Spring mathematics

tournament|1997
Problem 8'1. Given the equation jx ; aj + 15 = 6jx + 2j, where a
is a real parameter.
(a) Prove that for any value of a the equation has exactly two
distinct roots x1 and x2.
(b) Prove that jx1 ; x2j
6 and nd all values of a for which
jx1 ; x2j = 6.
Solution: (a) When x  ;2, the equation is equivalent to jx ;
aj = ;6x ; 27, which has a solution only if ;6x ; 27
0, i. e., if
x  ; 29 . Considering the cases of both x
a and x  a shows that
if x  ;2, the given equation has a unique root x1, and if x > ;2,
it has a unique root x2:
8 a ; 27 9 8 3 ; a
a  1
><
a < ;2 ><
7
x1 = > a + 27 9 x2 = > a +5 3
a > 12
:;
a
; :
5 2 7 2
1
(b) It follows from (a) that
8 156 ; 12a 9
>>
>< 35
a < ; 2
jx1 ; x2j = > 6
; 29  a  12
>> 12a + 204 1
:
a >
35 2
It remains to be seen that 156 ; 12a > 6 when a < ; 9 and
35 2
12a + 204 > 6 when a > 1 . Therefore jx ; x j
6 and equal-
35 2 1 2

ity obtains only when a 2 ; 92
12 ].

Problem 8'2. Let O be the intersecting point of the diagonals

of the convex quadrilateral ABCD and let 6 DAC = 6 DBC . The
midpoints of AB and CD are respectively M and N and P and Q
are points on AD and BC respectively such that OP ? AD and
OQ ? BC . Prove that MN ? PQ.

Solution: Denote by E and F the midpoints of AO and BO. Then

4PEM = 4MFQ since MF = 21 AO = PE , QF = 21 OB = ME
and 6 MEP = 6 MEO + 6 OEP = 6 MEO + 26 DAC = 6 MFO +
26 DBC = 6 MFQ (MFOE is a parallelogram). Therefore MP =
MQ. The case when E and F are interior points for 6 PMQ is
treated similarly (prove that there are no other possibilities). By
analogy we conclude that NP = NQ. Hence M and N lie on the
axis of symmetry of PQ and so MN ? PQ.

2
Problem 8'.3. Find all natural numbers n such that there exists
an integer number x for which 499(1997n + 1) = x2+x.

Solution: Let n be a solution of the problem. Then (2x + 1)2 =

1996  1997n + 1997. If n = 1 we get (2x + 1)2 = 19972 and so
2x + 1 = 1997. Therefore x = 998 and x = ;999 satisfy the
conditions of the problem. Let n
2. Now (2x + 1)2 is divisible
by 1997, which is a prime number, and so (2x + 1)2 is divisible by
19972 . But this is impossible, since 1996  1997n +1997 is not divisible
by 19972 when n
2. The only solution is n = 1.

Problem 8.1. Find all values of the real parameter m such that
the equation (x2 ; 2mx ; 4(m2 + 1))(x2 ; 4x ; 2m(m2 + 1)) = 0 has
exactly three distinct roots.

Solution: Suppose m satis es the conditions of the problem and

the equations
x2 ; 2mx ; 4(m2 + 1) = 0
x2 ; 4x ; 2m(m2 + 1) = 0
share the root x0. After subtracting we get (2m ; 4)x0 = (2m ;
4)(m2 +1) and so x0 = m2 +1 (note that if m = 2, the two equations
coincide). Substituting x0 in any of the equations gives the equation
(m2 + 1)(m2 ; 2m ; 3) = 0 with roots m = ;1 and m = 3. Direct
veri cation shows that the condition is satis ed only for m = 3.
Let now (1) and (2) share no roots. Since D1 = 4 + 5m2 > 0
(1) always has two distinct roots and therefore (2) should have equal

3
roots. Thus D2 = 4 + 2m(m2 + 1) = 0 and so m = ;1. But this
case has already been considered. Thus we determine that m = 3.

Problem 8.2. The area of the equilateral triangle ABC is 7. Points

M and N are chosen respectively on AB and AC so that AN = BM .
Denote by O the intersecting point of the straight lines BN and CM .
The area of BOC is 2.
(a) Prove that MB : AB = 1 : 3 or MB : AB = 2 : 3.
(b) Find 6 AOB .

Solution: (a) Denote MB AB = x. Therefore SABN = 7x = SBMC

and so SBOM = 7x ; 2 and SAMON = SBOC = 2. Further SCON =
; 7x) ,
7 ; 2 ; 2 ; (7x ; 2) = 5 ; 7x, SANO = 1 ;x x  SCNO = x(51 ; x
1 ; x 1 ; x
SAMO = x  SBOM = x (7x ; 2). It follows from SAMON =
SANO + SAMO that 2 = x(51 ;; 7x) + 1 ; x (7x ; 2), and thus 9x2 ;
x x
9x + 2 = 0: The roots of the above equation are x1 = 31 and x2 = 23 .
(b) Since 4ABN  = 4BMC, we get 6 BOM = 6 BCM +
6 CBO = 6 MBO + 6 CBO = 60
. Further 6 MAN + 6 MON = 180

and therefore the quadrilateral AMON is inscribed in a circle. Let
MB = 1 , i. e., AM = 2BM = 2AN . Denote by Q the mid-
AB 3
point of AM . Triangle AQN is isosceles and has an angle equal to
60
, so it is equilateral. Therefore Q is the circumcentre of AMON
and 6 AOM = 6 ANM = 90
. Thus 6 AOB = 150
. Similarly, if
4
MB = 2 , i. e., 2AM = MB = AN , we get 6 AMN = 6 AON = 90
,
AB 3
so 6 AOB = 90
.

Problem 8.3. Given n points, n
5, in the plane such that

no three lie on a line. John and Peter play the following game:
On his turn each of them draws a segment between any two points
which are not connected. The winner is the one after whose move
every point is an end of at least one segment. If John is to begin the
game, nd the values of n for which he can always win no matter
how Peter plays.

Solution: Call a point isolated if it is not an end of a segment.

John wins exactly when there are 1 or 2 isolated points before his
last move. Peter is forced to reach the above only if before his move
there are exactly 3 isolated points and any of the remaining n ; 3
points are connected by a segment. Indeed, if there are at least 4
isolated points he could connect one of them with a non-isolated
point. If the isolated points are 3 but not all of the remainig n ; 3
points are connected he could draw a missing segment. Since the
number of segments with ends in n ; 3 points is (n ; 3)(2 n ; 4) ,
we determine that John wins only when (n ; 3)(2 n ; 4) is an odd
integer number. This is true when n is of the form 4k + 1 or 4k + 2.

Problem 9.1. Let f (x) = x2 ; 2ax ; a2 ; 43 where a is real param-

eter. Find all values of a such that the inequality jf (x)j  1 holds
for any x in the interval 0
1].

5
Solution: Let M and m be the maximum and minimum values
of f (x) in the interval 0
1]. Then the condition of the problem is
equivalent to M  1 and m
;1. There are three cases to consider.
Case 1: a 2 0
1]. Then m = f (a) = ;2a2 ; 43 and M = f (0) =
;a2 ; 43 or M = f (1) = ;a2 ; 2a + 14 . It follows from m
;1 and
p p
M  1 that a 2 ; 42
42 ] and a 2 (;1
; 32 ] ; 12
1). In this
p
case the solution is a 2 0
42 ].
Case 2: a < 0. Now m = f (0) and M = f (1). From m
;1
and M  1 we get that a 2 ; 12
12 ]: Therefore a 2 ; 21
0).
Case 3: a > 1. Now m = f (1) and M = f (0). It follows from
m
;1 and M  1 that a 2 ; 25
12 ] which is a contradiction with
a > 1. p
1
Thus the solution is a 2 ; 2
42 ].

Problem 9.2. Let I and G be the incentre and the centre of

4ABC with sides AB = c, BC = a, CA = b.
(a) Prove that (if a > b) the area of CIG equals (a ; b)r where r
6
is the inradius of ABC:
(b) If a = c +1 and b = c ; 1, prove that the segment IG is parallel
to AB and nd its length.

6
Solution: (a) We shall use the usual notation for a triangle. Let
CL and CM be respectively the bisector and the median from C .
It follows from CG = 32 CM that SCIG = 32 SCIM . Thus SCIM =
SCLM ; SILM = LM2 hc ; LM2  r = LM 2 (hc ; r). We nd from
AL + BL = c and BL AL = b that AL = bc and so LM = AM ;
a a+b
c bc c ( a ;
AL = 2 ; a + b = 2(a + b) . Also, hc ; r = 2cS ; r = 2pr
b )
c ;r =
r(2p ; c) = r(a + b) . Therefore S = 2 S = 2  c(a ; b) 
CIG
c c 3 CIM 3 4(a + b)
r(a + b) = (a ; b)r .
c 6
(b) The distances from I and G to AB are respectively r and
hc . Hence r = S = chc = chc = hc and so IGkAB . Therefore
3 p 2p 3c 3
the altitude from C of triangle CIG equals 23 hc = 2r. Thus SCIG =
IG  r. On the other hand SCIG = (a ;6 b)r = r3 and so IG = 13 .

Problem 9.3. Let n
n be an even number and A be a subset

of f1
2
: : :
ng. Consider the sums of the form "1x1 + "2x2 + "3x3,
where x1
x2
x3 are integer numbers in A (not necessarily distinct),
"1
"2
"3 (at least one of which is not 0) belong to f;1
0
1g and
none of the elements of A appears with coecients 1 and ;1 in any
of the sums. Call A a `free' set if n divides none of the above sums.
(a) Construct a `free' set having  n4 ] elements (x] is the least in-
teger number less than or equal to x).
7
 (b) Prove that no set of  n4 ] + 1 elements is `free'.

Solution: (a) The set A = f1
3
: : :
2 n4 ] ; 1g is `free' and has

 n4 ] elements (prove it!).
(b) Let n = 4m and suppose that A f1
2
: : :
ng is a `free' set
having  n4 ] + 1 = m + 1 elements. Without loss of generality assume
that A f1
2
: : :
2mg (since we can replace x 2 A by n ; x). We
shall show that there exist two elements of A whose sum is equal
to another element of A. Indeed, let a1 < a2 <    < am+1 be the
elements of A and consider the set B = fa1 + ai : i = 1
2
: : :
m +
1g. There are 2m + 2 integer numbers a1
a2
: : :
am+1
2a1
a1 +
a2
: : :
a1+am+1 from A B and they lie in the interval a1
a1+am+1],
which contains exactly am+1 + 1  2m + 1 integer numbers. This
gives ai = a1 + aj for some i
j . But then a1 + aj ; ai = 0, which
is impossible. The case n = 4m + 2 is settled in a similar fashion.
Notice that 2m + 1 cannot be an element of a `free' set.

Problem 10.1. Find the least natural 3number a such that the
equation cos (a ; x) ; 2 cos (a ; x)+cos 2a  cos( x
2 x +
2a 3
 )+2 = 0
has a root.

Solution: The roots of the our equations are the common roots of
cos (x ; a) = 1 and cos 32x
a  cos( x +  ) + 1 = 0. The roots of
2a 3
the rst one are x = a + 2n
n = 0
1
2
: : : and the roots of the
second one are x = 2a(k ; 1 )
k = 0
1
: : : : Therefore a = 6n
3 6k ; 5
8
for some integer numbers n and k. It is easy to see now that the
least natural number with the required property is a = 6.

Problem 10.2. Point F lies on the base AB of a trapezoid ABCD

and is such that DF = CF . Let E be the intersecting point of AC
and BD and O1 and O2 are circumcentres of ADF and FBC respec-
tively. Prove that the straight lines FE and O1O2 are orthogonal.

Solution: Let k1 and k2 be circles with centres O1 and O2 and let

the intersecting points of the two circles be points P and Q. It is well
known that PQ ? O1O2. On the other hand, if L is an arbitrary
point and two lines through L intersect k1 and k2 in points A, B
and C , D respectively. Then L 2 PQ if and only if LA  LB =
LC  LD. Let k1 and k2 be the circumscribed circles of 4AFD
and 4FBC and let G be the intersecting point of FE with CD.
Denote by C1 and D1 those points on DC for which AD1jjCF and
BC1jjDF , i. e., such that the quadrilaterals AFCD1 and BFDC1
are parallelograms. Using that FD = FC we get 6 CFB = 6 FCD =
6 FDC = 180
; 6 BC1C . This means that F , B , C and C1 lie on
a circle and so the line DC intersects k2 in C and C1. By analogy
6 AFD = 6 FDC = 6 FCD = 6 AD1D and so line DC meets k1
in points D and D1. In accordance with the initial notes FE is
perpendicular to OO1 if and only if GC  GC1 = GD  GD1 . It follows
from 4GCE  4FAE , 4GDE  4FBE and 4DCE  4BAE
that GC =
AF EA AB
CE = DC and GD = DE = DC . Thus GC =
BF EB AB
DC  AF and GD = BF  DC . On the other hand GC = jDC ;
AB AB 1 1

DGj = jBF ;DGj = BF j1; DG BF j = BF j1 ; DC j = BF jAB ;DC j,

AB AB
9
GD1 = jCD1 ; CGj = jAF ; CGj = AF j1 ; CG
AF j = AF j 1 ; DC j =
AB
AF jAB ;DC j. Therefore GC GC = DC  AF  BF jAB ;DC j =
AB 1
AB 2
GD  GD1 .

Problem 10.3. Find all natural numbers n for which a convex n-

gon can be partitioned into triangles through its diagonals in such a
way that there is an even number of diagonals from each vertex. (If
there is a vertex with no digonals through it, assume that there is
an even number (zero) of diagonals from this vertex).

Solution: It is easy to see by induction that if an n-gon is par-

titioned into triangles through d non-intersecting diagonals then
n = d + 3. Let n be a natural number and A1A2    An is a convex
n-gon which can be partitioned into triangles through d diagonals
in a way that there is an even number of diagonals through each
vertex. Since n = 3 is a solution we may assume that n
4.
It is clear that at least one side of each triangle is a diagonal of the
n-gon. We say that a triangle is of type tk , (k = 1
2
3) if exactly k of
its sides are diagonals. Denote by xk the number of triangles of type
tk . It is easy to see that 2x1 + x2 = n = d +3 abd x1 +2x2 +3x3 = 2d.
It follows now that x1 = x3 + 2, so x1 > 0. Therefore there exists
a triangle two of whose sides are sides of the n-gon. Let that be
Aj;1Aj Aj+1. Diagonal Aj;1Aj+1 is a side of another triangle|e. g.,
Aj;1Aj+1As. Assume that Aj;1As or Aj+1As is a side of the n-gon.
If it is Aj;1As then s = j ; 2. It follows now that there are no
diagonals from Aj;1 distinct from Aj;1Aj+1 because such a diagonal
intersects Aj+1As. This contradicts the premise that there is an even

10
number of diagonals from each vertex. Therefore both Aj;1As and
Aj+1As are diagonals so Aj;1Aj+1As is of type t3. Hence there is a
triangle of type t3 adjacent to each triangle of type t1. If distinct
triangles of type t1 are adjacent to distinct triangles of type t3 then
x1  x3 = x1 ; 2 < x1, a contradiction. Therefore there are at
least two triangles of type t1 adjacent to one and the same triangle
of type t3. Without loss of generality assume these are the triangles
A1AnAn;1 and An;1An;2An;3 . Consider the polygon A1A2    An;3.
Obviously the diagonals partition this polygon into triangles and
there is an even number of diagonals through each vertex.
Conversely, if the polygon A1A2    An;3 can be partitioned in the
required way, then adding the vertices An;2, An;1 , An and diagonals
An;3An;1 and A1An;1 shows that the same is true for the polygon
A1A2    An.
Therefore a natural number n
6 is a solution if and only if
n ; 3 is a solution. It is easy to see that n = 3 is a solution, whereas
n = 4 and n = 5 are not. Thus all natural numbers satisfying the
conditions of the problem are n = 3k, k = 1
2
: : :.

Problem 11.1. For any real number b denote by f (b) the maximal
value of j sin x + 3 + 2sin x + bj. Find the minimal value of f (b).

Solution: Substitute t = sin x and g(t) = t + 3 +2 t + b. Since

g(t) is an increasing function in the interval ;1
1], it follows that
f (b) = max(jg(;1)j
jg(1)j) = max(jbj
jb + 23 j). Now from the graph
of the function f (b) we conclude that min f (b) = f (; 34 ) = 43 .
11
 Problem 11.2. A convex quadrilateral ABCD is such that 6 DAB =
6ABC = 6 BCD. Let H and O be respectively the orthocentre and
the circumcentre of 4ABC . Prove that H , O and D lie on a line.

Solution: Let 6 CAB = , 6 ABC =  , 6 BCA =  . Note that

 <  and  <  . There are three cases to consider:  < 90
,
 = 90
and  > 90
. Suppose rst that  < 90
. Then O and
H are interior points for 4ABC and 6 ACO = 6 CAO = 6 HCB =
6 HAB = 90
;  . Therefore O is an interior point for 4HAC and
6 HAO =  ;  = 6 ACD, 6 HCO =  ;  = 6 CAD, 6 HAD =
6 HCD = 2 ; 90
. It follows from the Sine Theorem for 4AHD,

4CHD and 4ACD that sin sin 6 AHD = AD , sin 6 HCD = HD ,

6 HAD HD sin 6 CHD CD
sin 6 CAD = CD . By multiplying the above equalities we get
sin 6 ACD AD
sin 6 AHD sin 6 HCO sin 6 CAO = sin 6 CHD sin 6 HAO sin 6 ACO:
It follows now from Ceva's Theorem that AO, CO and HD intersect
in a point and so H , O and D lie on a line. In the case of  = 90

we obtain that H B , O is the midpoint of AC and AHCD is a
rectangle. Therefore H , O and D lie on a line. Finally, let  > 90
.
In this case B and O are interior points for 4AHC and 4ADC
respectively. Similarly to the case  < 90
we get that the points H ,
O and D lie on a line.

Problem 11.3. For any natural number n
3 denote by m(n)

the maximum number of points which can be placed inside or on
the outline of a regular n-gon with side 1 in a way that the distance
between any two of them is greater than 1. Find all n for which
m(n) = n ; 1.
12
Solution: We prove rst that m(n) = n ; 1 for n = 4
5
6. Let n
be one of the above numbers and let B1
B2
  
Bn be points satis-
fying the conditions of the problem for the regular n-gon A1A2    An
of side 1 and centre O. It is obvious that OBi  1 and therefore
no three points O, Bi, Bj , 1  i = 6 j  n lie on a line. Further-
more at least one of the angles OBi Bj is less than 2n  90
and
it follows from the Cosine Theorem that BiBj2  OBi2 + OBj2 ;
2OBi OBj cos 2n = Bi0Bj0 2 where Bi0 and Bj0 are points on the seg-
ments OA1 and OA2 such that OBi0 = OBi , Bj0 = OBj . When
n = 4
5
6 the greatest side in 4OA1A2 is A1A2 = 1 and therefore
BiBj  Bi0Bj0  1, which is a contradiction. Thus m(n)  n ; 1
if n = 4
5
6. It is easily seen that for these n there exist n ; 1
points on the outline of a regular n-gon A1A2    An with the re-
quired property. Therefore m(n) = n ; 1 if n = 4
5
6. We shall
prove now that if n
7 then m(n)
n. Let Bi 2 AiAi+1 be points
such that AiBi = 2i;1 ", 1  i  n ; 1 where 0 < " < n1;1 is arbi-
2
trary chosen. From n
7 we obtain cos 6 A1A2A3 = ; cos 2 < ; 1 :
n 2
It follows now that B1B22 > (1 ; ")2 +4"2 +2"(1 ; ")2 = 1+3"2 > 1:
Similarly BiBi+1 > 1 when 1  i  n ; 2. Further, it is clear
that B1Bn;1 > A1An = 1. Since OA1 = OA2 =    = OAn > 1,
AiAj > 1 when ji ; j j
2 and Bi ! Ai when " ! 0, it follows
that we can make " so small that OBi > 1 when 1  i  n ; 1 and
BiBj > 1 when ji ; j j
2. Then the points B1
B2
  
Bn;1
O
satisfy the conditions of the problem and so m(n)
n when n
7.
Since it is obvious that m(3) = 1, we come to the conclusion that
m(n) = n ; 1 only when n = 4
5
6.

13