Sie sind auf Seite 1von 3

18/Feb/2015

IMC SEMINAR
Selection Test 1
Solutions
Problem 1 Alice and Bob play the following game: they start with an empty 2014 2014
matrix and take turns writing numbers in each of the 20142 positions. Once the matrix is filled,
Alice wins if the determinant is nonzero and Bob wins if the determinant is zero. If Alice goes
first, does either player have a winning strategy?
Solution. This is Putnam, 2008 A2. Bob wins using one of the following strategies.
First solution: Pair each entry of the first row with the entry directly below it in the
second row. If Alice ever writes a number in one of the first two rows, Bob writes the same
number in the other entry in the pair. If Alice writes a number anywhere other than the first
two rows, Bob does likewise. At the end, the resulting matrix will have two identical rows, so
its determinant will be zero.
Second solution: (by Manjul Bhargava) Whenever Alice writes a number x in an entry
in some row, Bob writes number x in some other entry of the same row. At the end the
resulting matrix will have all rows summing to zero, so it cannot have full rank.
Problem 2 Let P (x) be a polynomial of degree n such that P (x) = Q(x)P 00 (x), where Q(x)
is a quadratic polynomial and P 00 (x) is the second derivative of P (x). Show that if P (x) has at
least two distinct roots then it must have n distinct roots.
Solution. This is Putnam Problem B2 (1999).
First solution: Suppose that P does not have n distinct roots; then it has a root of multiplicity at least 2, which we may assume is x = 0 without loss of generality. Let xk be the
greatest power of x dividing P (x), so that P (x) = xk R(x) with R(0) 6= 0; a simple computation
yields
P 00 (x) = (k 2 k)xk2 R(x) + 2kxk1 R0 (x) + xk R00 (x).
Since R(0) 6= 0 and k 2, we conclude that the greatest power of x dividing P 00 (x) is xk2 .
But P (x) = Q(x)P 00 (x), and so x2 divides Q(x). We deduce (since Q is quadratic) that Q(x)
is a constant C times x2 ; in fact, C = 1/(n(n 1)) by inspection of the leading-degree terms
of P (x) and P 00 (x).P
Now if P (x) = nj=0 aj xj , then the relation P (x) = Cx2 P 00 (x) implies that aj = Cj(j 1)aj
for all j; hence aj = 0 for j n 1, and we conclude that P (x) = an xn , which has all identical
roots.
Second solution (by Greg Kuperberg): Let f (x) = P 00 (x)/P (x) = 1/Q(x). By hypothesis,
f has at most two poles (counting multiplicity).
Recall that for any complex polynomial P , the roots of P 0 lie within the convex hull of
P . To show this, it suffices to show that if the roots of P lie on one side of a line, say on

the positive side of the imaginary axis, then P 0 has no roots on the other side. That follows
because if r1 , . . . , rn are the roots of P ,
n

P 0 (z) X 1
=
P (z)
z ri
i=1

and if z has negative real part, so does 1/(z ri ) for i = 1, . . . , n, so the sum is nonzero.
The above argument also carries through if z lies on the imaginary axis, provided that z is
not equal to a root of P . Thus we also have that no roots of P 0 lie on the sides of the convex
hull of P , unless they are also roots of P .
From this we conclude that if r is a root of P which is a vertex of the convex hull of the
roots, and which is not also a root of P 0 , then f has a single pole at r (as r cannot be a root
of P 00 ). On the other hand, if r is a root of P which is also a root of P 0 , it is a multiple root,
and then f has a double pole at r.
If P has roots not all equal, the convex hull of its roots has at least two vertices.
Problem 3 Start with a finite sequence a1 , a2 , . . . , an of positive integers. If possible, choose
two indices j < k such that aj does not divide ak and replace aj and ak by gcd(aj , ak ) and
lcm(aj , ak ), respectively. Prove that if this process is repeated
a) it must eventually stop;
b) the final sequence does not depend on the choices made.
Solution. This is Putnam, 2008 A3.
a) First we note that for any integer m > k such that k - m we have mk = gcd (m, k) lcm (m, k)
and lcm (m, k) > m. From this one can see that the total product of all numbers is invariant
under the procedure. Assume the procedure never stops. The last element an can only increase
or stay unchanged. As it cannot be bigger than the total product, then there is a moment of a
time after which it will stay unchanged. After this moment the element an1 always increases
or stays unchanged. So there is a moment of time after which an1 will not change. Repeating
the above argument inductively we conclude that the process stops eventually.
b) The proof of this part can also be used for proving the first part. Let p1 , p2 , . . . , pN be
(j)
all prime factors of aj s and k with j = 1, n and i = 1, N be a power of pk in aj prime
factorization.Of course some of s could be equal to zero. We will describe our procedure in
terms of changes of s.
Consider our process: if j < k and aj - ak , then pair
N
Y
m=1

(j)
m

pm ,

N
Y

(k)
m

pm ,

m=1

will be changed by pair


N
Y




!
N
(j) (k)
(j) (k)
Y
min m ,m
max m ,m
pm
,
pm
.
m=1
m=1

Condition aj - ak is equivalent to the existence of at least one index m0 such that






(j)
(k)
(k)
(j)
(j)
(k)
min m
,

<

=
max

.
m
m
m
m
m
0
0
0
0
0
0
So the step of our process leads to the following: at least for one prime number lower degree
will move to the left along the sequence, and bigger one to the right, while the opposite move
(j)
is impossible for any prime pm . If we have the sequence of as where for all j sequences m
are ordered in non-decreasing order, then for any j < k we have aj | ak and we cannot make
any more steps. Otherwise we can find a pair which we can use for the procedure and continue
(j)
the process until getting the unique sequence with m ordered in non-decreasing order.
Problem 4 Let x1 , . . . , xn be positive real numbers and n 3. For convenience denote
xn+1 = x1 and xn+2 = x2 . Prove that
n
X
k=1

n
xk
.
xk+1 + xk+2
4

Solution. Let xk1 be the largest number among x1 , . . . , xn . Look at the k1th term of our sum:
xk 1
th
xk +1 +xk +2 and denote the larger term in the denominator by xk2 . Look at the k2 term of
1

k2
our sum: xk +1 +x
and denote the larger term in the denominator by xk3 , etc. We obtain a
k2 +2
2
sequence of indices k1 , k2 , . . ..
Since our inequality is cyclic, without loss of generality we can think that k1 = 1. At
each step of the above procedure we move to the right only by 1 or 2 indices, that is kj1 =
kj + 1 or kj + 2. Therefore after a certain number of steps we arrive at the index n 1 or n,
that is k` = n 1 or n for some `. Moreover, ` n/2, as we always move by at most two
indices. Notice that the (n 1)th and the nth term of our sum contains in the denominator x1 ,
which is the largest of all the xi . Thus, k`+1 = 1 = k1 .
Applying the AM-GM inequality yields

`
`
X
X
xk j
xkj
xk

xk+1 + xk+2
x
+ xkj +2
2xkj+1
j=1 kj +1
j=1
k=1
v
r
u l
u Y xkj
n
`
` 1
`
`t
=`
= .
2xkj+1
2
4
2`

n
X

j=1

Das könnte Ihnen auch gefallen