Scribd Logo
Upload

Welcome to Scribd!

  • 15 views

    G13 1942 PUTNAM Web Solution

    Uploaded by

    mokonoani
    The document contains solutions to problems from the 1942 Putnam exam. Problem B5 asks about the function f(x) = x/(1 + x^6sin^2x) and solving the integral from 0 to infinity of f(x). The solution sketches the curve and shows that the integral converges by bounding the function near the spikes at x=nπ and elsewhere. Choosing k=2 1/4 ensures the whole integral is bounded.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    G13 1942 PUTNAM Web Solution

    Uploaded by

    mokonoani
    15 views10 pages
    The document contains solutions to problems from the 1942 Putnam exam. Problem B5 asks about the function f(x) = x/(1 + x^6sin^2x) and solving the integral from 0 to infinity of f(x). The solution sketches the curve and shows that the integral converges by bounding the function near the spikes at x=nπ and elsewhere. Choosing k=2 1/4 ensures the whole integral is bounded.

    Original Title

    G13-1942-PUTNAM-Web-Solution

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    The document contains solutions to problems from the 1942 Putnam exam. Problem B5 asks about the function f(x) = x/(1 + x^6sin^2x) and solving the integral from 0 to infinity of f(x). The solution sketches the curve and shows that the integral converges by bounding the function near the spikes at x=nπ and elsewhere. Choosing k=2 1/4 ensures the whole integral is bounded.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    15 views10 pages

    G13 1942 PUTNAM Web Solution

    Uploaded by

    mokonoani
    The document contains solutions to problems from the 1942 Putnam exam. Problem B5 asks about the function f(x) = x/(1 + x^6sin^2x) and solving the integral from 0 to infinity of f(x). The solution sketches the curve and shows that the integral converges by bounding the function near the spikes at x=nπ and elsewhere. Choosing k=2 1/4 ensures the whole integral is bounded.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 10

    5th Putnam 1942

    Problem A1

    ABCD is a square side 2a with vertices in that order. It rotates in the first quadrant with A remaining on the
    positive x-axis and B on the positive y-axis. Find the locus of its center.

    Solution

    Answer: the segment (a, a) to (a√2, a√2).

    Easy.

    Let AB make an angle θ with the x-axis. Then we find that the coordinates of the center to be x = y = a cos θ
    + a sin θ. But a cos θ + a sin θ = a √2 sin(θ + π/4).

    5th Putnam 1942

    © John Scholes
    [email protected]
    15 Sep 1999
    5th Putnam 1942

    Problem A2

    a and b are unequal reals. What is the remainder when the polynomial p(x) is divided (x - a)2(x - b).

    Solution

    Suppose the remainder is cx2 + dx + e. We have p(a) = ca2 + da + e, p(b) = cb2 + db + e. Also, differentiating,
    we get p'(a) = 2ca + d. Solving, c = p'(a)/(a - b) - p(a)/(a - b)2 + p(b)/(a - b)2, d = (2a p(a) - 2a p(b) - (a2 -
    b2)p'(a) )/(a - b)2, e = p(a) - a2(p(a) - p(b))/(a - b)2 + ab p'(a) /(a - b).

    5th Putnam 1942

    © John Scholes
    [email protected]
    5 Mar 2002
    5th Putnam 1942

    Problem A3

    Does ∑n≥0 n! kn/(n + 1)n converge or diverge for k = 19/7?

    Solution

    The nth term divided by the n-1th term is k n nn-1/(n+1)n = k/(1 + 1/n)n which tends to k/e. But k/e < 1, so
    the series converges by the ratio test.

    5th Putnam 1942

    © John Scholes
    [email protected]
    5 Mar 2002
    5th Putnam 1942

    Problem A4

    Let C be the family of conics (2y + x)2 = a(y + x). Find C', the family of conics which are orthogonal to C. At
    what angle do the curves of the two families meet at the origin?

    Solution

    For most points P in the plane we can find a unique conic in the family passing through the point. Thus we
    should be able to find the gradient of members of the family at (x, y) in a formula which is independent of a.
    We then use this to get a formula for the gradient of the orthogonal family and solve the resulting first-order
    differential equation to get the orthogonal family.

    Thus we have 8y y' + 4x y' + 4y + 2x = ay' + a = (y' + 1)(2y + x)2/(y + x). So y'(2y+x)( 4(x+y) - (2y+x) ) =
    (2y+x)2 - (x+2y)(x+y), or y'(2y+x)(2y+3x) = -2x(2y+x), so y' = -x/(2y+3x). Hence the orthogonal family
    satisfies y' = (2y+3x)/x. So y'/x2 - 2y/x3 = 3/x2. Integrating y = bx2 - 3x. These are all parabolas.

    All members of both families pass through the origin. Changing coordinates to X = x + 2y, Y = y - 2x, the
    equation of a member of the first family becomes X2 = a(3X-Y)/5 or Y = - 5/a (X - 3a/10)2 + 9a/20. This has
    gradient 3 (in the new system) at the origin. In the old system the tangent is y = -x. The orthogonal set
    obviously has gradient -3 at the origin. If the angle between them is k, then tan k = (-1 +3)/(1+3) = 1/2. So k
    = tan-11/2.

    5th Putnam 1942

    © John Scholes
    [email protected]
    5 Mar 2002
    5th Putnam 1942

    Problem A5

    C is a circle radius a whose center lies a distance b from the coplanar line L. C is rotated through π about L to
    form a solid whose center of gravity lies on its surface. Find b/a.

    Answer

    (π + √(π2+2π-4))/(2π-4) = about 2.9028

    Solution

    The solid is half a torus. We can divide it into a large number of thin disks. Each disk has variable thickness,
    with thickness proportional to the distance from L. So we must integrate to find the distance of the centroid
    of the disk from L. Take the density to be kd, where d is the distance from L.

    Take x to be distance along the line perpendicular to x, and θ to be the angle between the radius vector and
    the x-axis. We have x = a cos θ, so dx = - a sin θ dθ. The mass is ∫0π 2a sin θ (a sin θ dθ) k(b + a cos θ) =
    2a2bk ∫0π sin2θ dθ + 2a3k ∫0π sin2θ cos θ dθ = a2bkπ + 0. So the mass times the centroid distance is ∫0π 2a2k
    sin2θ (a cos θ + b)2 dθ = 2a4k ∫0π sin2θ cos2θ dθ + 4a3bk ∫0π sin2θ cos θ dθ + 2a2b2k ∫0π sin2θ dθ = ½a4k ∫0π
    sin22θ dθ + 0 + a2b2kπ = ka2π(a2/4 + b2). So the centroid distance is b + a2/4b. Thus we can regard the mass
    as uniformly spread over a semicircle radius b + a2/4b.

    We need another integration to find the distance of the mass of a semicircle radius r from its center. It is (1/
    πr) ∫0π r2 sin θ dθ = 2r/π. Thus the cm of the half-torus is a distance (2/π)(b + a2/4b) from L. We want it to be
    a distance b-a from L so that it lies on the surface. Thus (2/π)(b + a2/4b) = b - a, so (2π-4)b2 - 2πab - a2 = 0.
    Hence b/a = (π + √(π2+2π-4))/(2π-4) = about 2.9028.

    5th Putnam 1942

    © John Scholes
    [email protected]
    19 January 2004
    Last corrected/updated 19 Jan 04
    5th Putnam 1942

    Problem B1

    S is a solid square side 2a. It lies in the quadrant x ≥ 0, y ≥ 0, and it is free to move around provided a vertex
    remains on the x-axis and an adjacent vertex on the y-axis. P is a point of S. Show that the locus of P is part
    of a conic. For what P does the locus degenerate?

    Solution

    Let A be the vertex that moves along the x-axis and B the vertex that moves along the y-axis. Suppose that
    when AB is horizontal P has coordinates b, c. In the general configuration let be the angle BAO be θ. Then P
    has coordinates x = (2a - b) cos θ + c sin θ, y = b sin θ + c cos θ. Hence cx - (2a - b)y = (b2 + c2 - 2ab) sin θ,
    bx - cy = (2ab - b2 - c2) cos θ. Squaring and adding we eliminate θ to get: (b2 + c2) x2 - 4ac xy + (4a2 + b2 +
    c2 - 4ab) y2 = (b2 + c2 - 2ab)2, which is the equation of a conic. So the locus of P must form part of this
    conic.

    The conic degenerates if b2 + c2 = 2ab. In this case, the equation becomes 2ab x2 - 4ac xy + (4a2 - 2ab)y2 =
    0, or bx2 - 2c xy + (2a - b)y2 = 0, or b2x2 - 2bc xy + c2 y2 = 0, or bx = cy. So in this case the locus lies on a
    straight line. We may write the condition b2 + c2 = 2ab as (a - b)2 + c2 = a2, which shows that such P lie on
    the semicircle diameter AB.

    5th Putnam 1942

    © John Scholes
    [email protected]
    5 Mar 2002
    5th Putnam 1942

    Problem B2

    Let Pa be the parabola y = a3x2/3 + a2x/2 - 2a. Find the locus of the vertices of Pa, and the envelope of Pa.
    Sketch the envelope and two Pa.

    Solution

    We can write the equation of Pa as (y + 35a/16) = (a3/3)(x + 3/4a)2, so the vertex is x = -3/4a, y = -35a/16.
    The locus of the vertex is xy = 105/64.

    The graph shows P3, P2, P1, P1/2, P1/3 and the two hyperbolae yx = -7/6, yx = 10/3. It shows that for positive
    a, the parabolas touch the lower branches of the hyperbolae. For negative a they touch the upper branches.

    That is not hard to verify. We claim that Pa and xy = 10/3 touch at x = -2/a, y = -5a/3. The point obviously
    lies on xy = 10/3. We have (a3/3)(x + 3/4a)2 = (1/3)(-2 + 3/4)2 a = 25a/48 = (-5a/3 + 35a/16), so the point also
    lies on Pa. The gradient of xy = 10/3 at the point is -10/(3x2) = -5a2/6. The gradient of Pa at the point is
    2a3x/3 + a2/2 = -5a2/6.

    Similarly, we claim that Pa and xy = -7/6 touch at x = 1/a, y = -7a/6. The point obviously lies on xy = -7/6.
    We have (a3/3)(x + 3/4a)2 = (a/3)(1 + 3/4)2 = 49a/48 = (-7a/6 + 35a/16), so the point also lies on Pa. The
    gradient of xy = -7/6 at the point is 7/(6x2) = 7a2/6. The gradient of Pa at the point is 2a3x/3 + a2/2 = a2(2/3 +
    1/2) = 7a2/6.
    It is less clear how you get the hyperbolas. One standard approach is to look for the singular points of the
    mapping f(a,t) = (t, a3t2/3 + a2t/2 - 2a). The matrix for the derivative is:
    1 0
    2a3t/3+a2/2 a2t2+at-2

    which has zero determinant when at = 1 or -2, so xy = 7/6 or -10/3.

    Thanks to Dave Rusin

    5th Putnam 1942

    © John Scholes
    [email protected]
    20 January 2004
    Last corrected/updated 20 Jan 04
    5th Putnam 1942

    Problem B4

    A particle moves in a circle through the origin under the influence of a force a/rk towards the origin (where r
    is its distance from the origin). Find k.

    Solution

    The equations of motion are r (θ ')2 - r'' = a/rk, r2θ ' = A (conservation of angular momentum).

    If the particle moves in a circle as described, then we can write its orbit as r = B cos θ. Differentiating, r' = -
    B θ ' sin θ = -AB/r2 sin θ. Differentiating again, r'' = -AB cos θ A/r2 + 2AB/r3 sin θ r' = -A2/r3 - 2A2B2/r5
    sin2θ = -A2/r3 - 2A2B2/r5 (1 - r2B2) = A2/r3 - 2A2B2/r5. So substituting back in the equation of motion we
    get: 2A2B2/r5 = a/rk. Hence k = 5.

    Note that this is unphysical, since we require infinite velocity as we reach the origin.

    5th Putnam 1942

    © John Scholes
    [email protected]
    5 Mar 2002
    5th Putnam 1942

    Problem B5

    Let f(x) = x/(1 + x6sin2x). Sketch the curve y = f(x) and show that ∫0∞ f(x) dx exists.

    Solution

    Obviously f(x) is positive for positive x. But it has an infinite number of spikes at x = nπ. The spike at nπ is
    height nπ, so we have to show that the integral is bounded above.

    We have sin x > ½x near x = 0 (certainly for x < π/3). So |sin x| > 1/(nπ)k except possibly for |x| < 2/(nπ)k.
    Let In be the interval centered on nπ width 4/(nπ)k. For x ∈ In we have f(x) < 2nπ, so the integral of f(x) over
    the interval is less than 8/(nπ)k-1. The total integral over all such intervals is bounded provided that k > 2.
    Outside such intervals, x6 sin2x > ½ x6/x2k, so f(x) < 2/x2k-5. Hence the interval of f(x) over 0 to ∞ excluding
    the intervals In is bounded provided k > 5/2. By taking k = 2 1/4, for example, we get that the whole integral
    is bounded.

    5th Putnam 1942

    © John Scholes
    [email protected]
    5 Mar 2002

    You might also like