(Real) Locally Convex Topological Vector Space. by the Dual Space X ∗, Or

(Real) Locally Convex Topological Vector Space. by the Dual Space X ∗, Or

CHAPTER V DUAL SPACES DEFINITION Let (X, T ) be a (real) locally convex topological vector space. By the dual space X∗, or (X, T )∗, of X we mean the set of all continuous linear functionals on X. By the weak topology on X we mean the weakest topology W on X for which each f ∈ X∗ is continuous. In this context, the topology T is called the strong topology or original topology on X. EXERCISE 5.1. (a) Prove that X∗ is a vector space under pointwise operations. (b) Show that W ⊆ T . Show also that (X, W) is a locally convex topological vector space. (c) Show that if X is infinite dimensional then every weak neigh- borhood of 0 contains a nontrivial subspace M of X. HINT: If V = n −1 n ∩i=1fi (Ui), and if M = ∩i=1 ker(fi), then M ⊆ V. (d) Show that a linear functional f on X is strongly continuous if and only if it is weakly continuous; i.e., prove that (X, T )∗ = (X, W)∗. (e) Prove that X is finite dimensional if and only if X∗ is finite di- mensional, in which case X and X∗ have the same dimension. EXERCISE 5.2. (a) For 1 < p < ∞, let X be the normed linear 0 space Lp(R). For each g ∈ Lp (R) (1/p + 1/p0 = 1), define a linear functional φg on X by Z φg(f) = f(x)g(x) dx. 81 82 CHAPTER V p0 Prove that the map g → φg is a vector space isomorphism of L (R) onto X∗. (b) By analogy to part a, show that L∞(R) is isomorphic as a vector space to L1(R)∗. (c) Let c0 be the normed linear space of real sequences {a0, a1,... } for which lim an = 0 with respect to the norm defined by k{an}k = max |an|. ∗ 1 1 Show that c0 is algebraically isomorphic to l , where l is the linear space ∗ of all absolutely summable sequences {b0, b1,... }. HINT: If f ∈ c0, n n define bn to be f(e ), where e is the element of c0 that is 1 in the nth position and 0 elsewhere. (d) In each of parts a through c, show that the weak and strong topologies are different. Exhibit, in fact, nets (sequences) which converge weakly but not strongly. ∞ 1 (e) Let X = L (R). For each function g ∈ L (R), define φg on X by R ∗ φg(f) = fg. Show that φg is an element of X . Next, for each finite R Borel measure µ on R, define φµ on X by φµ(f) = f dµ. Show that ∗ 1 φµ is an element of X . Conclude that, in this sense, L (R) is a proper subset of (L∞)∗. (f) Let ∆ be a second countable locally compact Hausdorff space, and let X be the normed linear space C0(∆) equipped with the supremum norm. Identify X∗. (g) Let X1,...,Xn be locally convex topological vector spaces. If Ln ∗ Ln ∗ X = i=1 Xi, show that X is isomorphic to i=1 Xi . THEOREM 5.1. (Relation between the Weak and Strong Topolo- gies) Let (X, T ) be a locally convex topological vector space. (1) Let A be a convex subset of X. Then A is strongly closed if and only if it is weakly closed. (2) If A is a convex subset of X, then the weak closure of A equals the strong closure of A. (3) If {xα} is a net in X that converges weakly to an element x, then there exists a net {yβ}, for which each yβ is a (finite) con- vex combination of some of the xα’s, such that {yβ} converges strongly to x. If T is metrizable, then the net {yβ} can be chosen to be a sequence. PROOF. If A is a weakly closed subset, then it is strongly closed since W ⊆ T . Conversely, suppose that A is a strongly closed convex set and let x ∈ X be an element not in A. Then, by the Separation Theorem, there exists a continuous linear functional φ on X, and a real number s, such that φ(y) ≤ s for all y ∈ A and φ(x) > s. But then DUAL SPACES 83 the set φ−1(s, ∞) is a weakly open subset of X that contains x and is disjoint from A, proving that A is weakly closed, as desired. If A is a convex subset of X, and if B is the weak closure and C is the strong closure, then clearly A ⊆ C ⊆ B. On the other hand, C is convex and strongly closed, hence C is weakly closed. Therefore, B = C, and part 2 is proved. Now let {xα} be a weakly convergent net in X, and let A be the convex hull of the xα’s. If x = limW xα, then x belongs to the weak closure of A, whence to the strong closure of A. Let {yβ} be a net (sequence if T is metrizable) of elements of A that converges strongly to x. Then each yβ is a finite convex combination of certain of the xα’s, and part 3 is proved. DEFINITION. Let X be a locally convex topological vector space, and let X∗ be its dual space. For each x ∈ X, define a functionx ˆ on X∗ byx ˆ(f) = f(x). By the weak∗ topology on X∗, we mean the weakest topology W∗ on X∗ for which each functionx, ˆ for x ∈ X, is continuous. THEOREM 5.2. (Duality Theorem) Let (X, T ) be a locally convex topological vector space, and let X∗ be its dual space. Then: (1) Each function xˆ is a linear functional on X∗. (2) (X∗, W∗) is a locally convex topological vector space. (Each xˆ is continuous on (X∗, W∗).) (3) If φ is a continuous linear functional on (X∗, W∗), then there exists an x ∈ X such that φ =x ˆ; i.e., the map x → xˆ is a linear transformation of X onto (X∗, W∗)∗. (4) The map x → xˆ is a topological isomorphism between (X, W) and ((X∗, W∗)∗, W∗). PROOF. If x ∈ X, then xˆ(af + bg) = (af + bg)(x) = af(x) + bg(x) = axˆ(f) + bxˆ(g), proving part 1. By the definition of the topology W∗, we see that eachx ˆ is continuous. Also, the set of all functions {xˆ} separate the points of X∗, for if f, g ∈ X∗, with f 6= g, then f −g is not the 0 functional. Hence there exists an x ∈ X for which (f −g)(x), which isx ˆ(f)−xˆ(g), is not 0. Therefore, the weak topology on X∗, generated by thex ˆ’s, is a locally convex topology. See part c of Exercise 3.11. 84 CHAPTER V Now suppose φ is a continuous linear functional on (X∗, W∗), and let M be the kernel of φ. If M = X∗, then φ is the 0 functional, which is 0ˆ. Assume then that there exists an f ∈ X∗, for which φ(f) = 1, whence f∈ / M. Since φ is continuous, M is a closed subset in X∗, and there exists a weak* neighborhood V of f which is disjoint from M. Therefore, ∗ by the definition of the topology W , there exists a finite set x1, . , xn of elements of X and a finite set 1, . , n of positive real numbers such that ∗ V = {g ∈ X : |xˆi(g) − xˆi(f)| < i, 1 ≤ i ≤ n}. Define a map R : X∗ → Rn by R(g) = (x ˆ1(g),..., xˆn(g)). Clearly R is a continuous linear transformation of X∗ into Rn. Now R(f) ∈/ R(M), for otherwise there would exist a g ∈ M such thatx ˆi(g) = xˆi(f) for all i. But this would imply that g ∈ V ∩ M, contradicting the choice of the neighborhood V. Also, R(M) is a subspace of Rn, so contains 0, implying then that R(f) 6= 0. Suppose R(M) is of dimension n j < n. Let α1, . , αn be a basis for R , such that α1 = R(f) and n αi ∈ R(M) for 2 ≤ i ≤ j + 1. We define a linear functional p on R by setting p(α1) = 1 and p(αi) = 0 for 2 ≤ i ≤ n. Now, p ◦ R is a continuous linear functional on X∗. If g ∈ M, then (p ◦ R)(g) = p(R(g)) = 0, since R(g) ∈ R(M), which is in the span of α2, . , αn. Also, (p◦R)(f) = p(R(f)) = 1, since R(f) = α1. So, p◦R is a linear functional on X∗ which has the same kernel M as φ and agrees with φ on f. Therefore, φ − p ◦ R = 0 everywhere, and φ = p ◦ R. n Let e1, . , en denote the standard basis for R , and let A be the n × n matrix relating the bases e1, . , en and α1, . , αn. That is, ei = Pn Pn j=1 Aijαj. Then, if α = (a1, . , an) = i=1 aiei, we have n X p(α) = aip(ei) i=1 n n X X = ai Aijp(αj) i=1 j=1 n X = aiAi1. i=1 DUAL SPACES 85 Therefore, φ(g) = p ◦ R(g) = p(R(g)) = p((xc1(g),..., xcn(g))) n X = Ai1xbi(g) i=1 n X = ( Ai1xbi)(g) i=1 n X\ = Ai1xi(g) i=1 =x ˆ(g), Pn where x = i=1 Ai1xi, and this proves part 3. We leave the proof of part 4 to the exercises. EXERCISE 5.3. Prove part 4 of the preceding theorem.

View Full Text

Details

  • File Type
    pdf
  • Upload Time
    -
  • Content Languages
    English
  • Upload User
    Anonymous/Not logged-in
  • File Pages
    18 Page
  • File Size
    -

Download

Channel Download Status
Express Download Enable

Copyright

We respect the copyrights and intellectual property rights of all users. All uploaded documents are either original works of the uploader or authorized works of the rightful owners.

  • Not to be reproduced or distributed without explicit permission.
  • Not used for commercial purposes outside of approved use cases.
  • Not used to infringe on the rights of the original creators.
  • If you believe any content infringes your copyright, please contact us immediately.

Support

For help with questions, suggestions, or problems, please contact us