Preliminary Exam 2016

Solutions to Morning Exam

Part I.
Solve four of the following five problems.
Problem 1. Find the volume of the “ice cream cone” defined by the inequalities
x2 + y 2 + z 2 6 1 and x2 + y 2 6 z 2 /3 for z > 0.
√
Solution: Since tan−1 3 = π/3 = π/2 − π/6, the volume is
Z 2π Z π/6 Z 1 √
ρ2 sin ϕ dρ dϕ dθ = (2π)(1/3)(1 − 3/2),
0 0 0
√
which is (2 − 3)π/3.
One can also do this problem in cylindrical coordinates: Since the intersection
of the surfaces r2 = z 2 /3 and r2 + z 2 = 1 projects to the circle r2 = 1/4, we get
Z 2π Z 1/2 Z √1−r2 Z 1/2 p √
√ r dz dr dθ = (2π) (r 1 − r2 − 3r2 ) dr,
0 0 3r 0

which is
1 r3 1/2
(2π)(− (1 − r2 )3/2 − √ ) 0
.
3 3
After some simplifications, we get the same answer as before.
Problem 2. Determine the radius of convergence and interval of convergence
2
of the power series n>1 (1 + 1/n)n xn .
P

Solution: Since limn→∞ (1+1/n)n |x| = e|x|, we see that the radius of convergence
is 1/e. Thus we must check convergence at the endpoints 1/e and −1/e. Now
2 1 1 1
log(1 + 1/n)n = n2 ( − 2 + 4 − . . . ),
n 2n 3n
which is n − 1/2 + O(1/n2 ). Thus
2
log((1 + 1/n)n (1/e)n ) = −1/2 + O(1/n2 ),
2
and consequently limn→∞ (1 + 1/n)n (1/e)n = e−1/2 . In particular, at 1/e the limit
of the nth term of the series is nonzero, and at −1/e the limit does not even exist.
So the interval of convergence is (−1/e, 1/e).
Problem 3. Prove that cos x0 = x0 for a unique x0 ∈ [0, 1], and show in addition
that π/6 < x0 < π/4.
Solution: Consider the function f (x) = cos x − x. We have f (0) = 1 > 0, and
since cos x is strictly decreasing on [0, π/2] we also have
f (1) = cos(1) − 1 < cos(π/2) − 1 = −1 < 0.
So the existence of x0 follows from the Intermediate Value Theorem (or the fact
that the continuous image of a connected set is connected). The uniqueness follows
from the fact that f 0 (x) = − sin x − 1 is strictly negative on [0, 1], whence f is
strictly decreasing on [0, 1]. [Remark: One could get the existence and uniqueness
simultaneously by appealing to the Contraction Fixed Point Theorem.]
1
2

For the second assertion, observe that

√
√ 3 3−π
f (π/6) = 3/2 − π/6 = >0
6
while
√
√ 2 2−π
f (π/4) = 2/2 − π/4 = < 0.
4
Since f is strictly decreasing on [0, 1] it follows that x0 ∈ (π/6, π/4).
Problem 4. Using standard techniques of integration, find antiderivatives on
some open interval where the integrand is defined and continuous:
(a) tan(cos2 x) sin(2x) dx.
R

Solution: Let u = cos2 x, so that du = −2 sin x cos xdx = − sin(2x)dx. We get

Z Z
tan(cos x) sin(2x) dx = − tan u du = log | cos u| + C = log | cos(cos2 x)| + C
2

R
(b) cos(log x) dx. (Here “log” is understood to be “ln.”)
Solution: Let u = cos(log x) and v = x, so that du = −x sin(log x)dx and
dv = dx. We get
Z Z
cos(log x) dx = x cos(log x) + sin(log x) dx.

A second integration by parts with u = sin(log x) and v = x then gives

Z
cos(log x) dx = x(cos log x + sin log x)/2 + C.

Problem 5. Find all solutions to the differential equation y 00 − y 0 − 6y = cos t

that are bounded on [0, ∞) and satisfy the condition y(0) = 0.
Solution: Since x2 − x − 6 = (x − 3)(x + 2), the general solution to the homoge-
neous equation is y(t) = ae3t + be−2t . Substituting y(t) = c cos t + d sin t into the
inhomogeneous equation, one finds that c = −7/50 and d = −1/50, so the general
solution to the inhomogeneous equation is
y(t) = ae3t + be−2t + (−7/50) cos t + (−1/50) sin t.
The boundedness on [0, ∞) holds if and only if a = 0, and given that a = 0, the
condition y(0) = 0 holds if and only if b = 7/50. So the above y(t) with a = 0,
b = 7/50, c = −7/50, and d = −1/50 is the unique function satisfying the stated
conditions.

Part II.
Solve three of the following six problems.
Problem 6. Let F(x, y, z) = (2x + 3y)i + (3x + 2y)j + zk.
(a) Compute ∇ × F.
Solution: Using the definition of ∇ × F as a symbolic determinant, we compute
i j k
 
 ∂ ∂ ∂ 
∇ × F = det  ,
∂x ∂y ∂z
2x + 3y 3x + 2y z
which is 0.
 3

t t
(b) Let C be the curve given parametrically R by r(t) = e cos ti + e sin tj for
0 6 t 6 2π. Find the value of the line integral C F · ds.
Solution: Since ∇ × F is zero and R3 is simply connected, F has a potential
function, which is easily computed to be
ϕ(x, y, z) = x2 + 3xy + y 2 + z 2 /2 + C.
Note also that r(0) = i and r(2π) = e2π i. Hence C F · ds = ϕ(e2π , 0, 0) − ϕ(1, 0, 0),
R

which is e4π − 1.
Problem 7. Fix an element c ∈ R, and define a function f : R → R by
f (x) = |x − c|. Show that f is uniformly continuous.
Solution: By the Triangle Inequality, ||x − c| − |y − c|| 6 |x − y|, or in other
words, |f (x) − f (y)| 6 |x − y|. So given ε > 0 let δ = ε; then |f (x) − f (y)| < ε
whenever |x − y| < δ.
Problem 8. The formula f (x, y, z) = (x + y 2 + z 2 , x2 + y + z 2 , x2 + y 2 + z)
defines a function f : R3 → R3 .
(a) Explain why there are open neighborhoods U and V = f (U ) of (0, 0, 0) ∈ R3
and a C 1 function g : V → U such that g(f (x, y, z)) = (x, y, z) for (x, y, z) ∈ U and
f (g(x, y, z)) = (x, y, z) for (x, y, z) ∈ V .
Solution: The Jacobian matrix of f at (0, 0, 0) is
 
1 2y 2z
2x 1 2z  = I,
(0,0,0)
2x 2y 1
the 3 × 3 identity matrix. In particular, this matrix is invertible, so by the Inverse
Function Theorem, there exist U , V , and g as above.
(b) Now let h(x, y, z) = (x + ey + ez − 2, ex + y + ez − 2, ex + ey + z − 2). Show
that if f is replaced by h then no such U , V , and g exist.
Solution: In this case, the Jacobian matrix of h at (0, 0, 0) is
1 ey ez
   
1 1 1
ex 1 ez  = 1 1 1 ,
(0,0,0)
ex ey 1 1 1 1
a matrix of row rank 1, not 3, and therefore not invertible. Denote this matrix by
A. If U , V , and g as above exist, and if B is the Jacobian matrix of g at (0, 0, 0),
then AB = I and hence det(A) det(B) = 1, a contradiction since det(A) = 0.
Problem 9. Let fn (x) = nxe−nx . Show that the sequence {fn } is pointwise
convergent on [0, 1] but not uniformly convergent.
Solution: If x = 0 then fn (x) = 0 for all n; otherwise
nx y
lim = lim y = 0
n→∞ enx y→∞ e

by L’Hôpital’s Rule. Thus {fn } converges pointwise to the zero function on [0, 1].
But the convergence is not uniform, for take ε < 1/e, let N > 1 be arbitrary, and
observe that if x = 1/(N + 1) and n = N + 1 then |fn (x) − 0| = 1/e and thus
|fn (x) − 0| > ε.
Problem 10. Let V be the real vector space of C ∞ functions f : R → R with
compact support (in other words, f vanishes outside some closed bounded interval).
4

Define an operator T : V → V by T (f ) = f 00 . Show that T is self-adjoint relative

to the L2 inner product on V . In other words, letting
Z ∞
hf, gi = f (x)g(x) dx
−∞

for f, g ∈ V , show that hT (f ), gi = hf, T (g)i.

Solution: Using integration by parts, we have
Z ∞ Z ∞
f (x)g 0 (x) dx = − f 0 (x)g(x) dx,
−∞ −∞

because both f and g vanish outside a closed bounded interval. Replacing g by g 0 we

obtain an expression on the right-hand which is symmetric in f and g. Therefore so
is the expression on the left-hand side, which is hf, T (g)i. In other words, hf, T (g)i
equals hg, T (f )i and hence hT (f ), gi. [Aternatively, just do integration by parts
again.]
Problem 11. Find the surface area of the torus described parametrically by
cos ϕ cos ϕ sin ϕ
r(θ, ϕ) = cos θ(1 + )i + sin θ(1 + )j + k (0 6 θ, ϕ 6 2π).
2 2 2
Solution: The surface area is by definition
Z 2π Z 2π
A= |rθ × rϕ | dθ dϕ
0 0
Now
 
i j k
− sin θ(1 + cos ϕ cos ϕ
rθ × rϕ = det  ) cos θ(1 + ) 0  ,
2 2
− sin ϕ − sin ϕ cos ϕ
 
cos θ( ) sin θ( )
2 2 2
cos ϕ cos ϕ cos ϕ cos ϕ cos ϕ sin ϕ
which is (cos θ)(1 + ) i + (sin θ)(1 + ) j + (1 + ) k. So
2 2 2 2 2 2
1 cos ϕ
|rθ × rϕ | = ( + ),
2 4
and consequently A = 2π 2 . [Alternatively, one could use the theorem of Pappus
that the surface area of a surface of revolution is 2πrL, where 2πr is the distance
traveled by the centroid of the curve being rotated and L is its length.]

Part III.
Solve one of the following three problems.
Problem 12. Let f be a C 2n function in some neighborhood of a point a ∈ R,
and suppose that f (k) (a) = 0 for 1 6 k 6 2n − 1. Show that if f (2n) (a) > 0 then f
has a local minimum at a.
Solution: By Taylor’s theorem, f is represented in some neighborhood of a by
the relevant Taylor polynomial of order 2n − 1 plus the remainder term:
(x − a)2n
f (x) = f (a) + f (2n) (c) ,
(2n)!
where c is strictly between a and x. Furthermore, since f (2n) is continuous in some
neighborhood of a and f (2n) (a) > 0, we see that f (2n) (c) > 0 for x near a. Of
 5

course also (x − a)2n > 0 for x 6= a. So for x near a but not equal to a it follows
that f (x) > f (a).
Problem 13. Let X and Y be metric spaces with respective metrics dX (∗, ∗)
and dY (∗, ∗), let x0 be a point of X, and let f : X → Y be a function.
(a) Consider the following definitions:
(A) f is continuous at x0 if for every ε > 0 there exists δ > 0 such that if x ∈ X
and dX (x, x0 ) < δ then dY (f (x), f (x0 )) < ε.
(B) f is continuous at x0 if for every sequence {xn }n>1 in X which converges
to x0 the sequence {f (xn )}n>1 converges to f (x0 ).
Show that these definitions are equivalent.
Solution: Assume that f is continuous at x0 according to definition (A), and
suppose that {xn }n>1 is a sequence in X which converges to x0 . We must show
that {f (xn )}n>1 converges to f (x0 ). Let ε > 0 be given. Choose δ > 0 as in (A),
and then choose N > 1 so that if n > N then dX (xn , x0 ) < δ. Since (A) is in force,
it follows that dY (f (xn ), f (x0 )) < , as desired.
Now assume that f is continuous at x0 according to definition (B), and let ε > 0
be given. Suppose that there does not exist δ > 0 with the property asserted in
(A). Then for every positive integer n we cannot take δ = 1/n, so there exists
xn ∈ X such that dX (xn , x0 ) < 1/n but dY (f (xn ), f (x0 )) > ε. Then {xn }n>1 is a
sequence in X which converges to x0 but {f (xn )}n>1 does not converge to f (x0 ),
contradicting (B). So the required δ > 0 does exist and (A) follows.
(b) Let I = [0, 2π) ⊂ R and T = {(x, y) ∈ R2 : x2 + y 2 = 1}, and consider
I and T as metric spaces by restricting the standard Euclidean metrics on R and
R2 respectively. Define g : I → T by g(x) = (cos x, sin x), and put f = g −1 . Is f
continuous at (1, 0) ∈ T ? Justify your answer using (B).
Solution: To see that f is not continuous at (1, 0), consider the sequence {xn }n>1
in T given by xn = (cos(1/n), sin(−1/n)), which converges to the point x0 = (1, 0).
However f (xn ) = 2π − 1/n, whence {f (xn )}n>1 does not converge to the point
f (x0 ) = 0 (or in fact to any point of I).
2
Problem 14. Define a real-valued function f on R by setting f (x) = e−1/x for
x 6= 0 and f (0) = 0.
2
(a) Show by induction on n that f (n) (x) = e−1/x Pn (1/x) for x 6= 0, where Pn
is a polynomial.
Solution: In the base case n = 0 we take P0 (x) = 1. Now suppose that for some
2
n > 0 we have f (n) (x) = e−1/x Pn (1/x) for x 6= 0. Then
2 2
f (n+1) (x) = e−1/x (2/x3 )Pn (1/x) + e−1/x Pn0 (1/x)(−1/x2 ).
2
Therefore f (n+1) (x) = e−1/x Pn+1 (1/x) with Pn+1 (y) = 2y 3 Pn (y) − y 2 Pn0 (y).
(b) Deduce that f is a C ∞ function on R and that f (n) (0) = 0 for all n.
Solution: The fact that f is a C ∞ function for x 6= 0 is verifiable by inspection.
We show by induction on n that f (n) exists and is continuous at 0 and that f (n) (0) =
0. For n = 0 the continuity is obvious, and f (0) = 0 by definition. Now suppose
that for some n > 0 we know that f (n) exists and is continuous at 0 and that
f (n) (0) = 0. To verify that f (n+1) (0) exists, we compute the limit of the relevant
6

difference quotient. By (a), we have

2
f (n) (h) − f (n) (0) e−1/h Pn (1/h)
lim = lim ,
h→0 h h→0 h
2
and the right-hand side is limy→∞ e−y Pn (y)y provided this limit coincides with
2
limy→−∞ e−y Pn (y)y. It does, because both limits are 0. Thus f (n+1) (0) exists and
equals 0, and then f (n+1) is continuous at 0 by (a).