## Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

311

16. Banach Spaces II Theorem 16.1 (Open Mapping Theorem). Let X, Y be Banach spaces, T ∈ L(X, Y ). If T is surjective then T is an open mapping, i.e. T (V ) is open in Y for all open subsets V ⊂ X.

X Y Proof. For all α > 0 let Bα = {x ∈ X : kxkX < α} ⊂ X, Bα = X {y ∈ Y : kykY < α} ⊂ Y and Eα = T (Bα ) ⊂ Y. The proof will be carried out by proving the following three assertions. Y (1) There exists δ > 0 such that Bδα ⊂ Eα for all α > 0. Y (2) For the same δ > 0, Bδα ⊂ Eα , i.e. we may remove the closure in assertion 1. (3) The last assertion implies T is an open mapping. ∞ S En , the Baire category Theorem 15.16 implies there exists 1. Since Y = n−1

n such that E n 6= ∅, i.e. there exists y ∈ E n and > 0 such that B Y (y, ) ⊂ E n . Suppose ky 0 k < then y and y + y 0 are in B Y (y, ) ⊂ E n hence there exists X x0 , x ∈ Bn such that kT x0 − (y + y 0 )k and kT x − yk may be made as small as we please, which we abbreviate as follows kT x0 − (y + y 0 )k ≈ 0 and kT x − yk ≈ 0.

0

Hence by the triangle inequality, kT (x0 − x) − y 0 k = kT x0 − (y + y 0 ) − (T x − y)k ≤ kT x0 − (y + y 0 )k + kT x − yk ≈ 0

X with x0 − x ∈ B2n . This shows that y 0 ∈ E2n which implies B Y (0, ) ⊂ E2n . Since α the map φα : Y → Y given by φα (y) = 2n y is a homeomorphism, φα (E2n ) = Eα α Y Y Y and φα (B (0, )) = B (0, 2n ), it follows that Bδα ⊂ Eα where δ ≡ 2n > 0. Y 2. Let δ be as in assertion 1., y ∈ Bδ and α1 ∈ (kyk /δ, 1). Choose {αn }∞ ⊂ n=2 ¡ ¢ P Y X (0, ∞) such that ∞ αn < 1. Since y ∈ Bα1 δ ⊂ Eα1 = T Bα1 by assertion 1. n=1 X there exists x1 ∈ Bα1 such that ky − T x1 k < α2 δ. (Notice that ky − T x1 k can be ¡ ¢ Y X ¯ made as small as we please.) Similarly, since y−T x1 ∈ Bα2 δ ⊂ Eα2 = T Bα2 there X exists x2 ∈ Bα2 such that ky − T x1 − T x2 k < α3 δ. Continuing this way inductively, X there exists xn ∈ Bαn such that

(16.1) Since

∞ P

ky − kxn k <

∞ P

k=1

n X

T xk k < αn+1 δ for all n ∈ N.

∞ P X xn exists and kxk < 1, i.e. x ∈ B1 . Passing

n=1

n=1

αn < 1, x ≡

n=1

X to the limit in Eq. (16.1) shows, ky − T xk = 0 and hence y ∈ T (B1 ) = E1 . X Therefore we have shown Bδ ⊂ E1 . The same scaling argument as above then X shows Bαδ ⊂ Eα for all α > 0. 3. If x ∈ V ⊂o X and y = T x ∈ T V we must show that T V contains a ball B Y (y, ) = T x + B Y for some > 0. Now B Y (y, ) = T x + B Y ⊂ T V iﬀ B Y ⊂ T V − T x = T (V − x). Since V − x is a neighborhood of 0 ∈ X, there exists X Y X α > 0 such that Bα ⊂ (V − x) and hence by assertion 2., Bαδ ⊂ T Bα ⊂ T (V − x) Y and therefore B (y, ) ⊂ T V with := αδ.

e. m) is a closed subspace such that each element of M has a representative in C([0. is bounded.312 BRUCE K. This shows that T x = y and proves that T is closed. DRIVER † Corollary 16. being the composition of bounded operators. Proof. Each proof requires that we ﬁrst show that (M.3 (Closed Graph Theorem). gx i for all f ∈ M. 1) I will give a two proofs of part 1. Suppose that M ⊂ L2 ([0. Suppose that T : H → H is a linear (not necessarily bounded) map such that there exists T ∗ : H → H such that Then T is bounded. y ∈ H. Example 16. Let X and Y be Banach space T : X → Y linear is continuous iﬀ T is closed i. Here is another example. Y i = hx. 1]). If T continuous and (xn . 1]) such that kfn − f k∞ → 0 as n → ∞. k · k∞ ) is a complete space. Proof. zi = hxn . y) = (x. X). zi as n → ∞. We will abuse notation and simply write M ⊂ C([0. a bijective linear transformation) then the inverse map. Proposition 16. i. T x). and since M is closed in L2 ([0. Then for any z ∈ H. T ∗ zi = hT x. Γ(T Hence T = π2 ◦ π1 |−1 ) is bounded. n → ∞. T −1 . So let {fn } ⊂ M and f ∈ C([0. Then kfn − fm kL2 ≤ kfn − fm k∞ → 0 as m. hT x. Γ(T ) ⊂ X × Y is closed. (Note that T −1 is automatically linear. y) ∈ H × H. T xn ) → (x. 1] there exists gx ∈ M such that Moreover we have kgx k ≤ A. y) ∈ X × Y as n → ∞ then T xn → T x = y which implies (x. 1]). Y are Banach spaces and T ∈ L(X. hT xn . To prove this suppose that xn ∈ H such that (xn . Conversely: If T is closed then the following diagram commutes Γ(T ) µ ¡ @ ¡ @ π2 Γ¡ @ R @ ¡ -Y X T where Γ(x) := (x. ∞) such that kf k∞ ≤ Akf kL2 for all f ∈ M. Let H be a Hilbert space. Proof. T xn ) → (x. 1]. T ∗ zi −→ hx. 1]). Y ) is invertible (i. T −1 ∈ L(Y. By passing to a f (x) = hf.4. zi = hy. (2) For all x ∈ [0. Γ(T As an application we have the following proposition. On the other hand limn→∞ hT xn . . The map π2 : X × Y → X is continuous and π1 |Γ(T ) : Γ(T ) → X is continuous bijection which implies π1 |−1 ) is bounded by the open mapping Theorem 16. Then (1) There exists A ∈ (0. If X. It suﬃces to show T is closed. T x) ∈ Γ(T ). (3) The subspace M is ﬁnite dimensional and dim(M) ≤ A2 . T ∗ Y i ∀ x.e. zi for all z ∈ H. zi = hy. L2 − limn→∞ fn = g ∈ M. zi as well and therefore hT x. 1]). To prove this it suﬃces to show M is a closed subspace of C([0.5.e.1.2.) Theorem 16.

1]. ii) Let j : (M.5. 2) For x ∈ [0. ∞) and W is a closed subspace of Lp (µ) such that W ⊂ Lp (µ)∩L∞ (µ). then kf k∞ ≤ A kf kp ≤ A kf k2 and if p ∈ (2.a. i)Let i : (M. With out loss of generality we may assume that µ(X) = 1. we conclude that g = f = j(f ) a. j = i−1 is bounded as well. k · k∞ ) be the identity map. M. k · k∞ ) → (M. Remark 16. ∞). gx i for all f ∈ M. Proof. The above example generalizes as follows. 0 The function G is called the reproducing kernel for M. we shows that W is a closed subspace of L∞ (µ) and hence by the open mapping theorem. As in Example 16.6. So f = g ∈ M. Then i is bounded and bijective. Hence there exists a unique element gx ∈ M such that f (x) = ex (f ) = hf. showing j is closed. gx i|2 = n X j=1 |fj (x)|2 and integrating this equation over x ∈ [0.e. µ) is a ﬁnite measure space. Then dim(W ) < ∞. Proposition 16. y) = gx (y) for all x. 1]). there exists a constant A < ∞ such that kf k∞ ≤ A kf kp for all f ∈ W. ³ ´1−2/p 2/p ≤ kf k2 A kf kp . Suppose that fn ∈ M such that fn → f in L2 and fn = j(fn ) → g in C([0. then kf kp ≤ kf k2 kf k∞ kf kp ≤ kf k2 2/p p 2 p−2 kf k∞ 1−2/p or equivalently. Then |ex (f )| ≤ |f (x)| ≤ kf k∞ ≤ Akf kL2 which shows that ex ∈ M∗ . By the open mapping theorem. let ex : M → C be the evaluation map ex (f ) = f (x). 1]. Then Z 1 f (x) = ex (f ) = f (y)G(x. Moreover kgx kL2 = kex kM∗ ≤ A. 3) Let {fj }n be an L2 — orthonormal subset of M. x. Now if 1 ≤ p ≤ 2. Then j=1 A2 ≥ kex k2 ∗ = kgx k2 2 ≥ M L n X j=1 |hfj . k · k2 ) → (M. 1] implies that n n XZ 1 X A2 ≥ |fj (x)|2 dx = 1=n j=1 0 j=1 which shows that n ≤ A2 . We will shows that j is a closed operator and hence bounded by the closed graph theorem. Now ﬁnish as in last line of proof i).REAL ANALYSIS LECTURE NOTES 313 subsequence if necessary we know that g(x) = limn→∞ fn (x) = f (x) for m .e. Suppose that (X. p ∈ [1. y)dy for all f ∈ M. Then as in the ﬁrst paragraph. G(x. Hence dim(M) ≤ A2 .5. y ∈ [0. k · k2 ) be the identity map.7. Keeping the notation in Example 16. Hence there exists A < ∞ such that kf k∞ = kj(f )k ≤ A kf k2 for all f ∈ M.

F is not meager iﬀ sup kAk < ∞. DRIVER † Let / ¯P D := (Q ¯+ iQ) and E = ∩c∈D Ec . A∈A A∈A (4) If X is a Banach space. dim(W ) ≤ B 2 . then sup kAxk ≤ M kxk < ∞ for all x ∈ X A∈A A∈A showing F = X. i. (2) If F is not meager. P Let {fn }N be an orthonormal subset of W and f = N cn fn with cn ∈ C. If M := sup kAk < ∞. ¯ N ¯ 2 2 / ¯ n=1 cn fn (x)¯ ≤ B |c| for all c ∈ D. By continuity it then follows for x ∈ E that ¯2 ¯N ¯ ¯X ¯ ¯ 2 cn fn (x)¯ ≤ B 2 |c| for all c ∈ CN . Then µ(E) = 0 and for x ∈ E. . 1. In particular if sup kAk = ∞ then sup kAxk = ∞ for x in a dense subset of X.e. Let X and Y be a normed vector spaces.314 BRUCE K. ¯ ¯ ¯ n=1 N and therefore that Taking cn = fn (x) in this inequality implies that ¯N ¯2 N ¯X ¯ X ¯ 2¯ |fn (x)| ¯ ≤ B 2 |fn (x)|2 for all x ∈ E / ¯ ¯ ¯ n=1 n=1 n=1 N X |fn (x)| ≤ B 2 for all x ∈ E. Y ) be a collection of bounded linear operators from X to Y. then sup kAk < ∞. A∈A A∈A Proof. For each c ∈ C . A ⊂ L(X.2) R = RA = F c = {x ∈ X : sup kAxk = ∞}. In particular if A∈A A∈A sup kAxk < ∞ for all x ∈ X then sup kAk < ∞. then sup kAk = ∞ iﬀ R is residual. ¯ ¯ ¯ n=1 from which we learn that kf kp ≤ A1−2/p kf k2 and therefore that kf k∞ ≤ AA1−2/p kf k2 so that in any case there exists a constant B < ∞ such that kf k∞ ≤ B kf k2 . Theorem 16.8 (Uniform Boundedness Principle). n=1 n=1 then °2 °N N ° °X X ° ° cn fn ° ≤ B 2 |cn |2 ≤ B 2 |c|2 ° ° ° n=1 n=1 ∞ PN 2 2 N where |c| := n=1 |cn | . / 2 Integrating this equation over x then implies that N ≤ B 2 . A∈A (3) If X is a Banach space. A∈A A∈A (1) If sup kAk < ∞ then F = X. F = FA = {x ∈ X : sup kAxk < ∞} and A∈A (16. / ¯2 ¯ N ¯ ¯X ¯ ¯ cn fn (x)¯ ≤ B 2 |c|2 . there is an exception set Ec such that for x ∈ Ec .

Then |fN (a)| ≤ MN kak 1 Then c ∈ ∞ .REAL ANALYSIS LECTURE NOTES 315 2. so R is residual iﬀ sup kAk = ∞.e. . For each n ∈ N. Since R = F c . F = X is not meager by the Baire Category Theorem 15. Let Bx (δ) be a ball such that Bx (δ) ⊂ En .8. we learn that kfN k = MN . . is equivalent to F is meager iﬀ sup kAk = ∞. then Z Z Z π 1 f (w)dw := f dm := f (eiθ )dθ. 3. i. let En ⊂ X be the closed sets given by \ En = {x : sup kAxk ≤ n} = {x : kAxk ≤ n}. Suppose that lim N X {cn }∞ n=1 ⊂ C is a sequence of numbers such that 1 N →∞ n=1 an cn exists in C for all a ∈ . ∞) is measurable. . Applications to Fourier Series. and by taking a = ek with k such MN = |ck | . 2π −π T T . So item 3.1. A∈A 4.9. N N 16. A∈A A∈A Then F = ∪∞ En which is assumed to be non-meager and hence there exists n=1 an n ∈ N such that En has non-empty interior. } . ¡ ¢∗ PN Proof. N } . Then for y ∈ X with kyk = δ we know x − y ∈ Bx (δ) ⊂ En . A∈A 3. Let T = S 1 be the unit circle in S 1 and m denote the normalized arc length measure on T. . kAyk ≤ kAxk + kA(x − y)k ≤ n + n = 2n. limN →∞ fN (a) exists for all a ∈ 1 and in particular. So by the Theorem 16. Now by assumption. follows from items 1. ∞ > sup kfN k = sup MN = sup {|cn | : n = 1. Example 16.16. . If X is a Banach space. So if f : T → [0. so that Ay = Ax − A(x − y) and hence for any A ∈ A. and 2 and the fact that F = X iﬀ sup kAxk < ∞ for all x ∈ X. sup kAk ≤ 2n/δ < ∞. R is A∈A A∈A residual iﬀ F is meager. 2. sup |fN (a)| < ∞ for all a ∈ N 1 . . Hence it follows that kAk ≤ 2n/δ for all A ∈ A. Let fN ∈ 1 be given by fN (a) = n=1 an cn and set MN := max {|cn | : n = 1. . Item 3.

¯ . there exists fk ∈ C(T. Recall that C(T ) is a complete metric space. R) such that fk (w) → sgndk (w) ¯ in L1 . so that dn (α) := k=−n n X αk = αn+1 − α−n α−1 with the convention that n X αn+1 − α−n αn+1 − α−n 1k . φn iφk (z) = hf. For n ∈ N let ¶ n n n X X X µZ k k hf. z)| = ∞ for all f ∈ Rz . DRIVER † Also let φn (z) = z n for all n ∈ Z. φn iz = f (w)w dw z k ¯ sn (f. L2 (T ) — limn→∞ sn (f. Proof.10. 2π 2π T −π −π 2 Z T |dn (w)| dw. Let Λn f := sn (f. For each z ∈ T. z) := k=−n = Z f (w) T k=−n Ã k=−n where dn (α) := Pn n X wk z k ¯ ! k=−n k=−n T dw = Z f (w)dn (z w)dw ¯ T αk . We will now show that the convergence is not pointwise for all f ∈ C(T ) ⊂ L2 (T ). we ﬁnd Dn (θ) := dn (eiθ ) = eiθ(n+1) − e−iθn eiθ(n+1/2) − e−iθ(n+1/2) = iθ − 1 e eiθ/2 − e−iθ/2 1 sin(n + 2 )θ = . Now αdn (α) − dn (α) = αn+1 − α−n . By symmetry considerations. By replacing fk by (fk ∧ 1) ∨ (−1) we may assume that kfk k∞ ≤ 1. Recall that {φn }n∈Z is an orthonormal basis for L2 (T ). it suﬃces to take z = 1 ∈ T. It now follows that ¯Z ¯ ¯ |Λn fk | ¯ ¯ fk (w)dn (w)dw¯ kΛn k ≥ ≥ ¯ ¯ kfk k∞ ¯ T R and passing to the limit as k → ∞ implies that kΛn k ≥ T |dn (w)| dw. ·) = f for all f ∈ L2 (T ). |α=1 = lim = 2n + 1 = α→1 α−1 α−1 k=−n Writing α = e .3) kΛn k = |dn (w)| dw = ¯ ¯ sin 1 θ ¯ dθ. Hence we ¯ have shown that ¯ Z Z π Z π ¯ ¯ sin(n + 1 )θ ¯ ¯ ¯ 1 2 ¯ ¯ ¯dn (e−iθ )¯ dθ = 1 (16. hence Rz is a dense subset of C(T ). Then ¯Z ¯ Z Z ¯ ¯ ¯ |f (w)dn (w)| dw ≤ kf k∞ ¯ |dn (w)| dw ¯ |Λn f | = ¯ f (w)dn (w)dw¯ ≤ ¯ ¯ T T T showing kΛn k ≤ Since C(T ) is dense in L1 (T ). 1). sin 1 θ 2 iθ Recall by Hilbert space theory. Proposition 16. there exists a residual set Rz ⊂ C(T ) such that supn |sn (f.316 BRUCE K.

we have °f ° ≤ kf k showing that Λf := f is a bounded 1 c0 1 ¯ ¯ ¯ ¯ ¯ˆ ¯ ¯ˆ ¯ ˆ lim sup ¯f (n)¯ = lim sup ¯f (n) − g (n)¯ ≤ kf − gk1 n→∞ n→∞ for all g ∈ C(T ). then T f (w)p(w. w)dw = 0 ¯ for all polynomials p in w and w. n ¯ ¯ Z (n+ 1 )π 2 ¯ ¯ dy ¯sin(n + 1 )θ¯ dθ = |sin y| → ∞ as n → ∞. R ˆ To see that Λ is injective. ¯ ¯ P ¯ ˆ ¯2 Proof.e. Since sin x is odd. Lemma 9. suppose f = Λf ≡ 0.8.REAL ANALYSIS LECTURE NOTES 317 Since sin x = Since for all x ≥ 0. |sin x| ≤ |x| for all x ∈ R. φn i = Z f (w)wn dw. Since L2 (T ) is dense in L1 (T ). Using this in Eq. . let ˆ f (n) := hf. ¯f (n)¯ < ∞ for all f ∈ L2 (T ) and in n∈Z ¯ ¯ ¯ˆ ¯ particular lim|n|→∞ ¯f (n)¯ = 0. (16. it follows that lim supn→∞ ¯f (n)¯ = 1 ˆ 0 for all f ∈ L¯ . ¯ ¯ θ 2 y 0 R1 = {f ∈ C(T ) : sup |Λn f | = ∞} is a residual set. Lemma 16. f ∈ c0 . By the Stone . ¯ ° ° ¯ˆ ¯ ° ˆ° ˆ Since ¯f (n)¯ ≤ kf k . ¯ T ˆ Then f ∈ c0 and the map f ∈ L1 (T ) → c0 is a one to one bounded linear transformation into but not onto c0 .Wierestrass and the dominated ¯ convergence theorem. For f ∈ L1 (T ). kΛn k ≥ 1 ¯ ¯ ¯ 2π −π π 0 2 ¯ θ 2θ ¯ Z π¯ ¯ ¯ 2 ¯sin(n + 1 )θ¯ dθ = ¯ π 0 2 ¯ θ Z π 0 Z x 0 cos ydy ≤ Z x 0 |cos y| dy ≤ x we learn that supn kΛn k = ∞. By Bessel’s inequality. Given f ∈ L1 (T ) and g ∈ L2 (T ) we have ¯ ¯ ¯ ¯Z ¯ ¯ˆ ¯ ¯ ¯ [f (w) − g(w)] wn dw¯ ≤ kf − gk ˆ ¯ ¯f (n) − g (n)¯ = ¯ 1 ¯ T and hence linear transformation from L1 (T ) to c0 . So by Theorem 16.3) implies that ¯ ¯ Z π ¯ Z π¯ ¯ sin(n + 1 )θ ¯ ¯ ¯ 1 2 ¯ dθ = 2 ¯ ¯sin(n + 1 )θ¯ dθ . i. this implies that Z f (w)g(w)dw = 0 T ¯ ¯ ¯ˆ ¯ for all g ∈ L2 (T ).7 now implies f = 0 a.e. See Rudin Chapter 5 for more details.11.

0= kxk=1 sup |u(x)| and this supremum is attained at x0 ∈ X with kx0 k = 1.This explains why kf k = kuk. ¯ µ ¶ d¯ x0 + itx0 ¯ u dt ¯0 kx0 + itx0 k ¯ ½ ¾ 1 d¯ ¯ (u(x0 ) + tu(ix0 )) = u(ix0 ) = dt ¯ |1 + it| 0 . this shows that kf k ≤ kuk so kf k = kuk. Therefore ran(Λ) 6= c0 . c0 ° ° °ˆ ° Taking f = dn . Hahn Banach Theorem. while f ((a + ib)x) = u(ax + ibx) − iu(iax − bx) = au(x) + bu(ix) − i(au(ix) − bu(x)) (a + ib)f (x) = au(x) + bu(ix) + i(bu(x) − au(ix)). |f | ≤ p iﬀ u ≤ p. In particular this implies there exists C < ∞ such that ° ° ° ˆ° (16. then v(ix) = Im f (ix) = Im(if (x)) = Ref (x) = u(x). the open mapping theorem would imply that Λ−1 : c0 → L (T ) is bounded. To understand better why kf k = kuk. then ∗ XR f (x) = u(x) − iu(ix).12.34 34 |f (x)| = f (λx) = u(λx) ≤ kuk kλxk = kukkxk.4) kf kL1 ≤ C °f ° for all f ∈ L1 (T ). More generally if p is a semi-norm on X. Therefore ∗ Conversely for u ∈ XR let f (x) = u(x) − iu(ix). For x ∈ X choose λ ∈ S 1 ⊂ C such that |f (x)| = λf (x) so Since x ∈ X is arbitrary. Let v(x) = Im f (x). This will be the content of the Hahn — Banach Theorem 16. Our next goal is to show that continuous dual X ∗ of a Banach space X is always large. 1 c0 16. Proposition 16. DRIVER † (16.318 BRUCE K. Because |u(x)| = |Ref (x)| ≤ |f (x)|. u(x0 ) = M. we ﬁnd °dn ° = 1 while limn→∞ kdn kL1 = ∞ contradicting Eq. Proof. we may assume that x0 . Then f (x) = u(x) + iv(x) = u(x) + iu(−ix) = u(x) − iu(ix). If Λ were surjective. If f ∈ X ∗ and ∗ u = Ref ∈ XR then (16. Since u has a maximum at since d | |1 dt 0 + it| = d | dt 0 √ 1 + t2 = 0. it follows that kuk ≤ kf k.5). Supppose that M = Replacing x0 by −x0 if necessary.4). notice that kxk=1 kxk=1 kf k2 = sup |f (x)|2 = sup (|u(x)|2 + |u(ix)|2 ). So f is complex linear.15 below.2. then f ∈ X ∗ and kukXR = kf kX ∗ . Let X be a complex vector space over C.5) ∗ Conversely if u ∈ and f is deﬁned by Eq. Proof. (16.

for example. b ∈ [−1.15 (Hahn-Banach). 2] + µ[−1. 2] = [−λ. Moreover. To check the last set inclusion let a. then µ ¶ λ µ λa + µb = (λ + µ) a+ b ∈ (λ + µ) [−1. i. Proof. 2] is a convex set and λ λ+µ x + y ∈ λ[−1. 1.REAL ANALYSIS LECTURE NOTES 319 For the last assertion. From this formula it is clear that p(cx) = cp(x) for all c ≥ 0 but not for c < 0. Step (1) We show for all x ∈ X \ M there exists and extension F to M ⊕ Rx with the desired properties. choose λ ∈ S 1 ⊂ C such that |f (x)| = λf (x). Equivalently put we must ﬁnd α ∈ R such that p(y + λx) − f (y) α≤ for all y ∈ M and λ > 0 λ p(z − µx) − f (z) α≥ for all z ∈ M and µ > 0. TODO: Add in the relationship to convex sets and separation theorems. µ So if α ∈ R is going to exist. indeed if p(x) = λ and p(y) = µ. 2λ]} . M ⊂ X be a subspace f : M → R be a linear functional such that f ≤ p on M . Let X be a real vector space. p satisﬁes the triangle inequality. |f (x)| = λf (x) = f (λx) = u(λx) ≤ p(λx) = p(x) Deﬁnition 16. p : X → R is a Minkowski functional if (1) p(x + y) ≤ p(x) + p(y) for all x. µ > 0 that p(y + λx) − f (y) f (z) − p(z − µx) ≤ µ λ or equivalently (16. Conversely if u ≤ p and x ∈ X.14. Then there exists a linear functional F : X → R such that F |M = f and F ≤ p. If F exists and α = F (x). then p(x) = x/2 and if x ≤ 0 then p(x) = −x.e. we have to prove. p(x) = inf {λ ≥ 0 : x ∈ λ[−1. for all y. λα ≤ p(y+λx)−f (y). 2]. .13 (Minkowski functional). 2] + µ λ+µ = 1. 2] and y ∈ µ[−1.e. it is clear that |f | ≤ p implies that u ≤ |u| ≤ |f | ≤ p. Suppose that X = R and Notice that if x ≥ 0. Theorem 16. Then holds for all x ∈ X. 2] so that which shows that p(x + y) ≤ λ + µ = p(x) + p(y).6) f (λz + µy) ≤ µp(y + λx) + λp(z − µx) = p(µy + µλx) + p(λz − λµx). z ∈ M and λ. then x ∈ λ[−1. 2] ⊂ (λ + µ) [−1. y ∈ X and (2) p(cx) = cp(x) for all c ≥ 0 and x ∈ X. ½ x/2 if x ≥ 0 p(x) = |x| if x ≤ 0. Example 16. 2] λ+µ λ+µ since [−1. see Reed and Simon Vol. then for all y ∈ M and λ ∈ R we must have f (y)+λα = F (y+λx) ≤ p(y+λx) i.

Step (2) Let us now write F : X → R to mean F is deﬁned on a linear subspace D(F ) ⊂ X and F : D(F ) → R is linear. Now applying Theorem 16. violating the maximality of F. Therefore F (y + λx) := f (y) + λα is the desired extension. Since isometries are necessarily ˆ injective.18. Theorem 16. Deﬁne F (x) = U (x) − iU (ix) then as in Proposition 16. (16. Then there exists F ∈ X 0 (X 0 is the algebraic dual of X) such that F |M = f and |F | ≤ p.e. f (x) = δ = d(x. Let u = Ref then u ≤ p on M and hence by Theorem 16. DRIVER † But f (λz + µy) = f (λz + µλx) + f (λz − λµx) ≤ p(λz + µλx) + p(λz − λµx) which shows that Eq. kF k ≤ 1.) Proof. We may now apply Zorn’s Lemma (see Theorem B. ∞) is a semi-norm.e.16. M ) and f = 0 on M .6) is true and by working backwards. since |f (m + x)| = δ = inf kx + mk. there exists 0 U ∈ XR such that U |M = u and U ≤ p on M . it follows that x kˆkX ∗∗ ≤ kxkX . Suppose that X is a complex vector space. Corollary 16. Then it is easily checked that D(G) is a linear subspace. Corollary 16. (This isometry need not be surjective. D(F ) = X for otherwise we could extend F by step (1). Proof. Notice that km + λxk = |λ|kx + m/λk ≥ |λ|δ and hence |f (m + λx)| = |λ|δ ≤ km + λxk which shows kf k ≤ 1. The linear map x ∈ X → x ∈ X ∗∗ where x(f ) = f (x) for all ˆ ˆ x ∈ X is an isometry. kf k = 1. F ≤ p on D(F )}. Deﬁne f : M ⊕ Cx → C by f (m + λx) ≡ λδ for all m ∈ M and λ ∈ C. For F. F = f on M and |F | ≤ p. a linearly ordered S subset of F) then Φ has an upper bound G ∈ F deﬁned by D(G) = D(F ) F ∈Φ and G(x) = F (x) for x ∈ D(F ). Since 1 = kf k ≤ kF k ≤ 1 we see kF k = kf k. and F < G for all F ∈ Φ. we are done.12. If Φ ⊂ F is a chain (i.15. G ∈ F. Proof. Let X be a normed space M ⊂ X be a closed subspace and x ∈ X \ M . x x This shows that x ∈ X → x ∈ X ∗∗ is an isometry. that is G is an extension of F.17 with M = {0} . which shows that kˆkX ∗∗ ≥ kxkX . . p : X → [0. Thus F is the desired extension of f.7) to conclude there exists a maximal element F ∈ F. Let F = {F : X → R : M ⊂ D(F ). <) is a partially ordered set. Then there exists f ∈ X ∗ such that kf k = 1.320 BRUCE K.17. and f : M → C is linear functional such that |f (x)| ≤ p(x) for all x ∈ M. i. Then (F. G : X → R we will say F < G if D(F ) ⊂ D(G) and F = G|D(F ) . there exists f ∈ X ∗ x such that kf k = 1 and |ˆ(f )| = f (x) = kxk . By m∈M Hahn-Banach theorem there exists F ∈ X ∗ such that F |M ⊕Cx = f and |F (x)| ≤ kxk for all x ∈ X. Necessarily. M ⊂ X is a linear subspace. In fact. Since |ˆ(f )| = |f (x)| ≤ kf kX ∗ kxkX for all f ∈ X ∗ . there exist an α ∈ R such that f (y) + λα ≤ p(y + λx).

Y ) −→ T x ∈ Y is continuous for all x ∈ X. where xi ∈ X and > 0 forms a neighborhood base for the weak—∗ topology on X ∗ at f ∈ X ∗ . m). m))∗ À L1 (R.20. Choosing µ = δx for some x ∈ (0. Y ) is the smallest topology such that T ∈ L(X.1] f dµ for all µ ∈ X ∗ . 1]. i.22 (Following Riesz and Nagy. M). 214). Suppose that µ is a σ — ﬁnite measure on a measurable space (X. Deﬁnition 16.1] showing f would have to be the constant function. A Banach space X is reﬂexive if the map x ∈ X → x ∈ X ∗∗ is ˆ surjective.1] X µ({x}) = x∈[0.24. [0. Proposition 10. X ∗ ∼ L∞ (R.REAL ANALYSIS LECTURE NOTES 321 Deﬁnition 16. (1) The weak topology on X is the topology generated by X ∗ . N ≡ ∩n {g ∈ X ∗ : |f (xi ) − g(xi )| < } i=1 Theorem 16. ∞). Y ) the normed space of bounded linear transformations from X to Y.e. The Banach space X := L1 (R. The argument in Example 13. sets of the form N = ∩n {x ∈ X : |fi (x) − fi (x0 )| < } i=1 ∗ where fi ∈ X and > 0 form a neighborhood base for the weak topology on X at x0 . m) is not reﬂexive.1. To prove this recall that X ∗ may be identiﬁed with complex measures µ on [0. Weak and Strong Topologies. see Theorem 13. Example 16. namely Z Z 1 F → µF → (f ∈ X → f dµF = f dF ).15 shows = (L∞ (R. (2) The weak-∗ topology on X ∗ is the topology generated by X. which clearly can not work. then Lp (X. m). p. Y ) is the smallest topology such that T ∈ L(X. (3) The strong operator topology on L(X.21. 1] which may be identiﬁed with right continuous functions of bounded variation (F ) on [0.14. 1) would then imply that Z f (x) = f δx = λ(δx ) = 1 [0. The Banach space X := C([0.23.14. 1]) is not reﬂexive. Example 16. Suppose there existed an f ∈ X such that is λ(µ) = [0.∗ compact.1] 0 Deﬁne λ ∈ X ∗∗ by λ(µ) = x∈[0. Example 16.19.e. so λ(µ) R the sum of the “atoms” of µ. i. .1] X (F (x) − F (x−)) . Every Hilbert space H is reﬂexive.25 (Alaoglu’s Theorem). Y ) −→ f (T x) ∈ C is continuous for all x ∈ X and f ∈ Y ∗ . This is a consequence of the Riesz Theorem. Example 16. If X is a normed space the unit ball in X ∗ is weak . M. As we have seen in Theorem 13.3. (4) The weak operator topology on L(X.15. Let X and Y be be a normed vector spaces and L(X. µ) is reﬂexive for all p ∈ (1. 16.

B ⊂ Ω. Hence τC ∗ = τ (ˆ|C ∗ : x ∈ X) ⊂ τρ . Suppose that X is a separable Banach space. For all x ∈ X let Dx = {z ∈ C : |z| ≤ kxk}. it suﬃces to show x is τρ — continuous for all x ∈ C. f ∈C ∗ yk → x uniformly on C ∗ and using yk is τρ — continuous for all k (as is easily ˆ ˆ ˆ checked) we learn x is also τρ continuous. DRIVER † topology. If f ∈ B := {f ∈ X ∗ : kf k ≤ 1}.4. i. Supplement: Quotient spaces. the map f ∈ X ∗ → (f (xn ) − g(xn )) ∈C is τw∗ continuous and since the sum in Eq. The transpose of A is the linear operator A† : Y ∗ → X ∗ deﬁned by . Let X and Y be Banach spaces and A : X → Y be a linear operator. n=1 Since f ∈C ∗ x ˆ sup |ˆ(f ) − yk (f )| = sup |f (x − yk )| ≤ kx − yk k → 0 as k → ∞. g) is τC ∗ — continuous.27. Since τC ∗ is the topology generated by {ˆ|C ∗ : x ∈ C} . it follows that f → ρ(f.y∈X c∈C Proof. Then n=1 (16. Then Dx ⊂ C is a Q compact set and so by Tychonoﬀ’s Theorem Ω ≡ Dx is compact in the product x∈X = x.e. τC ∗ := (τw∗ )C ∗ = {V ∩ C : V ∈ τw∗ } .35.7) is uniformly convergent for f ∈ C ∗ . Deﬁnition 16. We now wish to prove τC ∗ ⊂ τρ .26 (Alaoglu’s Theorem for separable spaces). |f (x)| ≤ kf k kxk ≤ kxk which implies ¯ that f (x) ∈ Dx for all x ∈ X. 16. Proof. and more reﬂexivity. The routine check that ρ is a metric is left to the reader. The topology on B inherited from ∗ the weak—∗ topology on X is the same as that relative topology coming from the ¯ product topology on Ω. adjoints. To prove this let πx (f ) = f (x) be the projection maps. ρ) is a compact metric space. But given x ∈ C x ˆ there exists a subsequence yk := xnk of {xn }∞ such that such that x = limk→∞ yk . ˆ x The compactness assertion follows from Theorem 16.25. Theorem 16.y∈X c∈C \ \ (πx+cy − πx − cπy ) −1 ({0}) which is closed because (πx+cy − πx − cπy ) : Ω → C is continuous. (16.322 BRUCE K. g) := ∞ X 1 |f (xn ) − g(xn )| 2n n=1 deﬁnes a metric on C ∗ which is compatible with the weak topology on C ∗ . Let τρ be the topology on C ∗ induced by ρ. This implies the open balls relative to ρ are contained in τC ∗ and therefore τρ ⊂ τC ∗ . C ∗ := {f ∈ X ∗ : kf k ≤ 1} is the closed unit ball in X ∗ and {xn }∞ is an countable dense subset of C := {x ∈ X : kxk ≤ 1} . For any g ∈ X and n ∈ N.7) ρ(f. Then B = {f ∈ Ω : f is linear} = {f ∈ Ω : f (x + cy) − f (x) − cf (y) = 0 for all x. So to ﬁnish the proof it suﬃces to show B is a closed subset of the compact space Ω. Moreover (C ∗ . y ∈ X and c ∈ C} \ \ = {f ∈ Ω : f (x + cy) − f (x) − cf (y) = 0} x. The compactness assertion may also be veriﬁed directly using: 1) sequential compactness is equivalent to compactness for metric spaces and 2) a Cantor’s diagonalization argument as in the proof of Theorem 10.

then f (x) = 0 for all f ∈ M 0 so that x ∈ M 0 . i. then there exists f ∈ X ∗ such that f |M = 0 while / ¯ ¡ ¢⊥ f (x) 6= 0. ° ° c Proposition 16. c ˆ (6) X is reﬂexive iﬀ X ∗ is reﬂexive. . If x ∈ M . f (x) = ψ(f ) 6= 0 so ¡ ⊥ ¢0 ¢0 ¡ ¯ f∈ N / . (4) N (5) nul(A) = ran(A† )⊥ and nul(A† ) = ran(A)0 . (1) kAk = sup kAxk = sup sup |f (Ax)| kxk=1 kxk=1 kf k=1 kf k=1 kxk=1 kf k=1 M 0 := {f ∈ X ∗ : f |M = 0} and (2) This is an easy consequence of the assumed continuity oﬀ all linear functionals involved.28 (Basic Properties). Moreover. Therefore ¡ ¢⊥ ¯ M ⊂ M 0 . Proof. ¡ ¢⊥ ¯ (3) M 0 = M . ¡ ¢⊥ (3) If x ∈ M. (1) kAk = °A† ° and A†† x = Ax for ˆ all x ∈ X. f ∈ M 0 yet f (x) 6= 0. ¡ ¢ ¯ ⊂ N ⊥ 0 with equality when X is reﬂexive. For x ∈ X (so x ∈ X ∗∗ ) we have ˆ ˆ x x x ˆ ψ 0 (ˆ) = ψ(ˆ) − fψ (ˆ) = fψ (x) − x(fψ ) = fψ (x) − fψ (x) = 0. ¡ ¡ ¢0 ¢0 ¯ (4) It is again simple to show N ⊂ N ⊥ and therefore N ⊂ N ⊥ .e. as above if f ∈ N there exists ψ ∈ X ∗∗ such that ψ|N = 0 while ψ(f ) 6= 0. (5) nul(A) = {x ∈ X : Ax = 0} = {x ∈ X : f (Ax) = 0 ∀ f ∈ X ∗ } © ª = x ∈ X : A† f (x) = 0 ∀ f ∈ X ∗ ª © = x ∈ X : g(x) = 0 ∀ g ∈ ran(A† ) = ran(A† )⊥ . More generally X ∗∗∗ = X ∗ ⊕ X 0 . For M ⊂ X and N ⊂ X ∗ let N ⊥ := {x ∈ X : f (x) = 0 for all f ∈ N }. we have x ∈ N ⊥ . = {f ∈ Y ∗ : f (Ax) = 0 ∀ x ∈ X} ª © = f ∈ Y ∗ : f |ran(A) = 0 = ran(A)0 . (2) M 0 and N ⊥ are always closed subspace of X ∗ and X respectively. ¯ ° ° ° ¯ ° = sup sup ¯A† f (x)¯ = sup °A† f ° = °A† ° . ψ = x for some x ∈ X and since g(x) = ψ(g) = 0 for ˆ ¯ all g ∈ N . On the other hand.REAL ANALYSIS LECTURE NOTES 323 ¡ † ¢ A f (x) = f (Ax) for f ∈ Y ∗ and x ∈ X. ran(A) = nul(A† )⊥ and if X is reﬂexive. then ran(A† ) = nul(A)0 . / ¯ ¯ If X is reﬂexive. © ª © ª nul(A† ) = f ∈ Y ∗ : A† f = 0 = f ∈ Y ∗ : (A† f )(x) = 0 ∀ x ∈ X x (6) Let ψ ∈ X ∗∗∗ and deﬁne fψ ∈ X ∗ by fψ (x) = ψ(ˆ) for all x ∈ X and set ˆ ψ 0 := ψ − fψ . Similarly. The null space of A is the subspace nul(A) := {x ∈ X : Ax = 0} ⊂ X. Moreover. This shows x ∈ M 0 / and we have ¡ 0 ¢⊥ ¯ shown M ⊂ M. Thus again N ⊥ ⊂ N .

Given α ∈ (0. ˆx then ψ = f ˆ c ˆ i. n→∞ x + M = 0 ∈ X/M.324 BRUCE K. kcx + M k = inf kcx + mk = |c| inf kx + m/ck = |c| kx + M k m∈M because m/c runs through M as m runs through M. If Y is another normed space and T : X → Y is a bounded linear transformation such that M ⊂ nul(T ). f = 0 so ψ = 0. If ψ ∈ X ∗ ∩ X 0 . Let x1 . But as above nul(J † ) = ran(J)0 which will be zero iﬀ ran(J) = X ∗∗ and since J is an isometry this is equivalent to saying ran(J) = X ∗∗ . . X/M the quotient space. π : X → X/M the projection map π(x) = x + M for x ∈ X and deﬁne the quotient norm on X/M by kπ(x)kX/M = kx + M kX/M = inf kx + mkX . DRIVER † ˆ c ˆ c ˆ This shows ψ 0 ∈ X 0 and we have shown X ∗∗∗ = X ∗ + X 0 . k·kX/M ) is a Banach space. then m∈M Proof. If X is ˆ = X ∗∗ and so X 0 = {0} showing X ∗∗∗ = X ∗ . kx1 + x2 + M k ≤ kx1 + M k + kx2 + M k. kπk ≤ 1. let x ∈ X \ M so that π(x) 6= 0. (X/M. m2 ∈ M then Taking inﬁnums over m1 . m∈M Then (1) (2) (3) (4) k·kX/M is a norm on X/M. i. Notice that fψ = J † ψ. then there exists a unique linear transformation S : X/M → Y such that T = S ◦ π and moreover kT k = kSk .e. m2 ∈ M then implies kx1 + x2 + M k ≤ kx1 + x2 + m1 + m2 k ≤ kx1 + m1 k + kx2 + m2 k.29. x = lim mn ∈ M = M. kπk = 1. Therefore X ∗∗∗ = X ∗ ⊕ X 0 as claimed. Let X be a Banach space. and we have completed the proof the (X/M. so that X is reﬂexive. where J : X → X ∗∗ is given by Jx = x. there exists m ∈ M such that kx + mk ≤ α−1 kπ(x)k . ˆ for some f ∈ X ∗ and 0 = f (ˆ) = x(f ) = f (x) for all x ∈ X. Since x ∈ M. x2 ∈ X and m1 . 1) Clearly kx + M k ≥ 0 and if kx + M k = 0. then there exists mn ∈ M ¯ such that kx + mn k → 0 as n → ∞. k · k) is a normed space. 2) Since kπ(x)k = inf m∈M kx + mk ≤ kxk for all x ∈ X. So we have again shown X ∗ is reﬂexive iﬀ X is reﬂexive. To see kπk = 1. i. X ∗ ˆ c reﬂexive. and the composition ˆ x ∈ X.e. x ∈ X. 1). The projection map π : X → X/M has norm 1. Conversely if X is reﬂexive we conclude that X ³ ´⊥ ⊥ ˆ ˆ therefore X ∗∗ = {0} = X 0 = X. then X ∗ ˆ 0 = {0} and is reﬂexive.e. ˆ ˆ J ˆ f ∈ X ∗ → f ∈ X ∗∗∗ → J † f ∈ X ∗ ³ ´ ˆ ˆ ˆx is the identity map since J † f (x) = f (Jx) = f (ˆ) = x(f ) = f (x) for all ˆ ˆ † Alternative proof. If c ∈ C\ {0} . Thus it follows that X ∗ → X ∗∗∗ is invertible iﬀ J † is its inverse which can happen iﬀ nul(J † ) = {0} . Theorem 16. M ⊂ X be a proper closed subspace.

x := (xn +mn ) exists in X and therefore n=1 by the continuity of π. (1) The weak — ∗ topology on X ∗∗ is generated by o n ˆ f : f ∈ X ∗ = {ψ ∈ X ∗∗ → ψ(f ) : f ∈ X ∗ } . Since α ∈ (0. N := ∩n {ψ ∈ X ∗∗ : |ψ(fk ) − λ(fk )| < } k=1 n (16. (3) Letting C and C ∗∗ be the closed unit balls in X and X ∗∗ respectively. (2) A basic weak .8) and so the induced topology on X is precisely the weak topology. . . As above there P exists mn ∈ M such that kπ(xn )k ≥ 1 kxn + mn k and hence kxn + mn k ≤ 2 ∞ P P 2 kπ(xn )k < ∞. ˆ (2) X ⊂ X ∗∗ is dense in the weak-∗ topology on X ∗∗ . or equivalently so that ˆ |ˆ(fk ) − λ(fk )| = |fk (x) − λ(fk )| < x for k = 1. Proof.30. Let X be a Banach space. P 3) Let π(xn ) ∈ X/M be a sequence such that kπ(xn )k < ∞. kπ(x + m)k kπ(x)k α kx + mk = ≥ =α kx + mk kx + mk kx + mk which shows kπk ≥ α. the map x ∈ X → x ∈ X is a homeoˆ ˆ with the relative morphism when X is equipped with its weak topology and X topology coming from the weak-∗ topology on X ∗∗ . Then ˆ (1) Identifying X with X ⊂ X ∗∗ . the weak — ∗ topology on X ∗∗ induces the ˆ weak topology on X. x6=0 kxk x∈M kxk / Theorem 16. . π(x) = n=1 showing X/M is complete. then ˆ C := {ˆ ∈ C ∗∗ : x ∈ C} is dense in C ∗∗ in the weak — ∗ topology on X ∗∗. 1) is arbitrary we conclude that kπ(x)k = 1. We must now ﬁnd x ∈ X such that x ∈ N . 2. x (4) X is reﬂexive iﬀ C is weakly compact. . be given. Since X is complete.∗ neighborhood of a point λ ∈ X ∗∗ is of the form for some {fk }k=1 ⊂ X ∗ and > 0. Moreover kSk = kT k because and kT k = kS ◦ πk ≤ kSk kπk = kSk ∞ X π(xn + mn ) = n=1 ∞ X π(xn ) kSk = sup kS(π(x))k kT xk = sup kπ(x)k x∈M / x∈M kπ(x)k / kT xk kT xk ≥ sup = sup = kT k . More explicitly.9) . 4) The existence of S is guaranteed by the “factor theorem” from linear algebra. n. So the induced topology on X is generated by ˆ {x ∈ X → x ∈ X ∗∗ → x(f ) = f (x) : f ∈ X ∗ } = X ∗ ˆ (16.REAL ANALYSIS LECTURE NOTES 325 Therefore. .

8).) As in the proof of item 2. . Exercises. . . . n. . fn (x)) ∈ Cn is surjective (why) and hence there exists x ∈ X such that (16. . ª © ∗ ¯ ¯ Since f1 . . Working as before. . Hence ¯ ¯ |λ(fi ) − fi (y/α)| ≤ ¯fi (y) − α−1 fi (y)¯ ≤ (1 − α−1 ) |fi (y)| 16. It will be left to the reader to verify that it suﬃces again to assume {fk }n is a linearly independent set. it follows that C is weak — ∗ closed and so by item 3. fn is a basis for (X/N ) we ﬁnd ¯ ¯Pn Pn ¯ ¯ ¯ | i=1 αi λ(fi )| i=1 αi fi (π(x)) °Pn ° Pn kπ(x)k = sup = sup ¯° ° α∈Cn \{0} α∈Cn \{0} k i=1 αi fi k i=1 αi fi P P |λ( n αi fi )| k n αi fi k i=1 Pn Pi=1 = sup ≤ kλk sup = 1. Conversely ˆ if C is weakly compact. . n α∈Cn \{0} k α∈Cn \{0} k i=1 αi fi k i=1 αi fi k which can be arbitrarily small (i. . n. λ(fn )) = (f1 (y). C is weakly compact in X.10) holds.326 BRUCE K. If X is reﬂexive.∗ compact and hence by item 1. . Since the {fk }k=1 are linearly independent. To prove this stronger assertion we may. fn (x)) = (f1 (π(x)). fn (π(x))). . . (Hint: Suppose that k=1 {f1 . The problem is that x may not be in C. 2. . .∗ open neighborhood of λ as in Eq.. . . ˆ Hence we have shown kπ(x)k ≤ 1 and therefore for any α > 1 there exists y = x + n ∈ X such that kyk < α and (λ (f1 ) . . To remedy this. . . . . ˆ ˆ (4) Let C := {ˆ : x ∈ C} ⊂ C ∗∗ ⊂ X ∗∗ . 2. . there exists C ∗∗ = C x ∈ C such that x = λ/ kλk .10) (f1 (x). the map x ∈ X → (f1 (x). So if λ ∈ X ∗∗ . fm } . . . there exists x ∈ X such that Eq. . let N := ¯ ∩n nul(fk ) = nul(T ).e. . . . fm } were a maximal linearly dependent subset of {fk }n . λ(fn )) as desired. . . . assume that {fk }k=1 is a linearly inn dependent set. π : X → X/N ∼ Cn be the projection map and fk = k=1 ∗ ¯ ◦ π for k = 1. . .e. (16. . by discardn ing some of the fk ’s if necessary. Then we have ∈ (X/N ) be chosen so that fk = fk produced x ∈ X such that ¯ ¯ (λ (f1 ) . weak—∗ ˆ ˆ ˆ = C. λ/ kλk ∈ C ∗∗ = C. . This shows λ = (kλk x) and therefore ˆ ˆ X = X ∗∗ . .. Since the weak — ∗ topology on ˆ X ∗∗ is Hausdorﬀ. given > 0. we need to ﬁnd x ∈ C such that Eq. . (3) Let λ ∈ C ∗∗ ⊂ X ∗∗ and N be the weak . . fn (y)). . . . then C ⊂ C ∗∗ is weak — ∗ compact being the continuous image of a continuous map. . then k=1 each fk with k > m may be written as a linear combination {f1 . (16. fn (x)) = T x = (λ (f1 ) . . . . λ(fn )) = (f1 (x).9). . (16. . DRIVER † In fact we will show there exists x ∈ X such that λ(fk ) = fk (x) for k = 1. . less than ) by choosing α suﬃciently close to 1.5. C = C ∗∗ is weak x . i. . . . .

34. Folland 5. Folland 5. for ∈ D choose x ∈ X such that kx k = 1 and | (x )| ≥ 1 k k. Exercise 16.9.1. Folland 5. Folland 5. Exercise 16. Exercise 16. More Examples of Banach Spaces. Exercise 16.18.35. 16.5.13. Folland 5.20. Exercise 16.29. Exercise 16. Folland 5.e.23. (Folland 5.5. Exercise 16.) Exercise 16. Folland 5.41. Exercise 16.7.33. Exercise 16. Baire Category Result Problems. Folland 5. Exercise 16.1.2. Exercise 16. i. Exercise 16. Folland 5.30.25. 2 Exercise 16.26. Folland 5.12. Exercise 16.21.) Hint: use the greedy algorithm.REAL ANALYSIS LECTURE NOTES 327 16.37.38.6.24. 16. Folland 5. Folland 5. Exercise 16.18.40.10. Exercise 16. k·k) is a Banach space.4. Exercise 16.14. Folland 5. Folland 5. M) be a measurable space and M (X) denote the space of complex measures on (X. . (Move to Section 14. M) and for µ ∈ M (X) let kµk ≡ |µk(X).9. Folland 5. Hahn-Banach Theorem Problems.17.2.5. suppose D ⊂ X ∗ \ {0} is a countable dense subset of X ∗ .39.11. Let (X.3.21.5. Show (M (X). Folland 5.8.15. Exercise 16.3. Show X is separable as well. Folland 5.19.11. Folland 5. Folland 5. Exercise 16. Folland 5.17. Let X be a Banach space such that X ∗ is separable.19.10.31.36. Exercise 16. Folland 5. Exercise 16.16. Folland 5.20.22. Exercise 16. Exercise 16.32.

Folland 5.328 BRUCE K. These conditions are equivalent to {f (xn )}n=1 being Cauchy for all ∞ f ∈ X ∗ and {fn (x)}n=1 being Cauchy for all x ∈ X respectively.25.49. Show every weakly compact subset X is norm bounded and every weak—∗ compact subset of X ∗ is norm bounded. Remark 16. Weak Topology and Convergence Problems. Exercise 16.26. ∞ ∞ .53.31.27.30.48. Exercise 16. Similarly a sequence {fn }∞ ⊂ X ∗ is weak—∗ Cauchy if for all V ∈ τw∗ such that 0 ∈ V.50. Exercise 16. xn − xm ∈ V for all m.28. n suﬃciently large. fn − fm ∈ V n=1 for all m. Exercise 16. Folland 5. Exercise 16.29.32. Exercise 16. land 5. n suﬃciently large.51.5. DRIVER † 16. Folland 5. Folland 5. Folland 5. A sequence {xn }n=1 ⊂ X is weakly Cauchy if for all V ∈ τw such that 0 ∈ V.4.47. Deﬁnition 16.31. Exercise 16.

- Existence of Equilibrium in Economies With Infinitely Many Commodities
- Asplund Spaces
- Mean Value Theorem
- Limits3a.pdf
- __Notes_on_functional_analysis.pdf
- banalg-1
- Tut2Soln
- 数学分析习题课讲义（复旦大学.pdf
- 16
- Chapter 2
- e 024025030
- Limit and Continuity.pdf
- bku
- Lecture 40
- [Kuttler_K.]_One_variable_advanced_calculus(Bokos-Z1).pdf
- notesc
- VUTU6OKM
- MathAnalysis5 13 Gs
- 38079
- Noncritical Belyi Maps
- Chap12optB
- LADE12 General Linear Transformations
- 1501.00038
- S Multi Space

Skip carousel

- tmp5E24
- UT Dallas Syllabus for math4301.001.11f taught by Wieslaw Krawcewicz (wzk091000)
- On a Result of Fixed Point Theorems for Commuting and Weakly Compatible Maps
- tmpA253.tmp
- tmp48DC.tmp
- Implementation of Efficient Algorithm for Exact Hausdorff Distance
- Common Fixed Point of Mappings Using Sequentially Weak Compatible Mapping

Sign up to vote on this title

UsefulNot usefulRead Free for 30 Days

Cancel anytime.

Close Dialog## Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

Close Dialog## This title now requires a credit

Use one of your book credits to continue reading from where you left off, or restart the preview.

Loading