Chapter 3.6

Test Functions and Distributions

[27/0%]
LaTeX
Problem 3.6.1

Suppose ff is a complex continuous function in Rn\mathbb {R}^{n}, with compact support. Prove that ψPj→f\psi P_{j} \rightarrow f uniformly on Rn\mathbb {R}^{n}, for some ψ∈D\psi \in \mathscr {D} and for some sequence {Pj}\left\{ P_{j}\right\} of polynomials.

?
Problem 3.6.2

Show that the metrizable topology for D(Ω)\mathscr {D}(\Omega ) that was rejected in Section 6.2 is not complete for any Ω\Omega.

?
Problem 3.6.3

If EE is an arbitrary closed subset of Rn\mathbb {R}^{n}, show that there is an f∈C∞(Rn)f \in C^{\infty }\left(\mathbb {R}^{n}\right) such that f(x)=0f(x) = 0 for every x∈Ex \in E and f(x)>0f(x) > 0 for every other x∈Rnx \in \mathbb {R}^{n}.

?
Problem 3.6.4

Suppose Λ∈D′(Ω)\Lambda \in \mathscr {D}^{\prime }(\Omega ) and Λϕ≥0\Lambda \phi \geq 0 whenever ϕ∈D(Ω)\phi \in \mathscr {D}(\Omega ) and ϕ≥0\phi \geq 0. Prove that Λ\Lambda is then a positive measure in Ω\Omega (which is finite on compact sets).

?
Problem 3.6.5

Prove that the numbers cαβc_{\alpha \beta } in the Leibniz formula are

cαβ=∏i=1nαi!βi!(αi−βi)! c_{\alpha \beta } = \prod _{i = 1}^{n} \frac{\alpha _{i}!}{\beta _{i}!\left(\alpha _{i}-\beta _{i}\right)!}
?
Problem 3.6.6
?
(a)

Suppose cm=exp⁡{−(m!)!},m=0,1,2,…c_{m} = \exp \left\{ -(m!)!\right\} , m = 0,1,2, \ldots. Does the series

∑m=0∞cm(Dmϕ)(0) \sum _{m = 0}^{\infty } c_{m}\left(D^{m} \phi \right)(0)

converge for every ϕ∈C∞(R)\phi \in C^{\infty }(R)?

(b)

Let Ω\Omega be open in Rn\mathbb {R}^{n}, suppose Λi∈D′(Ω)\Lambda_{i} \in \mathscr {D}^{\prime }(\Omega ), and suppose that all Λi\Lambda_{i} have their supports in some fixed compact K⊂ΩK \subset \Omega. Prove that the sequence {Λi}\left\{ \Lambda_{i}\right\} cannot converge in D′(Ω)\mathscr {D}^{\prime }(\Omega ) unless the orders of the Λj\Lambda_{j} are bounded. Hint: Use the Banach-Steinhaus theorem.

(c)

Can the assumption about the supports be dropped in (b)?

Problem 3.6.7

Let Ω=(0,∞)\Omega = (0, \infty ). Define

Λϕ=∑m=1∞(Dmϕ)(1m)[ϕ∈D(Ω)] \Lambda \phi = \sum _{m = 1}^{\infty }\left(D^{m} \phi \right)\left(\frac{1}{m}\right) \quad [\phi \in \mathscr {D}(\Omega )]

Prove that Λ\Lambda is a distribution of infinite order in Ω\Omega. Prove that Λ\Lambda cannot be extended to a distribution in RR; that is, there exists no Λ0∈D′(R)\Lambda_{0} \in \mathscr {D}^{\prime }(R) such that Λ0=Λ\Lambda_{0} = \Lambda in (0,∞)(0, \infty ).

?
Problem 3.6.8

Characterize all distributions whose supports are finite sets.

?
Problem 3.6.9
?
(a)

Prove that a set E⊂D(Ω)E \subset \mathscr {D}(\Omega ) is bounded if and only if

sup⁡{∣Λϕ∣:ϕ∈E}<∞ \sup \left\{ \left|\Lambda \phi \right|: \phi \in E\right\} < \infty

for every Λ∈D′(Ω)\Lambda \in \mathscr {D}^{\prime }(\Omega ).

(b)

Suppose {ϕj}\left\{ \phi_{j}\right\} is a sequence in D(Ω)\mathscr {D}(\Omega ) such that {Λϕj}\left\{ \Lambda \phi_{j}\right\} is a bounded sequence of numbers, for every Λ∈D′(Ω)\Lambda \in \mathscr {D}^{\prime }(\Omega ). Prove that some subsequence of {ϕj}\left\{ \phi_{j}\right\} converges, in the topology of D(Ω)\mathscr {D}(\Omega ).

(c)

Suppose {Λj}\left\{ \Lambda_{j}\right\} is a sequence in D′(Ω)\mathscr {D}^{\prime }(\Omega ) such that {Λjϕ}\left\{ \Lambda_{j} \phi \right\} is bounded, for every ϕ∈D(Ω)\phi \in \mathscr {D}(\Omega ). Prove that some subsequence of {Λj}\left\{ \Lambda_{j}\right\} converges in D′(Ω)\mathscr {D}^{\prime }(\Omega ) and that the convergence is uniform on every bounded subset of D(Ω)\mathscr {D}(\Omega ). Hint: By the Banach-Steinhaus theorem, the restrictions of the Λj\Lambda_{j} to DK\mathscr {D}_{K} are equicontinuous. Apply Ascoli's theorem.

Problem 3.6.10

Suppose {fi}\left\{ f_{i}\right\} is a sequence of locally integrable functions in Ω\Omega (an open set in Rn\mathbb {R}^{n}) and

lim⁡i→∞∫K∣fi(x)∣dx=0 \lim _{i \rightarrow \infty } \int _{K}\left|f_{i}(x)\right| d x = 0

for every compact K⊂ΩK \subset \Omega. Prove that then Dαfi→0D^{\alpha } f_{i} \rightarrow 0 in D′(Ω)\mathscr {D}^{\prime }(\Omega ), as i→∞i \rightarrow \infty, for every multi-index α\alpha.

?
Problem 3.6.11

Suppose Ω\Omega is open in R2\mathbb {R}^{2}, and {fi}\left\{ f_{i}\right\} is a sequence of harmonic functions in Ω\Omega that converges in the distribution sense to some Λ∈D′(Ω)\Lambda \in \mathscr {D}^{\prime }(\Omega ); explicitly, the assumption is that

Λϕ=lim⁡i→∞∫Ωfi(x)ϕ(x)dx[ϕ∈D(Ω)] \Lambda \phi = \lim _{i \rightarrow \infty } \int _{\Omega } f_{i}(x) \phi (x) d x \quad [\phi \in \mathscr {D}(\Omega )]

Prove then that {fi}\left\{ f_{i}\right\} converges uniformly on every compact subset of Ω\Omega and that Λ\Lambda is a harmonic function. Hint: If ff is harmonic, f(x)f(x) is the average of ff over small circles centered at xx.

?
Problem 3.6.12

Recall that δ\delta (the Dirac measure) is the distribution defined by δ(ϕ)=ϕ(0)\delta (\phi ) = \phi (0), for ϕ∈D(R)\phi \in \mathscr {D}(R). For which f∈C∞(R)f \in C^{\infty }(R) is it true that fδ′=0f \delta^{\prime } = 0? Answer the same question for fδ′′f \delta^{\prime \prime }. Conclude that a function f∈C∞(R)f \in C^{\infty }(R) may vanish on the support of a distribution Λ∈D′(R)\Lambda \in \mathscr {D}^{\prime }(R) although fΛ≠0f \Lambda \neq 0.

?
Problem 3.6.13

If ϕ∈D(Ω)\phi \in \mathscr {D}(\Omega ) and Λ∈D′(Ω)\Lambda \in \mathscr {D}^{\prime }(\Omega ), does either of the statements

ϕΛ=0,Λϕ=0 \phi \Lambda = 0, \quad \Lambda \phi = 0

imply the other?

?
Problem 3.6.14

Suppose KK is the closed unit ball in Rn,Λ∈D′(Rn)\mathbb {R}^{n}, \Lambda \in \mathscr {D}^{\prime }\left(\mathbb {R}^{n}\right) has its support in KK, and f∈C∞(Rn)f \in C^{\infty }\left(\mathbb {R}^{n}\right) vanishes on KK. Prove that fΛ=0f \Lambda = 0. Find other sets KK for which this is true. (Compare with Exercise 12).

?
Problem 3.6.15

Suppose K⊂V⊂Ω,KK \subset V \subset \Omega , K is compact, VV and Ω\Omega are open in Rn,Λ∈D′(Ω)\mathbb {R}^{n}, \Lambda \in \mathscr {D}^{\prime }(\Omega ) has its support in KK, and {ϕi}⊂D(Ω)\left\{ \phi_{i}\right\} \subset \mathscr {D}(\Omega ) satisfies

lim⁡i→∞[sup⁡x∈V∣(Dαϕi)(x)∣]=0(a) \lim _{i \rightarrow \infty }\left[\sup _{x \in V}\left|\left(D^{\alpha } \phi _{i}\right)(x)\right|\right] = 0 \tag {a}

for every multi-index α\alpha. Prove that then

lim⁡t→∞Λ(ϕi)=0 \lim _{t \rightarrow \infty } \Lambda \left(\phi _{i}\right) = 0
?
Problem 3.6.16

The preceding statement becomes false if VV is replaced by KK in the hypothesis (a). Show this by means of the following example, in which Ω=R\Omega = R. Choose c1>c2>⋯>0c_{1} > c_{2} > \cdots > 0, such that ∑cj<∞\sum c_{j} < \infty; define

Λϕ=∑j=1∞(ϕ(cj)−ϕ(0))(ϕ∈D(R)) \Lambda \phi = \sum _{j = 1}^{\infty }\left(\phi \left(c_{j}\right)-\phi (0)\right) \quad (\phi \in \mathscr {D}(R))

and consider functions ϕi∈D(R)\phi_{i} \in \mathscr {D}(R) such that ϕi(x)=0\phi_{i}(x) = 0 if x≤ci+1,ϕi(x)=1/ix \leq c_{i+1}, \phi_{i}(x) = 1 / i if ci≤x≤c1c_{i} \leq x \leq c_{1}. Show also that this Λ\Lambda is a distribution of order 1.

However, for certain K,VK, V can be replaced by KK in the hypothesis (a) of Exercise 15. Show that this is so when KK is the closed unit ball of Rn\mathbb {R}^{n}. Find other sets KK for which this is true.

?
Problem 3.6.17

If Λ∈D′(R)\Lambda \in \mathscr {D}^{\prime }(R) has order NN, show that Λ=DN+2f\Lambda = D^{N+2} f, for some continuous function ff. If Λ=δ\Lambda = \delta, what are the possibilities for ff?.

?
Problem 3.6.18

Express δ∈D′(R2)\delta \in \mathscr {D}^{\prime }\left(\mathbb {R}^{2}\right) in the form given by Theorem 6.27, as explicitly as you can.

?
Problem 3.6.19

Suppose Λ∈D′(Ω),ϕ∈D(Ω)\Lambda \in \mathscr {D}^{\prime }(\Omega ), \phi \in \mathscr {D}(\Omega ), and (Dαϕ)(x)=0\left(D^{\alpha } \phi \right)(x) = 0 for every xx in the support of Λ\Lambda and for every multi-index α\alpha. Prove that Λϕ=0\Lambda \phi = 0. Suggestion: Do it first for distributions with compact support, by the method used in Theorem 6.25.

?
Problem 3.6.20

Prove that every continuous linear functional on C∞(Ω)C^{\infty }(\Omega ) is of the form f→Λff \rightarrow \Lambda f, where Λ\Lambda is a distribution with compact support in Ω\Omega; this is a converse to (d)(d) of Theorem 6.24.

?
Problem 3.6.21

Let C∞(T)C^{\infty }(T) be the space of all infinitely differentiable complex functions on the unit circle TT in C\mathscr {C}. One may regard C∞(T)C^{\infty }(T) as the subspace of C∞(R)C^{\infty }(R) consisting of those functions that have period 2π2 \pi. Suppose

f(z)=∑n=0∞anzn f(z) = \sum _{n = 0}^{\infty } a_{n} z^{n}

converges in the open unit disc UU in C\mathcal{C}. Prove that each of the following three properties of ff implies the other two:

?
(a)

There exist p<∞p < \infty and γ<∞\gamma < \infty such that

∣an∣≤γ⋅np(n=1,2,3,…) \left|a_{n}\right| \leq \gamma \cdot n^{p} \quad (n = 1,2,3, \ldots )
(b)

There exist p<∞p < \infty and γ<∞\gamma < \infty such that

∣f(z)∣≤γ⋅(1−∣z∣)−p(z∈U) \left|f(z)\right| \leq \gamma \cdot (1-\left|z\right|)^{-p} \quad (z \in U)
(c)

lim⁡r→1∫−ππf(reiθ)ϕ(eiθ)dθ\lim_{r \rightarrow 1} \int_{-\pi }^{\pi } f\left(r e^{i \theta }\right) \phi \left(e^{i \theta }\right) d \theta exists (as a complex number) for every ϕ∈C∞(T)\phi \in C^{\infty }(T).

Problem 3.6.22

For u∈D′(R)u \in \mathscr {D}^{\prime }(R), show that

u−τxux→Du in D′(R) \frac{u-\tau _{x} u}{x} \rightarrow D u \quad \text{ in } \mathscr {D}^{\prime }(R)

as x→0x \rightarrow 0. (The derivative of uu may thus still be regarded as a limit of quotients).

?
Problem 3.6.23

Suppose {fi}\left\{ f_{i}\right\} is a sequence of locally integrable functions in Rn\mathbb {R}^{n}, such that

lim⁡i→∞(fi∗ϕ)(x) \lim _{i \rightarrow \infty }\left(f_{i} * \phi \right)(x)

exists, for each ϕ∈D(Rn)\phi \in \mathscr {D}\left(\mathbb {R}^{n}\right) and each x∈Rnx \in \mathbb {R}^{n}. Prove that then {Da(fi∗ϕ)}\left\{ D^{a}\left(f_{i} * \phi \right)\right\} converges uniformly on compact sets, for each multi-index α\alpha.

?
Problem 3.6.24

Let HH be the Heaviside function on RR, defined by

H(x)={1 if x>00 if x≤0 H(x) = \begin{cases} 1 & \text{ if } x > 0 \\ 0 & \text{ if } x \leq 0\end{cases}

and let δ\delta be the Dirac measure.

?
(a)

Show that (H∗ϕ)(x)=∫−∞xϕ(s)ds(H * \phi )(x) = \int_{-\infty }^{x} \phi (s) d s, if ϕ∈D(R)\phi \in \mathscr {D}(R).

(b)

Show that δ′∗H=δ\delta^{\prime } * H = \delta.

(c)

Show that 1∗δ′=01 * \delta^{\prime } = 0. (Here 1 denotes the locally integrable function whose value is 1 at every point and which is thought of as a distribution).

(d)

It follows that the associative law fails:

1∗(δ′∗H)=1∗δ=1 1 *\left(\delta ^{\prime } * H\right) = 1 * \delta = 1

but

(1∗δ′)∗H=0∗H=0 \left(1 * \delta ^{\prime }\right) * H = 0 * H = 0
Problem 3.6.25

Here is another characterization of convolutions analogous to Theorem 6.33. Suppose LL is a continuous linear mapping of D\mathscr {D} into C∞C^{\infty } which commutes with every DαD^{\alpha }, that is, (a)

LDαϕ=DαLϕ(ϕ∈D) L D^{\alpha } \phi = D^{\alpha } L \phi \quad (\phi \in \mathscr {D})

Then there is a u∈D′u \in \mathscr {D}^{\prime } such that

Lϕ=u∗ϕ L \phi = u * \phi

Suggestion: Fix ϕ∈D\phi \in \mathscr {D}, put

h(x)=(τ−xLτxϕ)(0)=(Lτxϕ)(x)(x∈Rn) h(x) = \left(\tau _{-x} L \tau _{x} \phi \right)(0) = \left(L \tau _{x} \phi \right)(x) \quad \left(x \in \mathbb {R}^{n}\right)

let DeD_{e} be the directional derivative used in the proof of Theorem 6.30, and show that

(Deh)(x)=(DeLτxϕ)(x)−(LτxDeϕ)(x) \left(D_{e} h\right)(x) = \left(D_{e} L \tau _{x} \phi \right)(x)-\left(L \tau _{x} D_{e} \phi \right)(x)

which is 0 if (a) holds. Thus h(x)=h(0)h(x) = h(0), which implies that τxL=Lτx\tau_{x} L = L \tau_{x}. Can the assumption that the range of LL is in C∞C^{\infty } be weakened?

?
Problem 3.6.26

If f∈L1((−∞,−δ)∪(δ,∞))f \in L^{1}((-\infty ,-\delta ) \cup (\delta , \infty )) for every δ>0\delta > 0, define its principal value integral to be

PV∫−∞∞f(x)dx=lim⁡δ→0(∫−∞−δ+∫δ∞)f(x)dx P V \int _{-\infty }^{\infty } f(x) d x = \lim _{\delta \rightarrow 0}\left(\int _{-\infty }^{-\delta }+\int _{\delta }^{\infty }\right) f(x) d x

if the limit exists. For ϕ∈D(R)\phi \in \mathscr {D}(R), put

Λϕ=∫−∞∞ϕ(x)log⁡∣x∣dx \Lambda \phi = \int _{-\infty }^{\infty } \phi (x) \log \left|x\right| d x

Show that

Λ′ϕ=PV∫−∞∞ϕ(x)dxxΛ′′ϕ=−PV∫−∞∞ϕ(x)−ϕ(0)x2dx \begin{aligned} & \Lambda ^{\prime } \phi = P V \int _{-\infty }^{\infty } \phi (x) \frac{d x}{x} \\ & \Lambda ^{\prime \prime } \phi = -P V \int _{-\infty }^{\infty } \frac{\phi (x)-\phi (0)}{x^{2}} d x \end{aligned}
?
Problem 3.6.27

Find all distributions u∈D′(Rn)u \in \mathscr {D}^{\prime }\left(\mathbb {R}^{n}\right) that satisfy at least one of the following two conditions:

?
(a)

τxu=u\tau_{x} u = u for every x∈Rnx \in \mathbb {R}^{n},

(b)

Dαu=0D^{\alpha } u = 0 for every α\alpha with ∣α∣=1\left|\alpha \right| = 1.