Banach space - Examples Bases and Further Reading

Classical Examples of Banach Spaces Introduction Before diving into abstract Banach space theory, it's essential to understand the concrete spaces that appear throughout functional analysis. These classical examples are the building blocks of the theory and the spaces you'll encounter most frequently in applications. Each example demonstrates how different notions of "size" or "distance" for functions and sequences naturally lead to Banach spaces. $L^p$ Spaces The $L^p(\Omega)$ spaces are fundamental in functional analysis, particularly in harmonic analysis, partial differential equations, and probability theory. For $1 \le p < \infty$, the space $L^p(\Omega)$ consists of measurable functions $f$ on a measure space $\Omega$ such that $|f|^p$ is integrable. These functions are considered equal if they agree almost everywhere. The norm is defined as: $$\|f\|p = \left(\int{\Omega} |f(x)|^p \, dx\right)^{1/p}$$ For $p = \infty$, $L^\infty(\Omega)$ consists of essentially bounded functions, with norm: $$\|f\|\infty = \text{ess\,sup}{x \in \Omega} |f(x)|$$ The essential supremum ignores sets of measure zero, which is crucial when working with equivalence classes of functions. Each $L^p(\Omega)$ is a Banach space under its respective norm. A key fact: $L^2(\Omega)$ is additionally a Hilbert space because the norm arises from the inner product $\langle f, g \rangle = \int\Omega f(x)\overline{g(x)} \, dx$. Why this matters: The $L^p$ spaces are "function spaces with controlled integrability." They allow us to study functions whose pointwise behavior might be wild, but whose aggregate size (measured by integration) is controlled. Sequence Spaces Sequence spaces are Banach spaces whose elements are infinite sequences of scalars. They're essential for understanding the structure of Banach spaces because many abstract constructions reduce to sequence space problems. $\ell^p$ spaces for $1 \le p < \infty$ consist of sequences $(an)$ of scalars such that $\sum{n=1}^\infty |an|^p < \infty$, with norm: $$\|a\|p = \left(\sum{n=1}^\infty |an|^p\right)^{1/p}$$ $\ell^1$ is the space of absolutely summable sequences $\ell^2$ is the space of square-summable sequences (this is a Hilbert space) $\ell^\infty$ consists of all bounded sequences $(an)$, with norm $\|a\|\infty = \supn |an|$. $c0$ consists of sequences that converge to zero, also with the supremum norm. This is a closed subspace of $\ell^\infty$. Why this matters: Sequence spaces are concrete and easier to visualize than function spaces. They also arise naturally when you expand functions in bases (which we'll discuss next). Spaces of Continuous Functions The space $C(K)$ consists of all continuous scalar-valued functions on a compact Hausdorff space $K$, equipped with the maximum norm: $$\|f\|\infty = \sup{x \in K} |f(x)|$$ Since $K$ is compact and $f$ is continuous, this supremum is always attained, so it's not just an essential supremum. When $K = [a,b]$ is a closed interval, $C([a,b])$ is the Banach space of continuous functions on that interval with the supremum norm. This space is separable (it has a countable dense subset—for example, polynomials with rational coefficients). Why this matters: $C(K)$ is important because every Banach space can be embedded into such a space—this is the content of the Banach-Mazur Theorem. It's also historically important and geometrically intuitive. <extrainfo> Sobolev and Hardy Spaces Sobolev spaces $W^{k,p}(\Omega)$ consist of functions in $L^p(\Omega)$ whose weak derivatives up to order $k$ also belong to $L^p(\Omega)$. These spaces incorporate both the size of a function and the size of its derivatives, making them essential in the study of partial differential equations. Hardy spaces $H^p$ consist of analytic functions on the unit disk with additional integrability conditions on their boundary values. They play a central role in complex analysis and harmonic analysis. Both of these are Banach spaces under appropriate norms, and both incorporate structural information (derivatives or analyticity) beyond just integrability. </extrainfo> Banach-Mazur Theorem Theorem: Every Banach space is isometrically isomorphic to a closed subspace of some $C(K)$ space. This remarkable result says that, up to isometric isomorphism, all Banach spaces can be viewed as spaces of continuous functions on compact Hausdorff spaces. This provides a unifying perspective and shows that $C(K)$ spaces are universal among Banach spaces. Schauder Bases and Approximation Properties Motivation and Definition In finite-dimensional vector spaces, bases are fundamental: every vector can be written uniquely as a finite linear combination of basis elements. A Schauder basis extends this idea to infinite-dimensional Banach spaces by allowing infinite series. A Schauder basis for a Banach space $X$ is a sequence $(en){n=1}^\infty$ such that every element $x \in X$ can be written uniquely as: $$x = \sum{n=1}^{\infty} an en$$ where $an$ are scalars and the series converges in the norm of $X$. The uniqueness means each $x$ corresponds to exactly one sequence of coefficients $(an)$. Key difference from algebraic bases: In infinite dimensions, we require convergence in the topology (the norm). This is what makes Schauder bases different from Hamel bases, and why they're more useful for Banach spaces. Coordinate Functionals and Biorthogonal Systems For each $n$, the coordinate functional $\varepsilonn : X \to \mathbb{F}$ is defined by: $$\varepsilonn(x) = an$$ where $an$ is the $n$-th coefficient in the expansion $x = \sum{n=1}^\infty an en$. An important fact: each $\varepsilonn$ is a bounded linear functional, so $\varepsilonn \in X^$ (the dual space). The boundedness follows from the properties of Schauder bases. The pair $(en, \varepsilonn)$ forms a biorthogonal system, meaning: $$\varepsilonn(em) = \delta{nm} = \begin{cases} 1 & \text{if } n = m \\ 0 & \text{if } n \neq m \end{cases}$$ This is the natural generalization of the orthonormality condition in Hilbert spaces, but for general Banach spaces where there's no inner product. Consequences of Having a Schauder Basis Separability: Any Banach space with a Schauder basis is separable. You can take all finite linear combinations of basis elements with rational (or algebraic) coefficients to get a countable dense subset. Bounded Approximation Property: If a space has a Schauder basis, the partial sum operators $PN(x) = \sum{n=1}^N an en$ converge to $x$ as $N \to \infty$. Moreover, there exists a constant $C > 0$ such that: $$\|PN(x)\| \le C \|x\|$$ for all $x$ and all $N$. This means the finite-rank operators formed by finite partial sums are uniformly bounded—they approximate the identity with bounded distortion. Why this matters: Having a Schauder basis gives you powerful structural information about a Banach space. Not every Banach space has a Schauder basis, so this is a strong property. Classical Bases in Common Spaces Understanding which classical spaces have explicit bases is important for building intuition. The Haar system forms a Schauder basis for $L^p(\mathbb{R})$ when $1 < p < \infty$. The Haar functions are piecewise constant functions with a nice recursive structure, making them computationally useful. The trigonometric system (sines and cosines, or complex exponentials $e^{inx}$) forms a Schauder basis for $L^p(\mathbb{T})$ when $1 < p < \infty$, where $\mathbb{T}$ is the circle (periodic functions on $[0, 2\pi)$). This connects to classical Fourier analysis. The Schauder system consists of piecewise linear "hat" functions on $[0,1]$. It forms a Schauder basis for the space $C([0,1])$ of continuous functions on the unit interval with the supremum norm. Why this matters: These examples show that the abstract concept of a Schauder basis has concrete realizations in spaces you encounter frequently. The Fourier system is particularly important because it shows that bases from classical analysis fit into the modern Banach space framework. James' Characterization of Reflexivity via Bases A Banach space is reflexive if $X = X^{}$ (the space equals its double dual). Reflexivity is an important property with many equivalent characterizations. For Banach spaces with a Schauder basis, James proved a beautiful characterization: A Banach space with a Schauder basis $(en)$ is reflexive if and only if the basis is both: Shrinking: The coordinate functionals $(\varepsilonn)$ form a Schauder basis for $X^$ (the dual space). Intuitively, the "span" of the functionals doesn't miss any functionals. Boundedly complete: Every bounded blockwise partial sum converges. More precisely: if $(x^{(k)})$ is a sequence of elements in $X$ formed by "blocks" of the basis (like $x^{(k)} = \sum{n=N{k-1}+1}^{Nk} an^{(k)} en$), and the partial sums of the combined series are bounded, then the series converges. Why this matters: This characterization gives a concrete way to verify reflexivity using the structure of the basis, rather than trying to construct an isomorphism between $X$ and $X^{}$ directly. <extrainfo> Differentiation in Banach Spaces When extending calculus to Banach spaces, two notions of derivative are important. Fréchet Derivative The Fréchet derivative is the natural generalization of the derivative to Banach spaces. For a map $F : X \to Y$ between Banach spaces, the Fréchet derivative at $x \in X$ is a bounded linear operator $DF(x) : X \to Y$ such that: $$\|F(x+h) - F(x) - DF(x)h\| = o(\|h\|)$$ as $\|h\| \to 0$. This is the same definition as in finite dimensions, but now applied to functions between Banach spaces. Gâteaux Derivative The Gâteaux derivative is a weaker notion. For $F : X \to Y$ at $x \in X$ in direction $h \in X$, it is: $$\delta F(x;h) = \lim{t \to 0} \frac{F(x+th) - F(x)}{t}$$ when the limit exists. This is a directional derivative—it measures how $F$ changes along a specific direction. Relation Between Derivatives Every Fréchet differentiable map is Gâteaux differentiable, but the converse is false. A function can have Gâteaux derivatives in all directions without having a Fréchet derivative. The Fréchet derivative is the stronger condition because it requires uniform control over all directions simultaneously. </extrainfo>