG,
x3  y3  z3  x4  y4  z4  2(x2  y2  z2 ). Other commended solvers: WONG
Raymond LO, Benjamin LUI), Thien
NGUYEN (Luong The Vinh High School, Sze Nga (Diocesan Girls’ School).
Solution. LI Pak Hin (PLK Vicwood
Dong Nai, Vietnam), Anna PUN Ying
K. T. Chong Sixth Form College),
(HKU Math), The 7B Mathematics Group
Paolo PERFETTI (Math Dept,
(Carmel Alison Lam Foundation Secondary
Università degli studi di Tor Vergata Olympiad Corner
School) and WONG Sze Nga (Diocesan
Roma, via della ricerca scientifica, (continued from page 1)
Girls’ School).
Roma, Italy) and Terence ZHU
(Affilated High School of South China Problem 3. Let A be a finite set of real
Problem 360. (Due to Terence ZHU,
Normal University). numbers. A1,A2,…,An are nonempty
Affiliated High School of Southern China
subsets of A satisfying the following
Normal University) Let n be a positive
Let x,y,z be real numbers satisfying the conditions:
integer. We call a set S of at least n distinct
conditions. For all real w, w2+3w+3 ≥
positive integers a n-divisible set if among (1) the sum of all elements in A is 0;
(w+3/2)2 implies (w2+3w+3)(w−1)2 ≥ 0.
every n elements of S, there always exist (2) for every xi∈Ai (i=1,2,…,n), we
Expanding, we get (*) w4+w3−2w2 ≥
two of them, one is divisible by the other. have x1+x2+⋯+xn > 0.
3w−3. Applying (*) to w=x,y,z and
adding, then using the conditions on Determine the least integer m (in terms of Prove that there exist 1≤i1<i2<⋯<ik ≤ n
x,y,z, we get n) such that every n-divisible set S with m such that

0  x3  y3  z3  x4  y4  z4 2(x2  y2  z2 ) elements contains n integers, one of them k


is divisible by all the remaining n−1 Ai1  Ai2    Aik  A.
n
 3( x  y  z )  9  0. integers.
Here |X| denotes the number of
Thus, for such x,y,z, we must have
Solution. Anna PUN Ying (HKU Math) elements in the finite set X.
equalities in the (*) inequality for x,y,z.
So x = y = z = 1 is the only solution. and proposer independently. Problem 4. Let n be a positive integer,
The smallest m is (n−1)2+1. First choose set S = {1,2,…,n}. For nonempty finite
Comments: For the idea behind this distinct prime numbers p1, p2, …, pn−1. For sets A and B of real numbers, find the
solution, we refer the readers to the i from 1 to n−1, let minimum of |A Δ S|+|B Δ S|+|C Δ S|,
article on the tangent line method (see where C =A+B ={a+b | a∈A, b∈B}, X
Math Excalibur, vol. 10, no. 5, page 1). 
Ai  pi , pi2 ,, pin 1  ΔY = {x | x belongs to exactly one of X
For those who do not know this method, or Y }, |X| denotes the number of
we provide the and let A be any nonempty subset of their elements in the finite set X.
union. Then A is n-divisible because
Proposer’s solution. Suppose (x,y,z) is among every n of the elements, by the Problem 5. Let n ≥ 4 be a given integer.
a solution. Let s=x+y+z and S=x2y+y2z pigeonhole principle, two of them will be For nonnegative real numbers
+z2x+xy2+yz2+zx2. By expansion, we in the same Ai, then one is divisible by the a1,a2,…,an, b1,b2,…,bn satisfying
have s(x2+y2+z2)−S= x3+y3+z3. Hence, other. However, among n elements, two of a1+a2+⋯+an = b1+b2 +⋯+bn > 0, find
s(x2+y2+z2)−S+x4+y4+z4 ≤ 2(x2+y2+z2), them will also be in different Ai’s and the maximum of
neither one is divisible by the other. So m
which is equivalent to ≤ (n−1)2 will not work.
n

(s−2)(x2+y2+z2)+x4+y4+z4 ≤ S. (*)
 a (a
i 1
i i  bi )
.
If m ≥ (n−1)2+1 and S is a n-divisible set n
Since S is the dot product of the vectors
with m elements, then let k1 be the largest  bi (ai  bi )
v =(x2,y2,z2,x,y,z) and w =(y,z,x,y2,z2,x2), i 1
element in S and let B1 be the subset of S
by the Cauchy Schwarz inequality,
consisted of all the divisors of k1 in S. Let Problem 6. Prove that for every given
S ≤ x2+y2+z2+x4+y4+z4 . (**) k2 be the largest element in S and not in B1. positive integers m,n, there exist
Combining (*) and (**), we conclude Let B2 be the subset of S consisted of all infinitely many pairs of coprime
(s−3)(x2+y2+z2) ≤ 0. Since s ≥ 3, we get the divisors of k2 in S and not in B1. Repeat positive integers a,b such that
s=3 and (*) and (**) are equalities. this to get a partition of S.
a+b | ama+bnb.
Hence, vectors v and w are scalar
multiple of each other. Since x,y,z are Assume there are at least n of these Bi set.

You might also like