Fiveable

🧐Functional Analysis Unit 10 Review

QR code for Functional Analysis practice questions

10.4 Characterizations of reflexivity

10.4 Characterizations of reflexivity

Written by the Fiveable Content Team • Last updated August 2025
Written by the Fiveable Content Team • Last updated August 2025
🧐Functional Analysis
Unit & Topic Study Guides

Reflexivity in Banach spaces is a powerful concept with far-reaching implications. It's characterized by the attainment of norms on dual spaces and the coincidence of weak and weak* compactness. These properties provide valuable tools for analyzing linear functionals and bounded sets.

The Eberlein-Šmulian theorem further connects reflexivity to weak convergence of bounded sequences. This relationship highlights the importance of reflexivity in understanding the behavior of sequences and subsets in Banach spaces, making it a crucial concept in functional analysis.

Characterizations of Reflexivity

James' theorem for reflexivity

  • Characterizes reflexivity in terms of the attainment of the norm on the dual space
    • Banach space XX is reflexive if and only if every continuous linear functional on XX attains its norm
  • Proof assuming XX is reflexive:
    • Let fXf \in X^* be a continuous linear functional on XX
    • Hahn-Banach theorem guarantees existence of xXx^{**} \in X^{**} such that x(f)=fx^{**}(f) = \|f\| and x=1\|x^{**}\| = 1
    • Reflexivity of XX implies existence of xXx \in X such that x(g)=g(x)x^{**}(g) = g(x) for all gXg \in X^*
    • In particular, f(x)=x(f)=ff(x) = x^{**}(f) = \|f\|, so ff attains its norm at xx
  • Proof assuming every continuous linear functional on XX attains its norm:
    • Let xXx^{**} \in X^{**} be a bounded linear functional on XX^*
    • Define f(x)=x(J(x))f(x) = x^{**}(J(x)), where J:XXJ: X \to X^{**} is the canonical embedding
    • fXf \in X^* and by assumption, there exists x0Xx_0 \in X such that f(x0)=ff(x_0) = \|f\|
    • For any gXg \in X^*, x(g)g(x0)=x(g)f(x0)xgJ(x0)|x^{**}(g) - g(x_0)| = |x^{**}(g) - f(x_0)| \leq \|x^{**}\| \|g - J(x_0)\|
    • Taking the supremum over all gg with g1\|g\| \leq 1 yields xJ(x0)xJ(x0)J(x0)=0\|x^{**} - J(x_0)\| \leq \|x^{**}\| \|J(x_0) - J(x_0)\| = 0
    • Thus, x=J(x0)x^{**} = J(x_0), so XX is reflexive

Weak compactness and reflexivity

  • Weakly compact subset KK of Banach space XX: every sequence in KK has a weakly convergent subsequence
    • Weak convergence: sequence (xn)(x_n) in XX converges weakly to xXx \in X if f(xn)f(x)f(x_n) \to f(x) for all fXf \in X^*
  • Reflexive spaces characterized by coincidence of weak and weak* compactness
    • In reflexive space, bounded set is weakly compact if and only if it is weakly* compact
  • Weakly compact sets in Banach space are bounded and closed in norm topology
  • In reflexive space, closed unit ball is weakly compact

Eberlein-Šmulian theorem in reflexivity

  • Eberlein-Šmulian theorem: for Banach space XX, the following are equivalent:
    1. XX is reflexive
    2. Every bounded sequence in XX has a weakly convergent subsequence
    3. Every bounded subset of XX is weakly relatively compact
  • Proof (1 ⇒ 2): Assume XX is reflexive, let (xn)(x_n) be bounded sequence in XX
    • Reflexivity of XX implies closed unit ball BXB_X is weakly compact
    • Sequence (xn/xn)(x_n / \|x_n\|) contained in BXB_X, so it has a weakly convergent subsequence
    • Corresponding subsequence of (xn)(x_n) is also weakly convergent
  • Proof (2 ⇒ 3): Assume every bounded sequence in XX has a weakly convergent subsequence
    • Let AXA \subseteq X be bounded, (xn)(x_n) be sequence in AA
    • By assumption, (xn)(x_n) has a weakly convergent subsequence, so AA is weakly relatively compact
  • Proof (3 ⇒ 1): Assume every bounded subset of XX is weakly relatively compact
    • In particular, closed unit ball BXB_X is weakly relatively compact
    • Since BXB_X is also weakly closed, it is weakly compact
    • Thus, XX is reflexive

Reflexivity vs Radon-Nikodým property

  • Banach space XX has Radon-Nikodým property (RNP) if for every finite measure space (Ω,Σ,μ)(\Omega, \Sigma, \mu) and every bounded linear operator T:L1(μ)XT: L^1(\mu) \to X, there exists a Bochner integrable function g:ΩXg: \Omega \to X such that T(f)=ΩfgdμT(f) = \int_\Omega f g d\mu for all fL1(μ)f \in L^1(\mu)
  • Every reflexive Banach space has RNP
    • Proof: Let XX be reflexive, T:L1(μ)XT: L^1(\mu) \to X be bounded linear operator
    • Define bounded linear functional φ\varphi on L1(μ)XL^1(\mu) \otimes X^* by φ(fx)=T(f),x\varphi(f \otimes x^*) = \langle T(f), x^* \rangle
    • Hahn-Banach theorem extends φ\varphi to bounded linear functional on L1(μ,X)L^1(\mu, X^*)
    • Reflexivity of XX implies L1(μ,X)L1(μ,X)L^1(\mu, X^*) \cong L^1(\mu, X)^*, so there exists gL1(μ,X)g \in L^1(\mu, X) such that φ(F)=ΩF(g)dμ\varphi(F) = \int_\Omega F(g) d\mu for all FL1(μ,X)F \in L^1(\mu, X^*)
    • In particular, T(f),x=Ωfg,xdμ\langle T(f), x^* \rangle = \int_\Omega f \langle g, x^* \rangle d\mu for all fL1(μ)f \in L^1(\mu) and xXx^* \in X^*
    • Thus, T(f)=ΩfgdμT(f) = \int_\Omega f g d\mu for all fL1(μ)f \in L^1(\mu), so XX has RNP
  • There exist non-reflexive Banach spaces with RNP (1\ell^1)
  • There exist reflexive Banach spaces without RNP (L1[0,1]L^1[0,1])
Pep mascot
Upgrade your Fiveable account to print any study guide

Download study guides as beautiful PDFs See example

Print or share PDFs with your students

Always prints our latest, updated content

Mark up and annotate as you study

Click below to go to billing portal → update your plan → choose Yearly → and select "Fiveable Share Plan". Only pay the difference

Plan is open to all students, teachers, parents, etc
Pep mascot
Upgrade your Fiveable account to export vocabulary

Download study guides as beautiful PDFs See example

Print or share PDFs with your students

Always prints our latest, updated content

Mark up and annotate as you study

Plan is open to all students, teachers, parents, etc
report an error
description

screenshots help us find and fix the issue faster (optional)

add screenshot

2,589 studying →