6th Putnam 1946

Problem A1

p(x) is a real polynomial of degree less than 3 and satisfies |p(x)| ≤ 1 for x ∈ [-1, 1]. Show that |p'(x)| ≤ 4 for x
∈ [-1, 1].

Solution

Straightforward.

Let p(x) = ax2 + bx + c, so p'(x) = 2ax + b. It is evidently sufficient to show that |2a + b| and |2a - b| ≤ 4. p(0)
= c, p(1) = a + b + c, p(-1) = a - b + c, so 2a + b = 3/2 p(1) + 1/2 p(-1) + 2 p(0). But |p(1)|, |p(-1)|, |p(0)| ≤ 1,
so |2a + b| ≤ 4. Similarly, 2a - b = 1/2 p(1) + 3/2 p(-1) + 2 p(0).

6th Putnam 1946

© John Scholes
[email protected]
3 Nov 1999
6th Putnam 1946

Problem A2

R is the reals. For functions f, g : R → R and x ∈ R define I(fg) = ∫1x f(t) g(t) dt. If a(x), b(x), c(x), d(x) are
real polynomials, show that I(ac) I(bd) - I(ad) I(bc) is divisible by (x - 1)4.

Solution

Let F(x) = I(ac) I(bd) - I(ad) I(bc). Then clearly F(1) = 0 (since all the integrals are over an empty range).
Differentiating, we get, F' = acI(bd) + bdI(ac) - adI(bc) - bcI(ad). So F'(1) = 0. Differentiating again: F'' =
a'cI(bd) + ac'I(bd) + abcd + b'dI(ac) + bd'I(ac) + abcd - a'dI(bc) - ad'I(bc) - abcd - b'cI(ad) - bc'I(ad) - abcd =
a'cI(bd) + ac'I(bd) + b'dI(ac) + bd'I(ac) - a'dI(bc) - ad'I(bc) - b'cI(ad) - bc'I(ad). So F''(1) = 1.

Differentiating again: F''' = a''cI(bd) + 2a'c'I(bd) + ac''I(bd) + a'bcd + abc'd + b''dI(ac) + 2b'd'I(ac) + bd''I(ac) +
ab'cd + abcd' - a''dI(bc) - 2a'd'I(bc) - ad''I(bc) - a'bcd - abcd' - b''cI(ad) - 2b'c'I(ad) - bc''I(ad) - ab'cd - abc'd =
a''cI(bd) + 2a'c'I(bd) + ac''I(bd) + b''dI(ac) + 2b'd'I(ac) + bd''I(ac) - a''dI(bc) - 2a'd'I(bc) - ad''I(bc) - b''cI(ad) -
2b'c'I(ad) - bc''I(ad). So F'''(1) = 1.

That is sufficient to prove the result. But notice that if we differentiate again, then just collecting the terms
that do not involve I(fg) we get (after some cancellation) 2a'bc'd + 2ab'cd' - 2a'bcd' - 2ab'c'd, which is not, in
general, zero for x = 1. So in general we do not have F(x) divisible by (x - 1)5.

6th Putnam 1946

© John Scholes
[email protected]
14 Dec 1999
6th Putnam 1946

Problem A3

ABCD are the vertices of a square with A opposite C and side AB = s. The distances of a point P in space
from A, B, C, D are a, b, c, d respectively. Show that a2 + c2 = b2 + d2, and that the perpendicular distance k
of P from the plane ABCD is given by 8k2 = 2(a2 + b2 + c2 + d2) - 4s2 - (a4 + b4 + c4 + d4 - 2a2c2 - 2b2d2)/s2.

Solution

Let Q be the point of the plane ABCD closest to P. Let O be the center of the square ABCD. Let QO make an
angle θ with AC. Then using the cosine rule we have: AQ2 = AO2 + OQ2 - 2AO·OQ cos θ, CQ2 = CO2 +
OQ2 + 2CO.OQ cos θ (*). Adding: AQ2 + CQ2 = 2AO2 + 2QO2. But a2 = AQ2 + k2 etc, so a2 + c2 = 2k2 + s2
+ 2QO2. Similarly, b2 + d2 = 2k2 + s2 + 2QO2. We have established that a2 + c2 = b2 + d2.

We have also established that 8k2 = 2(a2 + b2 + c2 + d2) - 4s2 - 8QO2 (**). The angle φ between QO and BD
is π/2 - θ. So cos φ = sin θ. Hence, going back to (*), AQ2 - CQ2 = -4 AO·OQ cos θ, and BQ2 - DQ2 = ± 4
AO·OQ sin θ. But AQ2 - CQ2 = a2 - c2 etc. So (a2 - c2)2 + (b2 - d2)2 = 16 AO2OQ2 = 8s2OQ2. Substituting in
(**) gives the required result.

6th Putnam 1946

© John Scholes
[email protected]
14 Dec 1999
6th Putnam 1946

Problem A4

R is the reals. f : R → R has a continuous derivative, f(0) = 0, and |f '(x)| <= |f(x)| for all x. Show that f is
constant.

Solution

Suppose f is not constant. Then take an interval [a, b] of length < 1/2 such that f(a) = 0, f(b) ≠ 0, and |f(b)| ≥
|f(x)| for x ∈ [a, b]. Now applying the mean value theorem to the interval gives an immediate contradiction.

6th Putnam 1946

© John Scholes
[email protected]
14 Dec 1999
6th Putnam 1946

Problem A5

Let T be a tangent plane to the ellipsoid x2/a2 + y2/b2 + z2/c2 = 1. What is the smallest possible volume for
the tetrahedral volume bounded by T and the planes x = 0, y = 0, z = 0?

Solution

Answer: √3 abc/2.

The normal at (x0, y0, z0) is (xx0/a2, yy0/b2, zz0/c2). So the tangent plane is x x0/a2 + y y0/b2 + z z0/c2 = 1.
This cuts the three axes at x = a2/x0, y = b2/y0, z = c2/z0. We can regard two of these lengths as defining the
base of the tetrahedron, and the third as forming its height. Hence its volume is a2b2c2/(6x0y0z0).

We wish to maximize x0y0z0. That is equivalent to maximising x02/a2 y02/b2 z02/c2. But we know that the
sum of these three numbers is 1, so their maximum product is 1/27 (achieved when they are all equal - the
arithmetic/geometric mean result). Hence x0y0z0 has maximum value abc/(3√3).

6th Putnam 1946

© John Scholes
[email protected]
14 Dec 1999
6th Putnam 1946

Problem A6

A particle moves in one dimension. Its distance x from the origin at time t is at + bt2 + ct3. Find an expression
for the particle's acceleration in terms of a, b, c and its speed v.

Solution

Differentiating, v = 3ct2 + 2bt + a (*), and hence the acceleration f = 6ct + 2b. So t = (f - 2b)/6c. Substituting
in (*) gives v = (f - 2b)2/12c + b(f - 2b)/3c + a. Hence 12cv = f2 - 4b2 + 12ac. So f = 2√(b2 + 3cv - 3ac).

6th Putnam 1946

© John Scholes
[email protected]
16 Dec 1999
6th Putnam 1946

Problem B1

Two circles C1 and C2 intersect at A and B. C1 has radius 1. L denotes the arc AB of C2 which lies inside C1.
L divides C1 into two parts of equal area. Show L has length > 2.

Solution

Let O1 be the center of C1, and O2 the center of C2. Let the line O1O2 meet the arc AB of C2 at P. If P lies
between O1 and O2, then the tangent to C2 at P divides C1 into two unequal parts and the area C1 ∩ C2 lies
inside the smaller part. Contradiction. So O1 must lie between P and O2. But now the arc AP is greater than
the segment AP, which is greater than AO1 = 1. Hence L > 2.

6th Putnam 1946

© John Scholes
[email protected]
14 Dec 1999
6th Putnam 1946

Problem B2

P0 is the parabola y2 = mx, vertex K (0, 0). If A and B points on P0 whose tangents are at right angles, let C
be the centroid of the triangle ABK. Show that the locus of C is a parabola P1. Repeat the process to define
Pn. Find the equation of Pn.

Solution

The gradient at the point (x1 , y1) is 1/2 m/y1. So the tangents at (x1 , y1), (x2 , y2) are perpendicular iff y1y2
= -m2/4. So we may take one point as (x, y) = (t2/m, t) and the other as (1/16 m3/t2, -1/4 m2/t). Hence the
centroid is (x, y), where x = 1/3 (t2/m + 1/16 m3/t2), y = 1/3 (t - 1/4 m2/t). But this lies on the parabola y2 =
m/3 (x - m/6). But since any value of t was possible and hence any value of y, any point of this parabola is a
centroid from some pair of points A, B.

Repeating, we get that the equation for Pn is y2 = m/3n(x - m/6(1 + 1/3 + ... + 1/3n-1) ) = m/3n(x - m/4 (1 -
1/3n) ).

6th Putnam 1946

© John Scholes
[email protected]
14 Dec 1999
6th Putnam 1946

Problem B3

The density of a solid sphere depends solely on the radial distance. The gravitational force at any point inside
the sphere, a distance r from the center, is kr2 (where k is a constant). Find the density (in terms of G, k and
r), and the gravitational force at a point outside the sphere. [You may assume the usual results about the
gravitational attraction of a spherical shell.]

Solution

A shell exerts no net attraction on a point inside and acts on a point outside as if all its mass was concentrated
at its center. So let the density at a radial distance r be ρ(r). Then G/r2 ∫0r 4πt2ρ(t) dt = kr2. Hence ∫0r t2ρ(t) dt
= k/(4Gπ) r4. Differentiating: r2ρ(r) = k/(Gπ) r3. So ρ(r) = k/(Gπ) r.

The force at the surface of the sphere (R) is kR2. Hence the force at a distance r > R from the center of the
sphere is kR4/r2. [We know the force is a simple inverse square law outside the sphere and it must be
continuous at the surface.]

6th Putnam 1946

© John Scholes
[email protected]
14 Dec 1999
6th Putnam 1946

Problem B4

Define an = 2(1 + 1/n)2n+1/( (1 + 1/n)n + (1 + 1/n)n+1). Prove that an is strictly monotonic increasing.

Solution

an = 2 (n+1)n+1/(nn (2n+1) ). Let f(n) = ln an/2 = (n + 1) ln(n + 1) - n ln n - ln(2n+1). Regard f as a function

So f ''(x) < 0 for all x > 0. Hence f '(x) is decreasing. But we can write f '(x) = ln(1 + 1/x) - 2/(2x + 1) which
tends to 0 as x tends to infinity. So f '(x) > 0 for all x > 0. Hence f(x) is strictly increasing for all positive x.
Hence f(n+1) > f(n) and so exp f(n+1) > exp f(n), which is the result we want.

6th Putnam 1946

© John Scholes
[email protected]
5 Mar 2002
6th Putnam 1946

Problem B5

Let m be the smallest integer greater than (√3 + 1)2n. Show that m is divisible by 2n+1.

Solution

This is a fairly well-known problem. (√3 - 1) < 1 and (we will show) (√3 + 1)2n + (√3 - 1)2n (*) is an integer.
Hence it must be the required integer.

But it is easy to check that (*) is the solution of a recurrence relation, in fact the relation un = 8un-1 - 4un-2, u0
= 2, u1 = 8. That establishes that (*) is an integer, and a trivial induction shows that (*) is divisible by 2n+1.

6th Putnam 1946

© John Scholes
[email protected]
14 Dec 1999
6th Putnam 1946

Problem B6

The particle P moves in the plane. At t = 0 it starts from the point A with velocity zero. It is next at rest at t =
T, when its position is the point B. Its path from A to B is the arc of a circle center O. Prove that its
acceleration at each point in the time interval [0, T] is non-zero, and that at some point in the interval its
acceleration is directly towards the center O.

Solution

This is almost trivial. Take polar coordinates with origin O. Let the radius of the circular arc be k. The
particle's radial acceleration is k (dθ/dt)2 towards O. Its tangential acceleration is k d2θ/dt2. Since the particle
is not at rest between A and B, (dθ/dt) is non-zero, so the radial acceleration is non-zero (and hence its total
acceleration).

(dθ/dt) is zero at A and at B, so its derivative (and hence the tangential acceleration) must be zero at some
point between.

6th Putnam 1946

© John Scholes
[email protected]
14 Dec 1999