2pt

Version : English
39th International Mathematical Olympiad
First Day { Taipei { July 15, 1998

Problem 1
In the convex quadrilateral ABCD, the diagonals AC and BD are perpendicular
and the opposite sides AB and DC are not parallel. Suppose that the point P ,
where the perpendicular bisectors of AB and DC meet, is inside ABCD. Prove
that ABCD is a cyclic quadrilateral if and only if the triangles ABP and CDP
have equal areas.
Solution:
Let AC and BD intersect at E . Suppose by symmetry that P is in 4ABE .
Denote by M and N the respective feet of perpendiculars from P to AC and BD.
Without assuming that PA = PB and PC = PD, we express the areas ABP ]
and CDP ] as follows1:
2ABP ] = 2ABE ] ; 2PAE ] ; 2PBE ]
= (AM + PN )(BN + PM ) ; (AM + PN )PM ; (BN + PM )PN
= AM  BN ; PM  PN
2CDP ] = 2CDE ] + 2PCE ] + 2PDE ]
= (CM ; PN )(DN ; PM ) + (CM ; PN )PM + (DN ; PM )PN
= CM  DN ; PM  PN:
Therefore
2(ABP ] ; CDP ]) = AM  BN ; CM  DN: ( )
itbpF 2:0159in1:9545in0in1 ; 1a:gif
We now assume that PA = PB and PC = PD. Suppose that ABCD is
cyclic. Then the uniqueness of P implies that it must be the circumcenter. So
M and N are the midpoints of AC and1 BD, respectively. Hence AM = CM
and BN = DN , so (*) implies ABP ] = CDP ]. Conversely, suppose that
ABP ] = CDP ]. Then, by (*), we have AM  BN = CM  DN . If PA 6= PC ,
assume by symmetry that PA > PC . Then AM > CM and also BN > DN ,
because PB > PD. Thus AM  BN > CM  DN , a contradiction. Hence
 Consider the median EM to the hypotenuse of the right triangle ABE . We
have \BEM = ', \AME = 2' and \EMP = 90 ; 2'. Likewise, \CEN = ,
\DNE = 2 and \ENP = 90 ; 2. Hence \MEN = 90 + ' + , and a
direct computation yields
\NPM = 360 ; (\EMP + \MEN + \ENP ) = 90 + ' +  = \MEN:
It turns out that, whenever AC ? BD, the quadrilateral EMPN has a pair of
equal opposite angles, the ones at E and P .
We now prove our claim. Since AB = 2EM and CD = 2EN , we have
ABP ] = CDP ] if and only if EM  PM = EN  PN , or, EM=EN = PN=PM .
On account of \MEN = \NPM , the latter is equivalent to 4EMN  4PNM .
This holds if and only if \EMN = \PNM and \ENM = \PMN , and these in
turn mean that EMPN is a parallelogram. But the opposite angles of EMPN
at E and P are always equal, as noted above. So it is a parallelogram if and only
if \EMP = \ENP  that is, if 90 ; 2' = 90 ; 2. We thus obtain a condition
equivalent to ' = , or to ABCD being cyclic.
Another Solution:
Choose a coordinate system so that the axes lie on the perpendicular lines
AC and BD, and so that the coordinates of A, B , C and D are (0 a), (b 0),
(0 c) and (d 0), respectively. By assumption, the perpendicular bisectors of AB
and CD have a unique common point. Hence the linear system formed by their
equations 2bx ; 2ay = b2 ; a2 and 2dx ; 2cy = d2 ; c2 has a unique solution,
and it is given by
x0 = ;c(b ;2(aad) ; + a(d2 ; c2)  y = ;d(b2 ; a2 ) + b(d2 ; c2) :
2 2

bc) 0
2(ad ; bc)
Naturally, (x0  y0) are the coordinates of P . Since it is interior to ABCD, 4ABP
and 4CDP have the same orientation. Then ABP ] = CDP ] if and only if
0 a 1 0 c 1
b 0 1 = d 0 1
x0 y0 1 x0 y0 1
2
This is equivalent to ax0 + by0 ; ab = cx0 + dy0 ; cd. Inserting the expressions
for x0 and y0, after the inevitable algebra work we obtain the equivalent condition

(ac ; bd)(a ; c)2 + (b ; d)2] = 0:

Now, the choice of the coordinate system implies that a and c have dierent
signs, as well as b and d. Hence the second factor is nonzero, so ABP ] = CDP ]
if and only if ac = bd. The latter is equivalent to AE  CE = BE  DE , where E
is the intersection point of the diagonals. However, it is a necessary and sucient
condition for ABCD to be cyclic.

3
 Problem 2
In a competition, there are a contestants and b judges, where b  3 is an odd
integer. Each judge rates each contestant as either \pass" or \fail". Suppose k
is a number such that, for any two judges, their ratings coincide for at most k
contestants. Prove that
k  b ; 1:
a 2b
Solution:
Since there are (b2) pairs of judges and each pair agrees on at most k contes-
tants, the total number of agreements is at most k (b2). For 1 i a, let the
i-th contestant be passed by xi judges and failed by yi judges, whence xi + yi = b.
Then the number of pairs of judges who agree on the this candidate is

(xi 2) + (yi2) = 1 (x2i + yi2 ; xi ; yi)

2 
1 1
 2 2 (xi + yi) ; b
2

= 14 (b ; 1)2 ; 1]:

Since b is odd, we may strengthen this lower bound to 41 (b ; 1)2 . It follows that

(xi 2) + (yi2)]  a(b ; 1) :

a 2
k (b2) 
X

i=1
4
This yields the desired result.
Comment:
The inequality in the problem is sharp. Let b = 2` + 1. Then it becomes
k
a
 2``+1 . Let a = (2` + 1`) and let each candidate be passed by a dierent
subset of ` judges. Then each judge passes exactly (2`` ; 1) contestants. Every
two judges pass (2` ; 1` ; 2) candidates in common, so that they fail

(2` + 1`) ; 2 (2`` ; 1) + (2` ; 1` ; 2) = (2`` ; 1) ; (2` ; 1` ; 2)

contestants in common. Hence we may take k = (2`` ; 1), and indeed ka = 2``+1 .
4
 The result is not valid for even b. Let b = 2` + 2. We still take a = (2` + 1`).
To the above example with 2` + 1 judges, add one who passes every candidate.
This judge agrees with every other one on (2`` ; 1) candidates also. Hence we
may take k = (2`` ; 1) again. However, now 2``+1 < b2;1b .

5
 Problem 3
For any positive integer n, let d(n) denote the number of positive divisors of n
(including 1 and n itself).
Determine all positive integers k such that
d(n2 ) = k
d(n)
for some n.
Solution:
For relatively prime positive integers a and b, d(ab) = d(a)d(b). Let n =
pk11 pk22  pkt t , where p1  p2 : : :  pt are distinct primes and k1  k2   kt are positive
integers. Then d(n) = (k1 + 1)(k2 + 1)  (kt + 1) while d(n2) = (2k1 + 1)(2k2 +
1)  (2kt + 1). It follows that d(n2 ) is always odd, so that the only acceptable
values for m are the odd integers. We now prove that every odd integer is
acceptable, that is,
m = 2kk1++11  2kk2++11  2kkt++11
1 2 t

for some positive integers k1 k2 : : :  kt.

A natural approach is by induction on m. The result holds for m = 1
since d(1 ) = 1. For any odd integer m > 1, if it is of the form 4s + 1, then
2

d(1)
4s
m = 2s + 1 (2s + 1). Since 2s + 1 < m, it is acceptable by the induction hypoth-
+1
esis. Hence m is also acceptable.
However, when m is of the form 4s + 3, m 2+ 1 is an even number. We have
to distinguish between the cases m = 8r + 3 and m = 8r + 7. In the former, we
have
m = 1224r + 9  12r + 5 (2r + 1):
r + 5 6r + 3
In the latter case, we have to split the numbers into two subclasses again. To
make this process to an end, we reformulate the above idea as follows.

6
 We claim that if x is acceptable, then so is 2k x ; 1 for any k  1. Let ` be
such that d(` ) = x. For k = 1, take n = px;1` where p is a prime not dividing
2

d(`)
`. Then d(nn)) = 2x x; 1  x = 2x ; 1. For k > 1, take
d ( 2

n = p21k;1 3x;2p22k;2 32 x;2  p23

k ;1
k;1 x;2 3k;1 x;1
pk `
where p1 p2 : : :  pk are distinct primes not dividing `. Then
d(n2) = 2k 3x ; 3  2k;132x ; 3  223k;1x ; 3  2  3k;1x ; 1  x = 2k x ; 1:
d(n) 2k;13x ; 1 2k;232x ; 1 2  3k;1x ; 1 3k;1x
This justies the claim.
We now prove that every odd integer is acceptable. We have already seen
that 1 is. For any odd integer m > 1, we can write m + 1 = 2k x for some odd
integer x < m. Since x is acceptable, m = 2k x ; 1 is also. Hence all odd integers
are acceptable.
Alternative Solution:
Let m be any odd positive integer, m > 1: We prove by induction that there
exists an n such that d(n ) = m: Put m + 1 = 2r m0 for some odd m0 < m: By
2

d(n)
induction hypothesis, there is some n0 such that d(n0) = m0: Put x0 = (2r ;
2

d(n0)
1)m0 ; 1 and xi = 2ix0 for i = 1 : : :  r: Note that 2xi = xi+1 for i = 0 : : :  r ; 1:
Let now p0 : : :  pr;1 be distinct primes, which are relatively prime to n0 : Consider
the number n = px0 0 : : : pxr;1
r ;1
 n0: We claim that n is what we search for. Indeed,
d(n2) = 2x0 + 1 : : : 2xi + 1 : : : 2xr;1 + 1  d(n20)
d(n) x0 + 1 xi + 1 xr;1 + 1 d(n0)
= x1 + 1 : : : xi+1 + 1 : : : xr + 1  m0
x0 + 1 xi + 1 xr;1 + 1
= xr + 1 m0:
x0 + 1
It remains to note that
xr + 1 m = xr + 1 m = 2r x0 + 1 = 2r (x0 + 1) ; 1 = 2r m ; 1 = m:
x0 + 1 0 (2r ; 1)m0 0 2r ; 1 2r ; 1 0
So, dd((nn)) = m as required.
2

Time Allowed : 4 21 hours.

Each problem is worth 7 points.
 Version : English
39th International Mathematical Olympiad
Second Day { Taipei { July 16, 1998

Problem 4
Determine all pairs (a b) of positive integers such that ab2 +b+7 divides a2 b+a+b.
Solution:
If x2 y + x + y is divisible by xy2 + y + 7, then so is the number

y(x2y + x + y) ; x(xy2 + y + 7) = y2 ; 7x:

Moreover, since x  1, we have y2 ; 7x < xy2 + y +7. It follows that if y2 ; 7x  0,
then y2 ; 7x = 0. In particular, 7 then divides y, so that

(x y) = (7z2  7z) for some positive integer z:

Conversely, if (x y) = (7z2  7z) with z being a positive integer, then x2 y + x + y =

7z(49z4 + z + 1) is divisible by xy2 + y + 7 = 7(49z4 + z + 1).
Suppose that y2 ; 7x < 0. Then the positive integer 7x ; y2, less than 7x, is
divisible by xy2 + y + 7. This is possible only if y = 1 or y = 2, since otherwise
xy2 + y + 7 > 9x.
For y = 1, we require 7x;1 being divisible by x+8. Now 7x;1 = 7(x+8);57,
and the factorization 57 = 1  57 = 3  19 shows that x = 11 or x = 49 are
the only possibilities. For (x y) = (11 1), we have xy2 + y + 7 = 19, which
divides x2y + x + y = 133. For (x y) = (49 1), we have xy2 + y + 7 = 57, which
divides x2 y + x + y = 2451.
For y = 2, we require 7x ; 4 being divisible by 4x + 9. Now 4(7x ; 4) =
7(4x + 9) ; 79, and 79 has no divisor of the form 4x + 9, so no solution arises.
Therefore, the complete solution is given by (x y) = (7z2  7z), z = 1 2 : : : ,
(x y) = (11 1) and (x y) = (49 1).
 Problem 5
Let I be the incentre of triangle ABC . Let the incircle of ABC touch the sides
BC , CA and AB at K , L and M , respectively. The line through B parallel to
MK meets the lines LM and LK at R and S , respectively. Prove that \RIS is
acute.
Solution:
Since the lines KL and RS are parallel, we have, in 4BKR
\BKR = 90 ; \A=2
\KBR = 90 ; \B=2
\BRK = 90 ; \C=2:
itbpF 1:8144in1:6293in0in3 ; 1b:gif
Hence, by the law of sine,
cos(\A=2)  BK:1
BR = cos( (1)
\C=2)
Similarly, we have, in 4BLS
\BLS = 90 ; \C=2
\BSL = 90 ; \A=2
\LBS = 90 ; \B=2
so that
BS = cos (\C=2)  BL = cos (\C=2)  BK:2 (2)
cos (\A=2) cos (\A=2)
Notice now that BI ? RS and IK ? AB: Then, on account of (1) and (2), we
obtain
IR2 + IS 2 ; RS 2 = ; (BR + BS )
;  ; 
BI 2 + BR2 + BI 2 + BS 2 2

2 BI 2 ; BR  BS
; 
=
2 BI 2 ; BK 2
; 
=
= 2IK 2 > 0:
So, by the law of cosine, \RIS is acute.
Alternative Solution:
Let W be the midpoint of KL and Q the midpoint of KM . Then Q 2 AI ,
W 2 BI , AI ?KM and BI ?KL. We rst prove that AW ?RI and CW ?SI .
itbpF 2:5581in2:418in0in3 ; 2:gif
Since \RBI = \RQI = 90 , the points R, B , I , Q are concyclic, which
implies \BRI = \BQI . Also, in the right triangles AIK and BIK , we have

IQ  IA = KI 2 = IW  IB
so that IQ=IW = IB=IA: Hence 4AIW and 4BIQ are similar. It follows that
\BRI = \BQI = \AWI . Since BR ? IW , this implies that RI ?AW . By a
similar argument, we can prove that SI ?CW . Thus, \RIS = 180 ; \AWC .
It remains to prove that \AWC is obtuse.
Let T be the midpoint of AC . Then 2;
;! = ;WC
WT ;! + ;WA
;! = ;!
LC + ;
;!. Since
KA
;!
LC and ; ;! are not collinear, we have
KA
WT < LC +2 KA = CM +2 AM = AC 2 :
This implies that W is inside the circle with diameter AC and so \AWC > 90 :
Therefore, \RIS is acute.
Comment:
Another proof for the fact AW ? RI is as follows. Since \RBI = \RQI =
90 , RBIQ is cyclic and its circumcircle is orthogonal to the diameter RI: Con-
sider the inversion
with respect to the incircle of 4ABC: Since
takes B and Q
into W and A respectively, it takes the circumcircle of RBIQ into the line AW .
Since
leaves the line RI invariant, we have AW ? RI:
 Problem 6
Consider all functions f from the set N of all positive integers into itself satisfying
; 
f t2 f (s) = s (f (t))2 
for all s and t in N . Determine the least possible value of f (1998).
Solution:
Denote by S the set of functions considered. Let f be any of them, and
let f (1) = a. Put t = 1 and s = 1 to obtain

f (f (s)) = a2 s f (at2) = f (t)]2 for all s t 2 N:

These together with the original equation give
; ; 
f (s)f (t)]2 = f (s)]2 f (at2) = f s2 f f (at2)
; 
= f (s2a2 at2) = f a(ast)2
= f (ast)]2 :

It follows that f (ast) = f (s)f (t) for any s t in particular, f (as) = af (s), and so

af (st) = f (s)f (t) for all s t 2 N:1 (3)

We now prove that f (t) is divisible by a for each t 2 N. For a given prime p,
denote by p and p the highest powers of p dividing a and f (t), respectively. It
follows by standard induction and (1) that f (t)]k = ak;1f (tk ) for all k 2 N. The
highest power of p dividing f (t)]k is pk  the one dividing ak;1 is p(k;1) . Hence
k  (k ; 1) for all k 2 N, which is possible only if   . The conclusion
holds for any prime p, and so, a divides f (t).
We may therefore set g(t) = f (t)=a, obtaining a new function g from N into
itself. The results proven above about f translate into

g(a) = a g(st) = g(s)g(t) g (g(s)) = s for all s t 2 N:2 (4)

In fact, g(st) = g(s)g(t) is equivalent to (1), and g (g(s)) = s follows from
ag (g(s)) = g(a)g (g(s)) = g (ag(s)) = g(f (s))
= f (f (s)) = a s = as:
2

a a
It is easy to infer from (2) that g (t2g(s)) = g(t2)g (g(s)) = s g(t)]2 for all s t 2 N.
So, g is also a function in S , and its values do not exceed the corresponding values
of f . Thus we may restrict attention on functions g satisfying (2). The essential
fact needed is that each function of this type takes a prime to a prime.
Indeed, let p be a prime, and let g(p) = uv for some positive integers u v.
Then, by (2), p = g (g(p)) = g(uv) = g(u)g(v), so that one of the numbers g(u)
and g(v) is 1. If, say, g(u) = 1, then u = g (g(u)) = g(1) = 1, which means that
g(p) is a prime.
To determine the desired minimum value, take any function g satisfying (2).
It is an injection (g(s) = g(t) implies s = g (g(s)) = g (g(t)) = t), and so
takes distinct primes to distinct primes. Hence a lower bound for g(1998) =
g(2  33  37) = g(2) g(3)]3 g(37) is obtained when g(2) g(3) g(37) are the three
smallest primes 2 3 5, with g(3) = 2. Thus, g(1998)  3  23  5 = 120 for
each g 2 S . There is, however, a function g 2 S with g(1998) = 120 this proves
that the minimum in question is 120. Set g(1) = 1, and dene g on the primes
in the following way: g(2) = 3, g(3) = 2, g(5) = 37, g(37) = 5 and g(p) = p for
each prime p 6= 2 3 5 37. The denition is then extended to an arbitrary t =
p1 1 p2 2  pk k 2 N by setting g(t) = g (p1 1 p2 2  pk k ) = g(p1)1 g(p2)2  g(pk)k .
The conditions in (2) are satised (with a = 1), so g 2 S . Clearly, g(1998) = 120,
which completes the proof.
Comment:
1. Actually, each function f considered above is of the form f (t) = ag(t),
where g is a multiplicative involution of N that is, a function satisfying (2)
with a = f (1). In turn, any multiplicative involution of N is uniquely
determined by its action on the primes.
 2. Divisibility properties are crucial for the solution: that a = f (1) divides f (t)
for each t 2 N, and that g takes a prime to a prime. They are unlikely to
be obtained in a straightforward manner, avoiding the careful study of the
functional equation. Standard and somewhat lengthy algebra work seems
inevitable to reveal the nature of the considered function.

Time Allowed : 4 21 hours.

Each problem is worth 7 points.