Real and Complex Analysis General Exam Fall 2007

Problem 1

Suppose that ( X,𝒜,μX, \mathcal{A}, \mu ) is a measure space, with 𝒜\mathcal{A} a σ\sigma-algebra, μ\mu a finite measure. Let 𝒜0𝒜\mathcal{A}_{0} \subset \mathcal{A} be a sub-algebra of sets which generates 𝒜\mathcal{A}; thus 𝒜\mathcal{A} is the smallest σ\sigma-algebra generated by 𝒜0\mathcal{A}_{0}.
Let 𝒜\mathcal{B} \subset \mathcal{A} be the collection of subsets of XX with the property that for every ϵ\epsilon and BB \in \mathcal{B}, there is an A𝒜0A \in \mathcal{A}_{0}, with

μ(AΔB)<ϵ.\mu(A \Delta B)<\epsilon .

(Here, Δ\Delta is the symmetric difference, CΔD=(CD)(DC)C \Delta D=(C-D) \cup(D-C).)
Show that \mathcal{B} is a σ\sigma-algebra containing 𝒜0\mathcal{A}_{0}, in particular that it is an algebra, and that it is closed under countable (increasing) unions.

Hint: You may assume that (do not prove!):

μ(CΔD)=μ(CcΔDc)μ((D1D2)Δ(E1E2))μ(D1ΔE1)+μ(D2ΔE2)|μ(CΔE)μ(DΔE)|μ(CΔD)\begin{aligned} \mu(C \Delta D) & =\mu\left(C^{c} \Delta D^{c}\right) \\ \mu\left(\left(D_{1} \cup D_{2}\right) \Delta\left(E_{1} \cup E_{2}\right)\right) & \leq \mu\left(D_{1} \Delta E_{1}\right)+\mu\left(D_{2} \Delta E_{2}\right) \\ |\mu(C \Delta E)-\mu(D \Delta E)| & \leq \mu(C \Delta D) \end{aligned}

Problem 2

(a) Let f(x)f(x) be a real-valued differentiable function on a neighborhood of [a,b][a, b], and assume that dfdx(a)<0\frac{d f}{d x}(a)<0, and dfdx(b)>0\frac{d f}{d x}(b)>0. Show that there exists a c(a,b)c \in(a, b) such that dfdx(c)=0\frac{d f}{d x}(c)=0.
(b) Suppose again that dfdx(a)<dfdx(b)\frac{d f}{d x}(a)<\frac{d f}{d x}(b) and that λ\lambda satisfies dfdx(a)<λ<dfdx(b)\frac{d f}{d x}(a)<\lambda<\frac{d f}{d x}(b). Show that there exists a c(a,b)c \in(a, b) with dfdx(c)=λ\frac{d f}{d x}(c)=\lambda.

Problem 3

Let ( X,μX, \mu ) be a measure space.
(a) Show that if gg is a non-negative square-integrable function on XX, then

{x:g(x)λ}g(x)dμ(x)1λXg2(x)dμ(x)\int_{\{x: g(x) \geq \lambda\}} g(x) d \mu(x) \leq \frac{1}{\lambda} \int_{X} g^{2}(x) d \mu(x)

Suppose now that ( X,μX, \mu ) is a finite measure space and let {fn}n=1,2,\left\{f_{n}\right\}_{n=1,2, \ldots} be a sequence of non-negative functions which are both integrable and square-integrable, i.e., in L1(X)L2(X)L^{1}(X) \cap L^{2}(X), that they converge pointwise to a function f(x)f(x), and that the limits

limnfn(x)dμ(x)=L and limnfn2(x)dμ(x)=M\lim _{n \rightarrow \infty} \int f_{n}(x) d \mu(x)=L \text { and } \lim _{n \rightarrow \infty} \int f_{n}^{2}(x) d \mu(x)=M

exist.
(b) Show that

limnfn(x)dμ(x)=f(x)dμ(x)\lim _{n \rightarrow \infty} \int f_{n}(x) d \mu(x)=\int f(x) d \mu(x)

Hints: Why is

f(x)dμ(x)liminffn(x)dμ(x)?\int f(x) d \mu(x) \leq \liminf \int f_{n}(x) d \mu(x) ?

To show an inequality in the other direction: Define cut-off functions:

fn,λ(x){fn(x),fn(x)λλ otherwise f_{n, \lambda}(x) \equiv\left\{\begin{array}{c} f_{n}(x), \quad f_{n}(x) \leq \lambda \\ \lambda \text { otherwise } \end{array}\right.

which converge pointwise to

fλ(x){f(x),f(x)λλ otherwise f_{\lambda}(x) \equiv\left\{\begin{array}{c} f(x), \quad f(x) \leq \lambda \\ \lambda \text { otherwise } \end{array}\right.

Show that, given ϵ\epsilon, there is a λ\lambda such that (again, μ\mu is a finite measure and use part (a))

limsupnfn(x)dμ(x)limsupnfn,λ(x)dμ(x)+ϵfλ(x)dμ(x)+ϵ\begin{aligned} \limsup _{n} \int f_{n}(x) d \mu(x) & \leq \limsup _{n} \int f_{n, \lambda}(x) d \mu(x)+\epsilon \\ & \leq \int f_{\lambda}(x) d \mu(x)+\epsilon \end{aligned}

which is clearly

f(x)dμ(x)+ϵ\leq \int f(x) d \mu(x)+\epsilon

Problem 4

Let ff be an entire function on the complex plane, and suppose there is a positive integer NN such that

f(z)zN0 as |z|\frac{f(z)}{z^{N}} \rightarrow 0 \text { as }|z| \rightarrow \infty

Find an upper bound on the number of zeros of ff (counting multiplicity) in the complex plane in terms of NN. Is the bound sharp?

Problem 5

Let \mathcal{H} be the Hilbert space of L2(D,μ)L^{2}(D, \mu) functions on the unit discD={z𝒞:|z|<1}\operatorname{disc} D=\{z \in \mathcal{C}:|z|<1\}, with μ\mu two-dimensional measure, dμ=dxdy=rdrdθd \mu=d x d y=r d r d \theta in polar coordinates, with fL2(D)\|f\|_{L^{2}(D)} its L2L^{2}-norm. Suppose that ff \in \mathcal{H} is moreover analytic in DD.
(a) Show that, for 0<r<10<r<1

f(0)=12π02πf(reiθ)dθf(0)=\frac{1}{2 \pi} \int_{0}^{2 \pi} f\left(r e^{i \theta}\right) d \theta

and that for 0<r1<10<r_{1}<1,

f(0)=1πr1202π0r1f(reiθ)rdrdθf(0)=\frac{1}{\pi r_{1}^{2}} \int_{0}^{2 \pi} \int_{0}^{r_{1}} f\left(r e^{i \theta}\right) r d r d \theta

(b) Show that

|f(0)|1πr1fL2(D)|f(0)| \leq \frac{1}{\sqrt{\pi} r_{1}}\|f\|_{L^{2}(D)}

for all ff \in \mathcal{H} which are analytic in DD.

Problem 6

Use the method of residues to evaluate

xsinxx2+4dx\int_{-\infty}^{\infty} \frac{x \sin x}{x^{2}+4} d x

Show all estimates.

Problem 7

Suppose that ff is analytic on A={z:1|z|3}A=\{z \in \mathbb{C}: 1 \leq|z| \leq 3\}, and assume that |f(z)|1|f(z)| \leq 1 for |z|=1|z|=1 and |f(z)|9|f(z)| \leq 9 for |z|=3|z|=3. Prove that |f(2i)|4|f(2 i)| \leq 4.