Frechet and countably normed spaces

[22/9%]
Show
LaTeX

Suppose in a countably-normed space XX each subsequent seminorm majorizes the preceding one, but is not equivalent to it. Prove that XX is not normable.

?

Prove that the spaces s,l2∞,C∞[0,1],A(D)s, l_{2}^{\infty }, C^{\infty }[0,1], \mathcal{A}(D) and C(R)C(\mathbb {R}) are separable Fréchet spaces.

?

Let XX be a countably-normed space, and let YY be a proper closed subspace of it. A vector x∈Xx \in X is called an ε\varepsilon-perpendicular to YY if d(x,0)=1d(x, 0) = 1 and d(x,Y)>1−εd(x, Y) > 1-\varepsilon. Prove that an ε\varepsilon-perpendicular to YY exists for any ε>0\varepsilon > 0.

?

Let XX be a countably-normed Fréchet space. Prove that any system of nested closed balls of the space XX has nonempty intersection (compare with problem 2.68).

?

Let XX be a countably-normed Fréchet space. Prove that if YY is a finite-dimensional subspace of XX, and xx is an arbitrary vector from XX, then there is a vector y0y_{0} in YY for which dist⁡(x,Y)=d(x,y0)\operatorname {dist}(x, Y) = d\left(x, y_{0}\right) (compare with problem 2.70).

?

(Analogue of the Banach theorem on the inverse operator). Let XX and YY be countably-normed Fréchet spaces, and T:X→YT: X \rightarrow Y a linear continuous bijective operator. Prove that TT has a linear continuous inverse operator.

?

(Analogue of the Banach-Steinhaus theorem). Let M\mathcal{M} be an arbitrary family of continuous linear operators from a countably-normed Fréchet space (X,{pn}1∞)(X,\left\{ p_{n}\right\}_{1}^{\infty }) into a countably-normed Fréchet space (Y,{qn}1∞)(Y,\left\{ q_{n}\right\}_{1}^{\infty }), such that for every x∈Xx \in X and n∈Nn \in \mathbb {N} the set {qn(Tx):T∈M}\left\{ q_{n}(T x): T \in \mathcal{M}\right\} is bounded. Prove that for every seminorm qnq_{n} there exist seminorms pk1,…,pkjp_{k_{1}}, \ldots , p_{k_{j}} and a constant Cn>0C_{n} > 0 such that qn(Tx)⩽Cnmax⁡{pk1(x),…,pkj(x)}q_{n}(T x) \leqslant C_{n} \max \left\{ p_{k_{1}}(x), \ldots , p_{k_{j}}(x)\right\} for all x∈Xx \in X and T∈MT \in \mathcal{M}.

?

Prove that the spaces E\mathcal{E} and S\mathcal{S} are metrizable (and, moreover, are Fréchet spaces), while the space D\mathcal{D} is not metrizable.

?

Let XX be completely regular and let C(X)C(X) be all continuous functions from XX into F\mathbb {F}, topologized by the seminorms pK(f)=sup⁡{∣f(x)∣:x∈K}p_K(f) = \sup \left\{ \left|f(x)\right| : x \in K\right\} for KK compact in XX. Then C(X)C(X) is metrizable if and only if X=⋃n=1∞KnX = \bigcup_{n=1}^{\infty } K_n, where each KnK_n is compact, K1⊆K2⊆⋯K_1 \subseteq K_2 \subseteq \cdots, and if KK is any compact subset of XX, then K⊆KnK \subseteq K_n for some nn. Verify this statement.

?

Let XX be locally compact and let C(X)C(X) be as in Exercise 2 above. Then C(X)C(X) is metrizable if and only if XX is σ\sigma-compact (that is, XX is the union of a sequence of compact sets). Also, the space H(G)H(G) of analytic functions on an open set G⊆CG \subseteq \mathbb {C}, topologized by uniform convergence on compact subsets, is metrizable. Verify these statements.

?

If X\mathscr {X} is a LCS, show that X\mathscr {X} is metrizable if and only if X\mathscr {X} is first countable. Is this equivalent to saying that {0}\left\{ 0\right\} is a GδG_\delta set?

?

With the notation of Exercise 6, show that (Cb(X),β)(C_b(X), \beta ) is metrizable if and only if XX is compact.

?

Prove the Open Mapping Theorem for Fréchet spaces.

?

Show that the space ss of all sequences of scalars (with the metric d(x,y)=∑n=1∞2−n∣x(n)−y(n)∣/(1+∣x(n)−y(n)∣)d(x,y) = \sum_{n=1}^{\infty } 2^{-n} \left|x(n)-y(n)\right| / (1 + \left|x(n)-y(n)\right|)) is linearly homeomorphic to C(N)C(\mathbb {N}) and describe s∗s^*.

?

In the space C(R)C(\mathbb {R}) of all continuous real functions on the line introduce the countable system of seminorms pN(f)=max⁡∣x∣≤N∣f(x)∣p_N(f)=\max_{\left|x\right|\leq N}\left|f(x)\right| and set

d(f,g)=∑N=1∞2−NpN(f−g)1+pN(f−g). d(f,g)=\sum _{N=1}^{\infty }2^{-N}\frac{p_N(f-g)}{1+p_N(f-g)}.

For what r>0r>0 is the ball of radius rr a convex set in C(R)C(\mathbb {R})?

?

Let LL be a complete LCS and Ω\Omega a domain in Rn\mathbb {R}^{n}. We define the space E(Ω,L)\mathcal E(\Omega ,L) as the collection of infinitely differentiable functions on Ω\Omega with values in LL. If {pα}α∈A\left\{ p_\alpha \right\}_{\alpha \in A} is the family of seminorms that determines the topology in LL, then the topology in E(Ω,L)\mathcal E(\Omega ,L) is given by the family of seminorms

pKlα(φ)=sup⁡x∈Kpα(∂lφ(x)), p_{Kl\alpha }(\varphi )=\sup _{x\in K}p_\alpha (\partial ^l\varphi (x)),

where KK is a compact subset of Ω\Omega and l=(l1,…,ln)l=(l_1,\ldots ,l_n) is a multi-index, with α∈A\alpha \in A. Prove that E(Ω,L)\mathcal E(\Omega ,L) is a complete LCS, metrizable if LL is metrizable.

?

Let Ω1\Omega_1 and Ω2\Omega_2 be domains in Rn\mathbb {R}^{n} and Rm\mathbb {R}^{m}. Then Ω1×Ω2\Omega_1\times \Omega_2 is a domain in Rn+m\mathbb {R}^{n+m}. In the notation of Problem 427 prove that

E(Ω1,E(Ω2))≃E(Ω1×Ω2)≃E(Ω2,E(Ω1)). \mathcal E(\Omega _1,\mathcal E(\Omega _2))\simeq \mathcal E(\Omega _1\times \Omega _2)\simeq \mathcal E(\Omega _2,\mathcal E(\Omega _1)).
?

Prove that the spaces C(Ω)C(\Omega ) (Section 1.44) do not have the Heine-Borel property.

?

Prove that the topology of C(Ω)C(\Omega ) does not depend on the particular choice of {Kn}\left\{ K_{n}\right\}, as long as this sequence satisfies the conditions specified in Section 1.44. Do the same for C∞(Ω)C^{\infty }(\Omega ) (Section 1.46).

?

In the setting of Section 1.46, prove that f→Dαff \rightarrow D^{\alpha } f is a continuous mapping of C∞(Ω)C^{\infty }(\Omega ) into C∞(Ω)C^{\infty }(\Omega ) and also of DK\mathscr {D}_{K} into DK\mathscr {D}_{K}, for every multi-index α\alpha.

?

Suppose μ\mu is a Borel probability measure on a compact Hausdorff space Q,XQ, X is a Fréchet space, and f:Q→Xf: Q \rightarrow X is continuous. A partition of QQ is, by definition, a finite collection of disjoint Borel subsets of QQ whose union is QQ. Prove that to every neighborhood VV of 0 in XX there corresponds a partition {Ei}\left\{ E_{i}\right\} such that the difference

z=∫Qfdμ−∑iμ(Ei)f(si) z = \int _{Q} f d \mu -\sum _{i} \mu \left(E_{i}\right) f\left(s_{i}\right)

lies in VV for every choice of si∈Eis_{i} \in E_{i}. (This exhibits the integral as a strong limit of "Riemann sums".) Suggestion: Take VV convex and balanced. If Λ∈X∗\Lambda \in X^{*} and if ∣Λx∣≤1\left|\Lambda x\right| \leq 1 for every x∈Vx \in V, then ∣Λz∣≤1\left|\Lambda z\right| \leq 1, provided that the sets EiE_{i} are chosen so that f(s)−f(t)∈Vf(s)-f(t) \in V whenever ss and tt lie in the same EiE_{i}.

?
Problem 3.3.26 from Rudin W. "Functional Analysis" [Rud91]More problems in Chapter 3.3 "Convexity"

Suppose Ω\Omega is a region in C,X\mathbb {C}, X is a Fréchet space, and f:Ω→Xf: \Omega \rightarrow X is holomorphic.

?
(a)

State and prove a theorem concerning the power series representation of ff, that is, concerning the formula f(z)=∑(z−a)ncnf(z) = \sum (z-a)^{n} c_{n}, where cn∈Xc_{n} \in X.

(b)

Generalize Morera's theorem to XX-valued holomorphic functions.

(c)

For a sequence of complex holomorphic functions in Ω\Omega, uniform convergence on compact subsets of Ω\Omega implies that the limit is holomorphic. Does this generalize to XX-valued holomorphic functions?

References

[БСШ17] Бородин П.А., Савчук А.М., Шейпак И.А. Задачи по функциональному анализу. МЦНМО, 2017.
[Con07] Conway J.B. A Course in Functional Analysis. 2nd ed., Springer, 2007.
[KiGv88] Kirillov A.A., Gvishiani A.D. Теоремы и задачи функционального анализа. 2-е изд., Наука, 1988.
[Rud91] Rudin W. Functional Analysis. 2nd ed., McGraw-Hill, 1991.