G13 1948 PUTNAM Web Solution
G13 1948 PUTNAM Web Solution
Problem A1
C is the complex numbers. f : C → R is defined by f(z) = |z3 - z + 2|. What is the maximum value of f on the
unit circle |z| = 1?
Solution
Answer: √13.
Put z = eiθ, and use cos 2θ = 2 c2 - 1, cos 3θ = 4 c3 - 3c, where cos θ = c, to get: |f(z)|2 = 6 - 2 cos(3θ - θ) + 4
cos3θ - 4 cos θ = 4(4c3 - c2 - 4c + 2). The cubic has stationary points where 12c2 - 2c - 4 = 0 or c = 2/3 or
-1/2. So the maximum value is at c = 1 or -1 (the endpoints), or 2/3 or -1/2 (the stationary points).
Substituting in, we find that the maximum is actually at c = -1/2 with value 13.
© John Scholes
[email protected]
4 Nov 1999
8th Putnam 1948
Problem A2
K is a cone. s is a sphere radius r, and S is a sphere radius R. s is inside K touches it along all points of a
circle. S is also inside K and touches it along all points of a circle. s and S also touch each other. What is the
volume of the finite region between the two spheres and inside K?
Solution
Slice off the top of a sphere radius r by a cut a distance d from the surface. A simple integration shows that
the volume removed is π d2(3r - d)/3.
Let h be the distance from the vertex of the cone along the axis to the nearest sphere. Similar triangles gives
(h+r)/r = (h+2r+R)/R. Hence h = 2r2/(R-r). Let the plane through the circle of contact between the small
sphere and the cone cut the cone's axis at a distance h+d from its vertex. Similarly, let the plane through the
circle of contact of the other sphere cut the cone's axis a distance D from the point of contact between the two
spheres. Let t be the distance from the cone's vertex to the circle of contact with the small sphere. Then t2 =
h(h+2r). By similar triangles (h+d)/t = t/(h+r), so d = 2r2/(R+r). Hence by similar triangles D = 2rR/(R+r).
Let v be the volume in the cone between the vertex and the small sphere. We find this as the volume of a
cone with circular base less the volume of a slice of sphere. The cone has height h+d. The square of the
radius of the base is (2rd-d2). Hence v = π/3 ( (h+d)(2rd-d2) - d2(3r-d) ). This simplifies to 4r5π/(3(R2-r2)).
Hence V the volume of the corresponding region between the vertex and the large sphere (assuming the small
sphere is temporarily removed) is 4r2R3π/(3(R2-r2)). Hence the required volume is V - v - 4πr3/3 =
4πr2R2/3(R+r).
Comment. There are more elegant ways of doing this, but this type of solid geometry is not currently in
fashion and no one is likely to remember them. In an exam one has no time to look for such things.
© John Scholes
[email protected]
11 Mar 2002
8th Putnam 1948
Problem A3
Solution
We have that b1 + 2b2 + 3b3 + ... + n bn = a1 - (n+1)an+1 + nan+2 = a1 - (n+1)(an+1 - an+2) - an+2. Also we are
given that all bi are non-negative. So S n bn is monotonic increasing and bounded above by a1. So it
converges to a limit L ≤ a1.
bn = (an - an+1) - (an+1 - an+2). Hence bn + bn+1 + bn+2 + ... + bn+m = (an - an+1) - (an+m+1 - an+m+2). But am
tends to zero, so (an+m+1 - an+m+2) tends to zero as m tends to infinity. Hence bn + bn+1 + bn+2 + ...
converges to an - an+1. Hence (n+1)(bn+1 + bn+2 + bn+3 + ... ) converges to (n + 1)(an+1 - an+2). Hence L ≥ b1
+ 2b2 + 3b3 + ... + n bn + (n+1)(bn+1 + bn+2 + bn+3 + ... ) = a1 - (n+1)an+1 + n an+2 + (n+1)(an+1 - an+2) = a1 -
an+2. But an+2 tends to zero. Hence L = a1.
© John Scholes
[email protected]
11 Mar 2002
8th Putnam 1948
Problem A4
Let D be a disk radius r. Given (x, y) ∈ D, and R > 0, let a(x, y, R) be the length of the arc of the circle center
(x, y), radius R, which is outside D. Evaluate limR→0 R-2 ∫D a(x, y, R) dx dy.
Solution
Answer: 4πr.
Let P be a point in the disk a distance x from its centre. Suppose that the circle centre P radius R cuts the disk
perimeter at A and B. Let angle APB be 2θ. We have r2 = x2 + R2 + 2xR cos θ. Hence the length of arc
outside the disk is 2R cos-1( (r2-x2-R2)/2xR). This applies for r-R <= x <= r. For smaller x the small circle
lies entirely in the disk. Thus we have to evaluate 1/R2 ∫ 2π x 2R cos-1( (r2-x2-R2)/2xR) dx. Put x = r - yR,
and we get 4πr ∫01 (1 - y R/r) cos-1( (2Rry - (y2+1)R2)/2R(r - yR) ) dy. As R tends to 0, (1 - y R/r) tends to 1
and (2Rry - (y2+1)R2)/2R(r - yR) tends to y, so we get 4πr ∫01 cos-1y dy = 4πr (y cos-1y - √(1-y2) )|01 = 4πr.
© John Scholes
[email protected]
11 Mar 2002
8th Putnam 1948
Problem A5
Let α1, α2, ... , αn be the nth roots of unity. Find ∏i<j (αi - αj)2.
Solution
Let αk = exp(2pi(k-1)/n). So α1 = 1, and the other αk are roots of xn-1 + xn-2 + ... + x + 1 = 0. Hence (1 - α2),
... , (1 - αn) are roots of (1 - x)n-1 + ... + (1 - x) + 1 = 0. The coefficient of x0 is n and the coefficient of xn-1 is
(-1)n-1. Hence ∏2n(1 - αk) = n.
Hence n/αk = (αk - α1)(αk - α2) ... (αk - αn), where the product excludes (αk - αk). Hence nn/(α2...αn) = ∏i not
equ j(αi - αj) = (-1)n(n-1)/2 ∏i<j(αi - αj)2. But (α2...αn) = (-1)n-1. Hence ∏i<j(αi - αj)2 = (-1)(n-1)(n-2)/2nn.
© John Scholes
[email protected]
11 Mar 2002
8th Putnam 1948
Problem A6
(1) On each element ds of a closed plane curve there is a force 1/R ds, where R is the radius of curvature. The
force is towards the center of curvature at each point. Show that the curve is in equilibrium.
(2) Prove that x + 2/3 x3 + 2·4/3·5 x5 + ... + 2·4· ... .2n/(3·5. ... .2n+1) x2n+1 + ... = (1 - x2)-1/2 sin-1x.
Solution
(1)
(2) Let f(x) = x + 2/3 x3 + 8/15 x5 + ... + 1/2 n! n!/2n+1! (2x)2n+1 + ... .
Hence x f(x) = x2 + 2/3 x4 + 8/15 x6 + ... + 1/4 n! n!/2n+1! (2x)2n+2 + ... .
Also f '(x) = 1 + 2x2 + 8/3 x4 + ... + n! n!/2n! (2x)2n + ... .
and (1 - x2) f '(x) = 1 + x2 + 2/3 x4 + ... + n-1! n-1!/2n! n/2 (2x)2n + ...
= 1 + x2 + 2/3 x4 + ... + 1/4 n-1!n-1!/2n-1! (2x)2n + ... .
Hence the derivative of √(1 - x2) f(x) is 1/√(1 - x2) (- x f(x) + (1 - x2) f '(x) ) = 1/√(1 - x2). So √(1 - x2) f(x) =
const + sin-1(x). Putting x = 0, we find that the constant is 0.
© John Scholes
[email protected]
11 Mar 2002
8th Putnam 1948
Problem B1
p(x) is a cubic polynomial with roots α, β, γ and p'(x) divides p(2x). Find the ratios α : β : γ.
Solution
p(x) = (x - α)(x - β)(x - γ). Hence p'(x) = 3x2 - 2(α+β+γ)x + βγ+γα+aβ. This must have roots α/2, β/2. Hence
2(α+β+γ)/3 = (α+β)/2 and αβ/4 = (βγ+γα+αβ)/3. This leads to α = (-2 + 2i/√3)γ, β = (-2 - 2i/√3)γ.
© John Scholes
[email protected]
11 Mar 2002
8th Putnam 1948
Problem B3
Show that [√n + √(n + 1)] = [√(4n + 2)] for positive integers n.
Solution
We have ( √n + √(n + 1) )2 = 2n + 1 + 2 √(n2 + n). Now n2 < n2 + n < n2 + n + 1/4 = (n + 1/2)2, so √(4n+1) <
√n + √(n+1) < √(4n+2). Hence [ √(4n+1) ] ≤ [ √n + √(n+1) ] ≤ [ √(4n+2) ]. But a square must be congruent to
0 or 1 mod 4, so [ √(4n+2) ] = [ √(4n+1) ]. Hence result.
√ John Scholes
[email protected]
11 Mar 2002
8th Putnam 1948
Problem B5
Find the area of the region { (x, y) : |x + yt + t2| ≤ 1 for all t ∈ [0, 1] }.
Solution
We cannot have |x| > 1, for then the inequality is not satisfied at t = 0. We cannot have y > 1 for then yt + t2 >
2 at t = 1 and hence x + yt + t2 > 1. Similarly, we cannot have y < -3 for then yt + t2 < -2 at t = 1 and hence x
+ yt + t2 < -1.
We can write x + yt + t2 = x - y2/4 + (t + y/2)2. So for -3 ≤ y ≤ -2, the maximum of (t + y/2)2 occurs at t = 0
with value y2/4, and its minimum is (1 + y/2)2 at t = 1. Hence the minimum of x + yt + t2 is 1 + x + y and the
maximum is x. So the region is bounded by x ≤ 1 and x + y ≥ -2.
For -2 ≤ y ≤ -1, the maximum of (t + y/2)2 is y2/4 at t = 0, but the minimum is 0 at t = -y/2. Hence the
maximum of x + yt + t2 is x and the minimum is x - y2/4. So the region is bounded by x ≤ 1 and x - y2/4 ≥ -1.
For -1 ≤ y ≤ 0, the maximum of (t + y/2)2 is (1 + y/2)2 and the minimum is 0 at t = -y/2. Hence the maximum
of x + yt + t2 is 1 + x + y and the minimum is x - y2/4. So the region is bounded by x + y ≤ 0 and x - y2/4 ≥
-1.
For 0 ≤ y ≤ 1, the maximum of (t + y/2)2 is (1 + y/2)2 and the minimum is y2/4. Hence the maximum of x +
yt + t2 is 1 + x + y and the minimum is x. So the region is bounded by x + y ≤ 0 and x ≥ -1.
Thus the region is the rhombus with vertices at (1, -1), (1, -3), (-1, -1), (-1, 1), except that the vertex (-1, -1) is
cut off by the curve x = y2/4 - 1, which forms the boundary between (0, -2) and (-1, 0). The rhombus has area
4 (base 2 and height 2). The area between the parabola and the lines x = -1 and y = -2 is ∫02 y2/4 dy = 8/12.
The area between the line y = -2 -x and the lines x = -1 and y = -2 is 1/2, so the area cut off the rhombus by
the parabola is 2/3 - 1/2 = 1/6. Hence the area required is 4 - 1/6 = 3 5/6.
© John Scholes
[email protected]
11 Mar 2002
8th Putnam 1948
Problem B6
(1) Take the origin O of the complex plane to be the vertex of a cube, so that OA, OB, OC are edges of the
cube. Let the feet of the perpendiculars from A, B, C to the complex plane be the complex numbers u, v, w.
Show that u2 + v2 + w2 = 0.
(2) Let (aij) be an n x n matrix. Suppose that for each i, 2 |aii| > ∑1n |aij|. By considering the corresponding
system of linear equations or otherwise, show that det aij ≠ 0.
Solution
(1)
(2) If the det is non-zero, then we can find xinot all zero so that ∑ xi aij = 0 for each j. Take k so that |xk| ≥ |xi|
for all i. Then |∑i not k xiaik| ≤ ∑i not k |xi aik| ≤ |xk| ∑i not k |aik| < |xk| |akk|, so we cannot have ∑ xiaik = 0.
Contradiction.
© John Scholes
[email protected]
11 Mar 2002