Measure Theory, Functional Analysis
2024-10-26
Preface
These notes accompany the Fall 2024 iteration of Math 114, a second-semester undergraduate course on real analysis and measure theory. These notes deviate quite a bit from lecture materials, in particular:
- I prefer a more principled construction of the Lebesgue measure as the restriction of
the exterior measure using the Carathéodory theorem.
- Sets declared measurable by the Carathéodory criteria are automatically closed under \(\sigma\)-algebra operations, so one only needs to show the measurability of rectangles to establish the measurability of the Borel sets.
- Materials are heavily supplemented by Sheldon Axler’s Measure, Integration and Real Analysis.
- Some time is spent on infinite-dimensional vector spaces and
results in Banach spaces, since
- nontrivial complexities arive when we generalize linear algebra constructions to infinite-dimensions (e.g. Hammel basis, span, and orthonormal basis).
- several theorems are important for convex analysis.
- Much more time is spent on the Fourier transform, with intuitions developed on the compact domain \(\partial D\cong S^1\) and later generalized to locally compact abelian groups.
- Other miscellany:
- Proof of Young’s inequality (proposition 7.2) using the Legendre transform to highlight the role of convexity.
Organization
- Abstract machinery of measures and \(\sigma\)-algebra.
- Lieb and Loss, chapter 1.
- The exterior measure and Lebesgue measure on \(\mathbb R^n\).
- Hunter.
- Lebesgue integral,
Littlewood’s three principles, and integral convergence theorems.
- Rudin, , chapter 11; Axler, chapter 2; Lieb and Loss, chapter 1.
- Iterated integrals: uniqueness of product measure
and Fubini’s theorem.
- Lieb and Loss, chapter 1.
- Spaces of functions:
normed vector spaces, dual space of bounded linear functionals,
and the Hahn-Banach theorem; Baire’s theorem and other Banach theorems.- Axler, Chapter 6.
- \(L^p\) spaces.
- Lieb and Loss, chapter 2; Axler, chapter 7.
- Fourier analysis: Example of the Fourier series
on a compact domain, the Haar measure, locally compact groups and their duals.
- Axler, Chapter 11, Rudin (Fourier analysis), Chapter 1, Su.
References for these notes
- Axler, Measure, Integration & Real analysis (Axler 2020).
- Lieb and Loss, Analysis (Lieb and Loss 2001).
- John K. Hunter, online notes on the Lebesgue measure.
- Rudin, Principles of Mathematical Analysis (Rudin et al. 1964).
- Rudin, Fourier analysis on groups (Rudin 2017).
- Dan Su, The Fourier Transform for Locally Compact Abelian Groups, online notes.
Main deliverables
- Monotone class theorem 2.1: freely-generated monotone class equals freely-generated \(\sigma\)-algebra for algebra of sets containing \(\Omega\).
- Carathéodory’s theorem 2.3: given an outer measure \(\mu\), the measurable sets \(B\subset \Omega\) satisfying \(\mu(\forall E) = \mu(E\cap B) + \mu(E\cap B^c)\) form a \(\sigma\)-algebra and obey countable additivity.
- Dominated convergence theorem 4.8:
dominated pointwise limit of measurable functions can be exchanged with integral.
- Hahn-Banach, hyperplane separation theorems: for every extension of a bounded linear functional, there exists a norm-minimizing one which preserves the original norm.
- Fubini’s theorem
- Baire’s theorem 6.4: for every cover of a complete metric space by closed sets, there exists an element with nonempty interior.
Proof techniques:
- Working with countable collection of approximations: given tolerance \(\epsilon\),
introduce approximation of index \(j\) with tolerance \(\epsilon 2^{-j}\)
to have additively bounded total tolerance.
- Examples: outer measure properties of exterior measure (proposition 3.1).
- Prove properties for characteristic functions, then generalize them
using approximation and limit theorems here.
- Examples: Luzin’s theorem,
monotone convergence theorem 4.7,
- Examples: Luzin’s theorem,
monotone convergence theorem 4.7,
- Sandwich bounds
- Lemma 2.2, Monotone convergence theorem.
- Argue the convergence of a subsequence satisfying \(\sum |f_k - f_{k-1}|<\infty\) when working with Cauchy sequences.
- Prime example for the use of monotone convergence theorem, Fatou’s lemma, and dominated convergence theorem: proposition 7.7.
- Using the monotone class theorem (example theorem 2.2): rewrite nested chain as disjoint union, then pass through the limit using the countable additivity of measures. Next apply the reduction from finite to \(\sigma\)-finite measures.
- Demonstrating a certain property (e.g. triviality): approximate by
a tractable dense subset.
- Orthonormality of the Fourier basis on \(L^2(\partial D)\) (proposition 8.7).
Mandatory todo list:
- Inductive generalization of Holder’s inequality.
- Put in some pset problems.
- Pset problem: separately continuous implies measurable.
- Dual of \(L^p\) space.
Optional:
- Lieb and Loss 2.9 to 2.18: compactness under weak convergence.
- Product measures and Fubini.
- Arnold-Kolmorogorov theorem.
Bibliography
Axler, Sheldon. 2020. Measure, Integration & Real Analysis. Springer Nature.
Lieb, Elliott H, and Michael Loss. 2001. Analysis. Vol. 14. American Mathematical Soc.
Rudin, Walter et al. 1964. Principles of Mathematical Analysis. Vol. 3. McGraw-hill New York.
Rudin, Walter. 2017. Fourier Analysis on Groups. Courier Dover Publications.