[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference rusure::math

Title:Mathematics at DEC
Moderator:RUSURE::EDP
Created:Mon Feb 03 1986
Last Modified:Fri Jun 06 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:2083
Total number of notes:14613

1382.0. "1990 IMO, Peking" by HERON::BUCHANAN (Holdfast is the only dog, my duck.) Wed Feb 06 1991 04:54

Q1.	A,C,B,D are four distinct points on a circle, such that the 
segments AB and CD cut at E.   M is a point on the segment BE, and is neither
B nor E.   The tangent through E of the circle passing through the three points
D,E & M cuts the straight lines through BC and AC at F and G respectively.

Calculate the quotient |EG|/|EF| as a function of t = |AM|/|AB|.


Q2.	On a circle are a set E of 2n-1 points (n an integer >=3).   If
exactly k of these points are coloured black, the colouration is said to be
"good" if there exist at least one pair of black points such that an *open*
arc [ie excluding the endpoints] contains exactly n points of E.

What is the smallest value of k for which every colouration in black of
exactly k points of E is good?


Q3.	Find all the integers n >= 1 such that (2^n + 1)/(n^2) is an integer.


Q4.	Let Q+ denote the strictly positive rationals.   Construct a function
f from Q+ to Q+ which satisfies:
		f(x*f(y)) = f(x)/y.


Q5.	Two players A & B alternately pick integers.   We designate by
n_1, n_3, ... , n_(2k+1) and n_2, n_4, ... , n_(2k+2) the numbers chosen
by A and B respectively.

An integer n_0 >= 2 is given:  A picks n_1 such that n_0 =< n_1 =< (n_0)^2,
then B picks n_2 such that n1/n2 is a perfect power of a prime number.   And
so on.   If B has chosen n_(2k), k>= 1, then A choses n_(2k+1) such that
n_(2k) =< n_(2k+1) =< (n_(2k))^2, and then B choses n_(2k+2) such that 
n_(2k+1)/n_(2k+2) is a perfect power of a prime number.

A wins if he succeeds in picking 1990.   B wins if he succeeds in picking 1.
(a) for which values of n_0 can A always win?
(b) for which values of n_0 can B always win?
(c) for which values of n_0 can each player prevent the other from winning?


Q6.	Prove that there exists a convex polygon with 1990 sides which
satisfies the following two properties:
(i) all the angles of the polygon are equal
(ii) the lengths of the sides form a permutation of the numbers
	1^2, 2^2, 3^2, ..., 1989^2, 1990^2.

Regards,
Andrew.
T.RTitleUserPersonal
Name
DateLines
1382.1I haven't looked at Q1...HERON::BUCHANANHoldfast is the only dog, my duck.Fri Feb 08 1991 13:50209
Q2.	On a circle are a set E of 2n-1 points (n an integer >=3).   If
exactly k of these points are coloured black, the colouration is said to be
"good" if there exist at least one pair of black points such that an *open*
arc [ie excluding the endpoints] contains exactly n points of E.

What is the smallest value of k for which every colouration in black of
exactly k points of E is good?

--------------------------------------------------------------------------------

A2.	Define a graph G, for which the vertex set is E, by stating that uv
is an edge in G iff an open arc linking u & v in the original circle contains
exactly n points.

	G must be regular of degree 2.   No point is linked to itself, or
twice to another, so G consists of a number of disjoint cycles.   What is
the size of those cycles?

	Starting from a point u, the length of a cycle is the minimum t
such that 2n-1 | t(n+1).   gcd(2n-1,n+1) = gcd(-3,n+1) 
	= 1 if n <> 2 mod 3	(=> t = 2n-1)
	= 3 if n == 2 mod 3	(=> t = (2n-1)/3)

	So, if n <> 2 mod 3, G consists of 1 odd cycle of 2n-1 elements.
Painting n of them black is necessary and sufficient to ensure that one edge
is incident to 2 black vertices.

	If n == 2 mod 3, G consists of 3 odd cycles of (2n-1)/3 elements.
Painting 3*[(2n-1)/6 - 1/2] + 1 of them black is necessary and sufficient to
ensure that one edge in one cycle is incident to 2 black vertices.   This
evaluates to n-1.

	So: for n <> 2 mod 3, k = n
	    for n == 2 mod 3, k = n-1

--------------------------------------------------------------------------------

Q3.	Find all the integers n >= 1 such that (2^n + 1)/(n^2) is an integer.

--------------------------------------------------------------------------------

A3.	Suppose p is prime, and n = q.p^a, where p does not divide q.
Firstly, it's obvious that p > 2, since 2^n+1 is odd.   Now we can write:

	p^(2a) | 2^(q.p^a) + 1

	The element coprime to p mod p^k form a cyclic group, G_p^k, order 
(p-1).p^(k-1) [dir /tit=Carmichael for proof, I think].   Let g generate 
G_p^(2a), and let 2 == g^r mod p^2a.   We know that -1 == g^((p-1)/2).p^(2a-1).
So:
	((p-1)/2).p^(2a-1) | r.q.p^a
=>	((p-1)/2).p^(a-1)  | r.q

p does not divide q, so p^(a-1) divides r.   If (p-1)/2 | r, then we have
a situation where:
(*)	2^(p^a) == -1 mod p^(2a)

We use this equation twice.   First, let's try p = 3.   Since (p-1)/2 = 1,
the hypothesis above is easily satisfied, and (*) holds.
	Suppose a >= 2.   Then
if	f(a) = (2^(3^a) + 1)
	f(a) = f(a-1)*(f(a-1)^2 - 3*f(a-1) +3]
So if 3^x|f(a), then 3^x+1|f(a+1), and no greater power.
	Since 3^2|f(1), have 3^(a+1)|f(a).   And hence (*) cannot hold for 
a > 1.   We have hence a = 1, if 3 divides n at all.

	Now. let p be the smallest prime > 3 dividing n.   The equation (*)
above implies:
	2 == -1 mod p	(since x^(p-1) == 1 mod p)
which is nonsense.   Therefore the hypothesis that (p-1)/2 | r is false, and
x = ((p-1)/2,q) <> 1.   But p is the smallest prime > 3.   So by the above
x = 1 or 3 => p = 3 (out) or 7.

	Suppose 2^n = -1 mod 7.   Then -2 is a square mod 7.   But this is
not so.   Therefore no such p exists.

n = 1 and 3 are the only solutions.

--------------------------------------------------------------------------------

Q4.	Let Q+ denote the strictly positive rationals.   Construct a function
f from Q+ to Q+ which satisfies:
		f(x*f(y)) = f(x)/y.

--------------------------------------------------------------------------------

A4.	Let p_i be the ith prime, then define:

	f(p_2i) = p_(2i+1)
	f(p_(2i+1)) = 1/p_2i

	Now extend this to a definition for all strictly positive rationals
through the requirement:
	f(x*y) = f(x)f(y)

	It's straightforward do check that this works.

--------------------------------------------------------------------------------

Q5.	Two players A & B alternately pick integers.   We designate by
n_1, n_3, ... , n_(2k+1) and n_2, n_4, ... , n_(2k+2) the numbers chosen
by A and B respectively.

An integer n_0 >= 2 is given:  A picks n_1 such that n_0 =< n_1 =< (n_0)^2,
then B picks n_2 such that n1/n2 is a perfect power of a prime number.   And
so on.   If B has chosen n_(2k), k>= 1, then A choses n_(2k+1) such that
n_(2k) =< n_(2k+1) =< (n_(2k))^2, and then B choses n_(2k+2) such that 
n_(2k+1)/n_(2k+2) is a perfect power of a prime number.

A wins if he succeeds in picking 1990.   B wins if he succeeds in picking 1.
(a) for which values of n_0 can A always win?
(b) for which values of n_0 can B always win?
(c) for which values of n_0 can each player prevent the other from winning?

--------------------------------------------------------------------------------

A5. (a)	If n_2k lies in {45,...,1990}, A can win immediately by saying 1990.
	If n_2k lies in {21,...,44}, A can win in two moves.   He first goes
to 420, and B's choices lie in {84,...,210}, which is a subset of {45,...,1990}
from which A wins as above.
	If n_2k lies in {16,...,20}, A can win in three moves.   He first goes
to 252, and B's choices lie in {28,...,126} which is a subset of {21,...,1990}
from which A wins as above.
	If n_2k lies in {10,...,15}, A can win in four moves.   He first goes
to 90, and B's choices lie in {18,...,45} which is a subset of {16,...,1990}
from which A wins as above.
	If n_2k lies in {8,9}, A can win in five moves.   He first goes to 60,
and B's choices lie in {12,...,30} which is a subset of {10,...,1990} from
which A wins as above.
	Thus A can win from n_2k in {8,...,1990}.
	If n_2k > 1990, then r > 3 can be found such that:
9.2^(r-1) < n_2k =< 9.2^r.   Then A plays to 9.2^r, and B's must reduce to
9 =< n_2(k+1) < n_2k.   This process is repeated until n_2l in {8,...,1990}
is reached, at which point A can win as above.
	So A can win from n_2k >= 8.

(b)	If n_2k = 2, then A's play lies in {2,3,4} and in each case B wins
in 1 move.
	If n_2k = 3, then A's play lies in {3,...,9}, and the choice which
is not a prime power (permitting an immediate win for B) is 6.   Then B goes
to 2 and wins as above.
	If n_2k = 4, then A's play lies in {4,...,16}, and those which are
not prime powers have at most two primes dividing, and one prime power
dividing is < 4.   So B goes to 2 or 3 and wins as above.
	If n_2k = 5, then A's play lies in {5,...,25}, and those which are
not prime powers have at most two primes dividing, and one prime power dividing
is < 5.   So B goes to 2,3 or 4, and wins as above.

	If n_2k = {2,3,4,5}, then B wins.

(c)	If n_2k = 6, then A's play lies in {6,...,36}.   Numbers with at
most two primes dividing lose as in (b) above.   So A's unique move is to
30, which has three primes dividing.   B must move to =< 7 or lose as in
(a).   So B goes to 6, A goes to 30, and so on.

	If n_2k = 7, then A's play lies in (7,...,49}.   Numbers with at
most two primes dividing lose as in (b) above.   So A must play to 30 or
42, which have three primes dividing.   30 is a draw as above.   On 42, B
must play to =< 7 and 6 is his only play.   So A goes to 30 and so on.

	Thus n_2k = 6,7 are draws.

	This exhausts all the possible starting positions.

--------------------------------------------------------------------------------

Q6.	Prove that there exists a convex polygon with 1990 sides which
satisfies the following two properties:
(i) all the angles of the polygon are equal
(ii) the lengths of the sides form a permutation of the numbers
	1^2, 2^2, 3^2, ..., 1989^2, 1990^2.

--------------------------------------------------------------------------------

A6.	Imagine we have such a polygon.   Then we can regard each of the
edges E_j, as a point in the complex plane:
	R_j*exp(2*pi*i*j/1990)
where R_j = |E_j|, for j=0,1989.

	That the 1990 points define a polygon is to say that the centre of
mass of the points is the origin.   That the polygon is convex is to observe
that each external angle is 2*pi*i*1990, and thus 1990 of them are necessary
to effect a rotation of 2*pi.   There is no possibility therefore that the
polygon crosses itself, and no internal angle is obtuse.

	So it remains to see if one can assign the R_j in such a way that
the centre of mass remains at 0, and so we get the correct set {R_j} = {j^2}
emerging.

	S_j = (j mod 199) + (j mod 5)*199
	S_j = S_k => j = k.
	0 =< S_j =< 1989 => S-j span every number from 0 to 994
	{S_j*exp(2*pi*j/1990)} has centre of mass 0, since it is composed
of 5 sets of 199 points and 199 sets of 5 points, each of which has centre
of mass 0, since the points are equally spread round the circle.

	Now examine each pair of points R_2j and R_(2j+1):

	R_2j = 4*S_j + 3 + (2*S_j + 1)^2 = (2*S_j + 2)^2
	R_(2j+1) = (2*S_j + 1)^2

	R_2j is always even, R_(2j+1) is always odd.   R_2j = R_2k => j=k.
R_(2j+1)=R(2k+1) => j=k.   Each R_j is a square, between 1^2 & 1990^2, thus
they span all possible values.   We have our polygon.

--------------------------------------------------------------------------------

Regards,
Andrew.