You are on page 1of 10

MASTERS EXAMINATION IN MATHEMATICS

PURE MATH OPTION, Fall 2020

Full points can be obtained for correct answers to 8 questions. Each numbered question
(which may have several parts) is worth 20 points. All answers will be graded, but the score
for the examination will be the sum of the scores of your best 8 solutions.

Use separate answer sheets for each question. DO NOT PUT YOUR NAME ON
YOUR ANSWER SHEETS. When you have finished, insert all your answer sheets into
the envelope provided, then seal it.

Any student whose answers need clarification may be required to submit to an oral exam-
ination.

Algebra

A1. Let R be a commutative ring with 1.

(a) SupposeT that P1 ⊇ P2 ⊇ · · · ⊇ Pn . . . is a decreasing sequence of prime ideals in R.


Show that ∞ i=1 Pi is a prime ideal in R.

(b) Suppose that Q1 and Q2 are two prime ideals in R, such that Q1 * Q2 and Q2 * Q1 .
Show that Q1 ∩ Q2 is not a prime ideal.

Solution.

(a) It is easy to see that P is an ideal. We’ll show that P is a prime ideal. Suppose that
x, y ∈ R and x · y ∈ P . Assume that x is not in P . Then there is a positive integer n such
that x ∈/ Pn . Then x ∈ / Pm for m ≥ n and x · y ∈ Pm for m ≥ n. Since Pm is a prime ideal,
we conclude that y ∈ Pm for every m ≥ n. It also follows that y ∈ Pi for i ≤ n. We concldue
that u ∈ P .

Choose a ∈ Q1 − Q2 and b ∈ Q2 − Q1 . Then a · b is in Q1 ∩ Q2 . Howevery a ∈


/ Q and
b∈
/ Q. Thus Q is not a prime ideal in R.

A2. Let G be a finite abelian group of order pn , where p is a prime number and n is a
k
positive integer. For each positive integer k, let Hpk = {x ∈ G : xp = 1}.

(a) Show that Hpk is a subgroup of G.

(b) Show that G is a cyclic group if and only if Hp is a cyclic group of order p.

1
Solution.

(a) It is easy to see that if x, y are in Hpk , then x · y −1 is in Hpk . Hence Hpk is a subgroup
of G.

(b)Assue G =< x > is a cyclic group of order pn . If y = xa is in Hp . Then y P = X ap = 1. It


n−1
follows that pn divides pa, or equivalently, pn−1 divides a. We conclude that y ∈< xp >.
We conclude that Hp is a cyclic subgroup of order p.

By the fundamental theorem of finite abelian group, we can write

G = H1 × H2 × · · · × Hr ,

where each Hj is a cyclic group of order pnj . Furthermore n1 ≥ · · · nr ≥ 1. G is cyclic if and


only r = 1. Set (Hj )p = {y ∈ Hj |y p = 1} for 1 ≤ j ≤ r. Then Hp = (H1 )p × · × (Hr )p . It
follows that Hp is a cyclic group if and only r = 1 if and only G is a cyclic group.

A3. Let Sn , the symmetric group of n elements. Consider β = (1, 2, · · · , n) ∈ Sn .

(a) Suppose that σ ∈ Sn . Show that σβσ −1 = (σ(1), σ(2), · · · , σ(n)).

(b) Consider the conjugacy class of β in Sn . Conj(β) = {x ∈ Sn : x is a conjugate of β}.


Show that |Conj(β)| = (n − 1)!

(c) Consider the centralizer of β. Cent(β) = {τ ∈ Sn : τ βτ −1 = β)}. Show that Cent(β) is


equal to the cyclic group generated by β.

Solution.

(a) Observe that σβσ −1 (σ(i)) = σ(i + 1), if 1 ≤ i < n. Similarly σβσ −1 (σ(n)) = σ(1). So
σβσ −1 = (σ(1), σ(2), · · · σ(n)).

(b) From (a), we see α is a conjugate of β if and only if α is an n−cycle. We conclude


that |Conj(β)| = (n − 1)! .

(c) Observe that |Sn | = |Conj(β]||Cent(β)|. We conclude that |Cent(β)| = n. Since < β >,
which is a cyclic group of order n, is clearly in Cent(β). We conclude that Cent(β) =< β >.

Complex Analysis

C1. Determine each function f (z) that is analytic on the entire complex plane and whose
imaginary part is nonnegative at every z.
2
Solution. Note that |e−if (z) | = eImf (z) ≤ 1. Then Liouville’s theorem implies e−if (z) is
constant throughout the complex plane, which means f (z) is a constant function with a
nonnegative imaginary part.

C2. Let C be the positively oriented closed contour |z| = 3. Calculate the integral
Z  
2
ez + sin(z 3 ) cos(z 5 ) + z dz
C

Solution.
2
Since ez + sin(z 3 ) cos(z 5 ) is entire, its integral is zero, so it remains to compute
|z|2
Z Z Z
dz
zdz = dz = 9 = 9 · 2πi = 18πi.
C C z C z

C3. Let D = {z | |z| ≤ 1} denote the closed disc of radius 1 centered at the origin. Let
r0 < 1 and let f (z) be an analytic function on D such that |f (z)| ≤ r0 for all z ∈ D. Show
that there is a unique z0 ∈ D such that f (z0 ) = z0 .

Solution.

Solution. Since |f (z)| ≤ r0 < 1 = |z| for all z on the circle |z| = 1, Rouché’s theorem
implies that f (z) − z and z have the same number of zeros in D1 . Hence f (z) − z has only
one zero, which means f has a unique fixed point.

Number Theory

N1.
(1) Define a prime natural number and define an irreducible natural number.
(2) Determine the unique prime factorization of 72168327.
(3) Prove that there exist infinitely many primes of the form 4n + 3.
(4) Prove that there exist infinitely many primes of the form 6n + 5.

Solution.

(1) A natural number p is prime if p|ab implies p|a or p|b whenever a, b ∈ N. A natural
number n is irreducible if ab = n implies a = 1 or b = 1 whenever a, b ∈ N.
(2) The digits of 72168327 sum to a multiple of 9, so it is divisible by 9. Doing the
division we see that 72168327 = 9 ∗ 8018703. The sum of the digits of 8018703 again
are divisible by 9, and we get 8018703 = 9 ∗ 890967. The sum of the digits of 890967
are divisible by 3, and we get 890967 = 3 ∗ 296989.
The last digit of 296989 is not 5 or 0, so it’s not divisible by 5. We next try 7,
and doing the division twice we obtain 296989 = 72 ∗ 6061. The alternating sum of
the digits of 6061 is 11, a multiple of 11, so it is divisible by 11. The division yields
3
6061 = 11 ∗ 551. Trying successive prime numbers gives 551 = 19 ∗ 29, the product
of two primes.
Putting this all together yields 72168327 has the prime factorization 35 72 ∗ 11 ∗ 19 ∗
29.
(3) Suppose there were only finitely many primes of the form 4n + 3, we will arrive at
a contradiction. Then we could list these primes as p1 , ..., pn . We then examine
A = (p1 ....pn )2 + 2. Since each pk ≡ −1 (mod 4), each p2k ≡ 1 (mod 4), so A ≡ 3
(mod 4). We look at A’s prime factorization A = q1 ....qm . A is odd so each qi is
odd. If each qi ≡ 1 (mod 4) their product A would also be 1 (mod 4) which it isn’t.
Hence some qi ≡ 3 (mod 4) and thus qi = pj for some j. But then qi would divide
A − (p1 ....pn )2 = 2, contradicting that qi ≡ 3 (mod 4).
(4) We argue similarly to the last part. Suppose there were only finitely many primes of
the form 6n + 5, we will arrive at a contradiction. Then we could list these primes
as p1 , ..., pn . We then examine A = (p1 ....pn )2 + 4. Since each pk ≡ −1 (mod 6),
each p2k ≡ 1 (mod 6), so A ≡ −1 (mod 6). We look at A’s prime factorization
A = q1 ....qm . A is odd so each qi is odd. If each qi ≡ 1 (mod 6) their product
A would also be 1 (mod 6) which it isn’t. Hence some qi ≡ −1 (mod 6) and thus
qi = pj for some j. But then qi would divide A − (p1 ....pn )2 = 4, contradicting that
qi ≡ −1 (mod 6).

N2.
(1) Define the Möbius function µ.
(2) Prove the fundamental property of the Möbius function.
(3) State the Möbius inversion formula.
(4) Prove that X
φ(n) = cx2 + O(x log x),
n≤x
X µ(d)
1
where c := 2
.
d≥1
d2

Solution.

(1) The Möbius function µ is defined by µ(n) = 1 if n = 1, µ(n) = (−1)k if n is the


product of k distinct primes, and µ(n) = 0 if the square of any prime divides n.
(2) The fundamental property of the Möbius function is that µ(1) = 1 and for n > 1 one
has X
µ(d) = 0
d|n
This can be proved as follows. The case n = 1 is immediate. For n > 1, let
n = pa11 ....pakk be the prime factorization of n. In the sum d|n µ(d) the nonzero
P
terms come from d = 1 and the factors of n which are products of distinct prime
factors. Thus adding according to the number of distinct prime factors of d we have
X k
X X
µ(d) = µ(1) + µ(pj ) + µ(pi pj ) + ... + µ(p1 ....pk )
d|n j=1 pi 6=pj
4
     
k k k k
=1− + − .... + (−1)
1 2 k
= (1 − 1)k
=0
(3) The Möbius
P inversion formula is that if f (n), g(n) are arithmetic functions with
g(n) = d|n f (d), then one has
X n
f (n) = g(d)µ( )
d
d|n
P
(4) We start with the formula φ(d) = n. Möbius inversion then gives
d|n
X n
φ(n) = d µ( )
d
d|n

X n
= µ(d)
d
d|n

Thus we have X XX n
φ(n) = µ(d)
n≤x n≤x d|n
d
X
= µ(d)q
(q,d):qd≤x
X X
= µ(d) q
d≤x q≤ x
d

  2  
X 1 x x
= µ(d) +O
d≤x
2 d d
   X 
1 2 X µ(d) 1
= x + O x
2 d≤x
d2 d≤x
d
 
1 2 X µ(d)
== x 2
+ O(x log x)
2 d≤x
d
 
1 2 X µ(d)
= x + O(x) + O(x log x)
2 d≥1
d2
 
1 2 X µ(d)
= x + O(x log x)
2 d≥1
d2

N3.
(1) Define the proper equivalence between binary, integral, quadratic forms.
5
(2) Let f (x, y) = ax2 + bxy + cy 2 ∈ Z[x, y] be a quadratic form of discriminant D.
Assuming that f is positive definite and reduced, prove that
r
|D|
|b| ≤ a ≤
3
and
|D| |D|
c≤ ≤ .
3a 3
(3) Determine the class number h(−8).
(4) Let p ≥ 3 be a prime. Prove that there exist integers x, y such that p = x2 + 2y 2 if
and only if p ≡ 1, 3( mod 8).

Real Analysis

R1. Prove that n≥1 sinxnnx converges uniformly on (2, ∞) to a differentiable function f (x),
P
and give an expression for f 0 (x).

Solution.

We start by noting that, since | sin(y)| ≤ 1 for all y ∈ R, we have



sin(nx) 1 1
xn ≤ xn ≤ 2n for all n ∈ N and x > 2 .

Since 0 < 21 < 1, the geometric series


P 1
2n
converges, and so the original series converges
uniformly to a function f by the Weierstrass M-test.
Differentiating each term of series, we obtain:

n cos(nx)xn − sin(nx) · nxn−1 n x cos(nx) − sin(nx)
 
d sin(nx)
= = ,
dx xn x2n xn+1
and so, as above,
 
d sin(nx)
≤ 2nx < n for all n ≥ 1 and x > 2 .
dx x n xn+1 2n−1
P n P d  sin(nx) 
Since the series 2n−1 converges, it again follows that the series of derivatives dx xn
converges uniformly on (2, ∞) to a function g. Integrating this uniformly convergent series
leads to the original series, and hence g = f 0 and
∞ 
0
X n x cos(nx) − sin(nx)
f (x) = n+1
.
n=1
x
R2. Let f : [a, b] → R be a function such that f (x) = 0 except at finitely many points.
Show that f is Riemann integrable by showing that for Revery  > 0 there is a partition P of
b
[a, b] such that U (f, P) − L(f, P) < . Then show that a f (x) = 0.

Solution.
6
Let a = x0 < x1 < · · · < xn = b be the points where f is not equal to zero, together with
the endpoints a and b, if they are not already included.
Let  > 0. Since f is nonzero only a finite number of points, it follows that f is bounded
– let M > 0 such that |f (x)| ≤ M for all x ∈ [a, b]. Choose

δ= .
4(n + 1)M
and a partition P = {a < a + δ < x1 − δ < x1 + δ < · < xn−1 + δ < b − δ < b}. Note that
f ≡ 0 on each sub-interval of the form [xj + δ, xj+1 − δ]. It then follows from their definition
that

U (f, P ) , L(f, P ) ≤ (n + 1)M · 2δ
and hence

U (f, P ) − L(f, P ) ≤ 4(n + 1)δM <  .
By definition, this implies that f in Riemann integrable. In addition,
Z b
− < L(f, P ) ≤ f (x) dx ≤ U (f, P ) <  ,
a

and since  > 0 is arbitrary, this implies that


Z b
f (x) dx = 0 .
a

R3. Consider the system of equations


(
x2 − y 2 − u3 + v 2 + 4 = 0
2xy + y 2 − 2u2 + 3v 4 + 8 = 0
(i) Verify that (x0 , y0 , u0 , v0 ) = (2, −1, 2, 1) is a solution of the system.
(ii) Prove that there are neighborhoods U of (2, −1) and V of (2, 1), and a unique function
f : U → V such that (x, y, f (x, y)) is a solution of the system for all (x, y) ∈ U .

Solution.

(i) We plug (2, −1, 2, 1) into the equations:


22 − (−1)2 − 23 + 12 + 4 = 4 − 1 − 8 + 1 + 4 = 0
and
2 · 2 · (−1) + (−1)2 − 2 · 22 + 3 · 14 + 8 = −4 + 1 − 8 + 3 + 8 = 0 ,
as claimed.
(ii) Denoting the left-hand side of each equation by fj (x, y, u, v), j = 1, 2, we obtain that
 ∂f1 ∂f1   2

−3u 2v
det ∂f∂u ∂v
∂f1 = det 3 = −36u2 v 3 + 8uv
∂u
2
∂v
−4u 12v
At (x0 , y0 , u0 , v0 ) = (2, −1, 2, 1), this determinant is −36 · 4 + 16 6= 0, and hence
the conclusion of the problem holds by the Implicit Function Theorem.
7
Topology

T1. Let X be a topological space and A a subset. Show that if A is connected then Ā is
connected. Give an example to show the converse need not hold.

Solution.

Suppose X is a topological space and A is a subset, where A is connected in X. Suppose


Ā is the union of two disjoint closed sets in the subspace topology. That is, we suppose
Ā = (B ∩ Ā) ∪ (C ∩ Ā) where B and C are closed in X and the sets B ∩ Ā and C ∩ Ā are
disjoint. Then intersecting with A we have that A = (B ∩ A) ∪ (C ∩ A) with the sets B ∩ A
and C ∩ A being disjoint. The connectedness of A implies that either A ⊂ B with A ∩ C = ∅
or A ⊂ C with A ∩ B = ∅. Without loss of generality we may assume the former holds.

Since B is closed and A ⊂ B, we must have Ā ⊂ B as well. Thus in the disjoint union
Ā = (B ∩ Ā) ∪ (C ∩ Ā) we must have C ∩ Ā = ∅. Since B and C were arbitrary closed sets
in X with Ā = (B ∩ Ā) ∪ (C ∩ Ā) being a disjoint union, we conclude Ā is connected.

That the converse doesn’t necessarily hold can be seen by taking X = R and A = (−1, 0)∪
(0, 1).

T2. Let τ1 and τ2 be topologies on a set S with τ1 strictly finer than τ2 . Show that it is not
possible for both topologies to be compact and Hausdorff.

Solution.

Suppose both topologies were compact and Hausdorff; we will arrive at a contradiction.
Let C ⊂ S be such that C is closed in τ2 but not τ1 . Let i : (S, τ2 ) → (S, τ1 ) denote the
inclusion map. Since τ2 is finer than τ1 , i−1 (U ) is open for any open U in τ1 , so the map i is
continuous.

Since C is closed in the compact topology τ2 , C is also compact in τ2 . Since i is continuous


i(C) is compact. So C is compact in τ1 as well. But compact subsets of Hausdorff spaces
are closed, so C is closed in τ1 , contradicting the definition of C and we are done.

T3. Let f : R → R satisfy f (x + n) = f (x) + n for all x ∈ R and n ∈ Z. Prove that if f


is continuous then it is surjective. Give an example to show the hypothesis of continuity is
needed.

Solution.

The conditions given implies that for n ∈ N one has that f (−n) = f (0) − n and f (n) =
f (0) + n. The continuity of f and the connectedness of the interval [−n, n] imply that
f ([−n, n]) is connected, and therefore an interval containing [f (0) − n, f (0) + n]. So R =
∪∞n=1 [f (0) − n, f (0) + n] ⊂ f (R). We conclude that f (R) = R.

8
That continuity is needed here can be seen by taking f (x) = bxc.

Logic

L1. Let LO be the theory of infinite linear orders


(a) Show that LO is axiomatizable
(b) Show that LO is not finitely axiomatizable.

Solution.

Part a) The following are the axioms:


• transitivity: ∀x∀y∀z(x < y ∧ y < z → x < z);
• irreflexivity: ∀x¬(x < x);
• total ordering: ∀x∀y(x < y ∨ y < x ∨ x = y);
• λk := (∃x1 )(∃x2 )...(∃xk )Λi<j xi 6= xj , for k > 1.
Part b) Suppose for contradiction that the theory of infinite linear orders is finitely ax-
iomatizable. Then there is some sentence φ, such that a model A |= φ iff A is a model of
infinite linear order. Let Σ be the axiomatization defined in part (1). Then Σ |= φ. By
compactness it follows that there is some finite subset ∆ of Σ, such that ∆ |= φ. Let n be
large enough such that if λk ∈ ∆, then k < n. Let A be a finite model of a linear order with
size at least n. Then A |= ∆, and so A |= φ. Contradiction with the assumption that φ
axiomatizes infinite linear orders.

L2. Show that the theory of (Q, <) is complete.

Solution.

By Vaught’s test it is enough to show that the theory of (Q, <) is ℵ0 -categorical. So fix
two countable models A, B of T h(Q, <). We will define an isomorphism between them.
Enumerate the universe of A by A = {an | n < ω} and the universe of B by B = {bn |
n < ω}. By induction on n, define functions fn : An → Bn , such that An ⊂ A, Bn ⊂ B,
fn ⊂ fn+1 , and each fn is a partial isomorphism, as follows.
Set A0 = {a0 }, B0 = {b0 } and f0 (a0 ) = b0 .
Odd stages n = 2m + 1: Set Bn = Bn−1 ∪ {bm }. If Bn = Bn−1 , let fn = fn−1 . Otherwise
let a∗ ∈ A \ An−1 be such that for all a ∈ An−1 ,
a < a∗ iff fn−1 (a) < bm .
Set An = An−1 ∪ {a∗ }, fn (a∗ ) = bm , fn−1 ⊂ fn . We can always find such an element a∗ , since
A is a dense linear order.
Even stages n = 2m > 0: Set An = An−1 ∪ {am }. If An = An−1 , let fn = fn−1 . Otherwise
let b∗ ∈ B \ Bn−1 be such that for all a ∈ An−1 ,
a < am iff fn−1 (a) < b∗ .
Set Bn = Bn−1 ∪ {b∗ }, fn (am ) = b∗ , fn−1 ⊂ fn . We can always find such an element b∗ , since
B is a dense linear order.
9
S
Finally set f = n fn . At even stages we guaranteed that for each m, am ∈ dom(f ),
and at odd stages we guaranteed that for each m, bm ∈ range(f ). Also by construction we
arranged that for all a, a0 ∈ A, a < a0 iff f (a) < f (a0 ). So f an isomorphism between A and
B.

L3. Let N = (N, <, +, 0, 1).


(a) Show there is a proper elementary extension M of N .
(b) Show that the order <M is not a wellorder.
(c) Show the class of wellorders is not axiomatizable.

Solution.

Part a) L = {<, +, 0, 1} is the language of N . Let c be a new constant and L0 = L ∪ {c}.


For each k > 0, define the L0 sentence
λk = “ 1| + ·{z
· · + 1} < c”.
k

Let Γ = T h(N ) ∪ {λk | k > 0}. We claim that Γ is consistent: If ∆ ⊂ Γ is finite, let n be
such that λk ∈ ∆ implies that k < n. Then let N 0 be the expansion of N to L0 , where c is
interpreted as n. Then N 0 satisfies ∆. By compactness it follows that Γ is consistent. Let
M be a model of Γ. Then M is a proper elementary extension.
Part b) Suppose for contradiction that <M is a wellorder. Let M denote the universe
of M and let I = M \ N. I is nonempty. We claim that I has no <-minimal element.
Otherwise, suppose d = min(I). Let
φ = “∀x(x 6= 0 → ∃y(y < x ∧ ¬∃z(y < z ∧ z < x)))”
i.e. φ say that every nonzero element has an immediate predecessor. Since N |= φ, so does
M. Let a be the immediate predecessor of d.
But we also have that M |= “∀x(x < y → (x + 1 < y ∨ x + 1 = y))”. So a + 1 = d, and
so a ∈ I. Contradiction with minimality of d,
Part c) Suppose for contradiction that Γ axiomatizes the class of wellorders. Then N =
(N, <, +, 0, 1) |= Γ. By part a) M is an elementary extension of N , and so M |= Γ. However,
by part b) <M is not a well order. Contradiction.

10

You might also like