Chapter 2.19

Holomorphic Fourier Transforms

[16/0%]
LaTeX
Problem 2.19.1

Suppose ff is an entire function of exponential type and

φ(y)=∫−∞∞∣f(x+iy)∣2dx \varphi (y) = \int _{-\infty }^{\infty }\left|f(x+i y)\right|^{2} d x

Prove that either φ(y)=∞\varphi (y) = \infty for all real yy or φ(y)<∞\varphi (y) < \infty for all real yy. Prove that f=0f = 0 if φ\varphi is a bounded function.

?
Problem 2.19.2

Suppose ff is an entire function of exponential type such that the restriction of ff to two nonparallel lines belongs to L2L^{2}. Prove that f=0f = 0.

?
Problem 2.19.3

Suppose ff is an entire function of exponential type whose restriction to two nonparallel lines is bounded. Prove that ff is constant. (Apply Exercise 9 of Chap. 12).

?
Problem 2.19.4

Suppose ff is entire, ∣f(z)∣<Cexp⁡(A∣z∣)\left|f(z)\right| < C \exp (A\left|z\right|), and f(z)=Σanznf(z) = \Sigma a_{n} z^{n}. Put

Φ(w)=∑n=0∞n!anwn+1 \Phi (w) = \sum _{n = 0}^{\infty } \frac{n!a_{n}}{w^{n+1}}

Prove that the series converges if ∣w∣>A\left|w\right| > A, that

f(z)=12πi∫ΓΦ(w)ewzdw f(z) = \frac{1}{2 \pi i} \int _{\Gamma } \Phi (w) e^{w z} d w

if Γ(t)=(A+ϵ)eit,0≤t≤2π\Gamma (t) = (A+\epsilon ) e^{i t}, 0 \leq t \leq 2 \pi, and that Φ\Phi is the function which occurred in the proof of Theorem 19.3. (See also Sec. 19.4).

?
Problem 2.19.5

Suppose ff satisfies the hypothesis of Theorem 19.2. Prove that the Cauchy formula

f(z)=12πi∫−∞∞f(ξ+iϵ)ξ+iϵ−zdξ(0<ϵ<y)(*) f(z) = \frac{1}{2 \pi \mathrm{i}} \int _{-\infty }^{\infty } \frac{f(\xi +\mathrm{i} \epsilon )}{\xi +i \epsilon -z} d \xi \quad (0 < \epsilon < y) \tag {*}

holds; here z=x+iyz = x+i y. Prove that

f∗(x)=lim⁡y→0f(x+iy) f^{*}(x) = \lim _{y \rightarrow 0} f(x+i y)

exists for almost all xx. What is the relation between f∗f^{*} and the function FF which occurs in Theorem 19.2? Is (∗)({ }^{*}) true with ϵ=0\epsilon = 0 and with f∗f^{*} in place of ff in the integrand?

?
Problem 2.19.6

Suppose φ∈L2(−∞,∞)\varphi \in L^{2}(-\infty , \infty ) and φ>0\varphi > 0. Prove that there exists an ff with ∣f∣=φ\left|f\right| = \varphi such that the Fourier transform of ff vanishes on a half line if and only if

∫−∞∞log⁡φ(x)dx1+x2>−∞ \int _{-\infty }^{\infty } \log \varphi (x) \frac{d x}{1+x^{2}} > -\infty

Suggestion: Consider f∗f^{*}, as in Exercise 5, where f=exp⁡(u+iv)f = \exp (u+i v) and

u(z)=1π∫−∞∞y(x−t)2+y2log⁡φ(t)dt u(z) = \frac{1}{\pi } \int _{-\infty }^{\infty } \frac{y}{(x-t)^{2}+y^{2}} \log \varphi (t) d t
?
Problem 2.19.7

Let ff be a complex function on a closed set EE in the plane. Prove that the following two conditions on ff are equivalent:

?
(a)

There is an open set Ω⊃E\Omega \supset E and a function F∈H(Ω)F \in H(\Omega ) such that F(z)=f(z)F(z) = f(z) for z∈Ez \in E.

(b)

To each α∈E\alpha \in E there corresponds a neighborhood VαV_{\alpha } of α\alpha and a function Fα∈H(Vα)F_{\alpha } \in H\left(V_{\alpha }\right) such that Fα(z)=f(z)F_{\alpha }(z) = f(z) in Vα∩EV_{\alpha } \cap E. (A special case of this was proved in Theorem 19.9).

Problem 2.19.8

Prove that C{n!}=C{nn}C\left\{ n!\right\} = C\left\{ n^{n}\right\}.

?
Problem 2.19.9

Prove that there are quasi-analytic classes which are larger than C{n!}C\left\{ n!\right\}.

?
Problem 2.19.10

Put λn=Mn−1/Mn\lambda_{n} = M_{n-1} / M_{n}, as in the proof of Theorem 19.11. Pick g0∈Cc(R1)g_{0} \in C_{c}\left(\mathbb {R}^{1}\right), and define

gn(x)=(2λn)−1∫−λnλngn−1(x−t)dt(n=1,2,3,…) g_{n}(x) = \left(2 \lambda _{n}\right)^{-1} \int _{-\lambda _{n}}^{\lambda _{n}} g_{n-1}(x-t) d t \quad (n = 1,2,3, \ldots )

Prove directly (without using Fourier transforms or holomorphic functions) that g=lim⁡gng = \lim g_{n} is a function which demonstrates that (e) implies (a) in Theorem 19.11. (You may choose any g0g_{0} that is convenient).

?
Problem 2.19.11

Find an explicit formula for a function φ∈C∞\varphi \in C^{\infty }, with support in [-2,2], such that φ(x)=1\varphi (x) = 1 if −1≤x≤1-1 \leq x \leq 1.

?
Problem 2.19.12

Prove that to every sequence {αn}\left\{ \alpha_{n}\right\} of complex numbers there corresponds a function f∈C∞f \in C^{\infty } such that (Dnf)(0)=αn\left(D^{n} f\right)(0) = \alpha_{n} for n=0,1,2,…n = 0,1,2, \ldots Suggestion: If φ\varphi is as in Exercise 11, if βn=αn/n!\beta_{n} = \alpha_{n} / n!, if gn(x)=βnxnφ(x)g_{n}(x) = \beta_{n} x^{n} \varphi (x), and if

fn(x)=λn−ngn(λnx)=βnxnφ(λnx) f_{n}(x) = \lambda _{n}^{-n} g_{n}\left(\lambda _{n} x\right) = \beta _{n} x^{n} \varphi \left(\lambda _{n} x\right)

then ∥Dkfn∥∞<2−n\left\| D^{k} f_{n}\right\|_{\infty } < 2^{-n} for k=0,…,n−1k = 0, \ldots , n-1, provided that λn\lambda_{n} is large enough. Take f=Σfnf = \Sigma f_{n}.

?
Problem 2.19.13

Construct a function f∈C∞f \in C^{\infty } such that the power series

∑n=0∞(Dnf)(a)n!(x−a)n \sum _{n = 0}^{\infty } \frac{\left(D^{n} f\right)(a)}{n!}(x-a)^{n}

has radius of convergence 0 for every a∈R1a \in \mathbb {R}^{1}. Suggestion: Put

f(z)=∑k=1∞ckeiλkx f(z) = \sum _{k = 1}^{\infty } c_{k} e^{i \lambda _{k} x}

where {ck}\left\{ c_{k}\right\} and {λk}\left\{ \lambda_{k}\right\} are sequences of positive numbers, chosen so that Σckλkn<∞\Sigma c_{k} \lambda_{k}^{n} < \infty for n=0,1,2,…n = 0,1,2, \ldots and so that cnλnnc_{n} \lambda_{n}^{n} increases very rapidly and is much larger than the sum of all the other terms in the series Σckλkn\Sigma c_{k} \lambda_{k}^{n}.

For instance, put ck=λk1−kc_{k} = \lambda_{k}^{1-k}, and choose {λk}\left\{ \lambda_{k}\right\} so that

λk>2∑j=1k−1cjλjk and λk>k2k \lambda _{k} > 2 \sum _{j = 1}^{k-1} c_{j} \lambda _{j}^{k} \text{ and } \lambda _{k} > k^{2 k}
?
Problem 2.19.14

Suppose C{Mn}C\left\{ M_{n}\right\} is quasi-analytic, f∈C{Mn}f \in C\left\{ M_{n}\right\}, and f(x)=0f(x) = 0 for infinitely many x∈[0,1]x \in [0,1]. What follows?

?
Problem 2.19.15

Let XX be the vector space of all entire functions ff that satisfy ∣f(z)∣≤Ceπ∣z∣\left|f(z)\right| \leq C e^{\pi \left|z\right|} for some C<∞C < \infty, and whose restriction to the real axis is in L2L^{2}. Associate with each f∈Xf \in X its restriction to the integers. Prove that f→{f(n)}f \rightarrow \left\{ f(n)\right\} is a linear one-to-one mapping of XX onto ℓ2\ell^{2}.

?
Problem 2.19.16

Assume ff is a measurable function on (−∞,∞)(-\infty , \infty ) such that ∣f(x)∣<e−∣x∣\left|f(x)\right| < e^{-\left|x\right|} for all xx. Prove that its Fourier transform f^\hat{f} cannot have compact support, unless f(x)=0f(x) = 0 a.e.

?