Loading [MathJax]/extensions/TeX/AMSmath.js

Saturday, April 28, 2012

Some results in real analysis

Proposition. Let f: [a, b] \rightarrow \mathbb{R} be a function such that left-hand limit f(x-) and right-hand limit f(x+) of f exist at every x \in [a, b]. Then the number of discontinuities of f is at most countable.

Proof: Define a real-valued function g on [a, b] by g(x) = \max\{|f(x) - f(x+)|, |f(x) - f(x-)|\}. Then f is continuous at x iff g(x) = 0. Let D_n = \{x \in [a, b]: g(x) \geq 1/n\}. The points of discontinuity of f is exactly \bigcup_{n = 1}^\infty D_n. We prove that each D_n is finite.
Suppose on the contrary that D_n is infinite. Then there is a sequence \{x_n\} of distinct points in D_n which converges to x \in [a, b]. Since f(x+), f(x-) exist, we can choose some \delta > 0 such that |f(y) - f(x+)| < 1/4n whenever 0 < y - x < \delta and |f(y) - f(x-)| < 1/4n whenever 0 < x - y < \delta. Therefore |f(y) - f(z)| < 1/2n whenever y, z \in (x - \delta, x) or y, z \in (x, x + \delta). We can find some k such that 0 < |x_k - x| < \delta. If x_k > x, then |f(y) - f(x_k)| < 1/2n whenever y \in (x, x + \delta), let y \rightarrow x_k from both left and right we get |f(x_k+) - f(x_k)| \leq 1/2n and |f(x_k-) - f(x_k)| \leq 1/2n, so g(x) \leq 1/2n < 1/n \leq g(x), contradiction. If x_k < x then similarly we also obtain a contradiction. Hence D_n must be finite. QED.

Proposition. Let f_n: U \rightarrow \mathbb{R} be a sequence of continuous functions on U \subseteq \mathbb{R}^d such that f_1 \leq f_2 \leq \cdots and \{f_n\} converges pointwise to a continuous function f on U. Then f_n \rightarrow f uniformly on compact subsets.

Proof: Note that f_1 \leq \cdots \leq f_n \leq \cdots \leq f. Let K \subseteq U be compact and \epsilon > 0. For any x \in K, since f_n(x) \rightarrow f(x), there is N_x \in \mathbb{N} such that f(x) - f_n(x) < \epsilon/3 whenever n \geq N_x. For any n \in \mathbb{N}, by uniform continuity of f and f_n on K, there is \delta_n > 0 such that for any x, y \in K, |f(x) - f(y)| < \epsilon/3 and |f_n(x) - f_n(y)| < \epsilon/3 whenever |x - y| < \delta_n. For every x \in K, if y \in B(x, \delta_{N_x}), then
\begin{align*} &f(y) - f_n(y) \leq f(y) - f_{N_x}(y) \\ &= [f(y) - f(x)] + [f(x) - f_{N_x}(x)] + [f_{N_x}(x) - f_{N_x}(y)] \\ &< \epsilon/3 + \epsilon/3 + \epsilon/3 = \epsilon \end{align*}
whenever n \geq N_x. Since \{B(x, \delta_{N_x}): x \in K\} is an open cover of K, then there are x_1, \ldots, x_k \in K such that K = \bigcup_{j = 1}^k B(x_j, \delta_{N_{x_j}}). Take N = \max\{N_{x_1}, \ldots, N_{x_k}\}. For n \geq N, if x \in K, then x \in B(x_j, \delta_{N_{x_j}}) for some 1 \leq j \leq k, so f(x) - f_n(x) < \epsilon. Hence f_n \rightarrow f uniformly on K. QED.


(To be continued)

Friday, April 27, 2012

Conformal equivalence of annuli

For 0 < r < R, let A(r, R) = \{z \in \mathbb{C}: r < |z| < R\}, the annulus centered at the origin with inner radius r and outer radius R. We are going to show that the conformal equivalence classes of such anuuli in the complex plane are parametrized by the ratio R/r.

Theorem. A(r_1, R_1) is conformally equivalent to A(r_2, R_2) if and only if R_1/r_1 = R_2/r_2.
Proof: Suppose R_1/r_1 = R_2/r_2. Then there exists k > 0 such that R_2 = k R_1 and r_2 = k r_1. The map f(z) = kz gives a conformal equivalence from A(r_1, R_1) onto A(r_2, R_2).
Conversely, let f be a biholomorphic map from A(r_1, R_1) onto A(r_2, R_2). By scaling if necessary,  we may assume r_1 = r_2 = 1. Let A_1 = A(1, R_1) and A_2 = A(1, R_2). Fix some 1 < r < R_2.
Let C = \{z \in \mathbb{C}: |z| = r\}.
Since f^{-1} is continuous, f^{-1}(C) is compact, so we can find some \delta > 0 such that A(1, 1 + \delta) \cap f^{-1}(C) = \emptyset. Let V = f(A(1, 1 + \delta)). Since f is continuous, V is connected, so either V \subseteq A(1, r) or V \subseteq A(r, R_2). By replacing f with R_2/f, we may assume the first case holds.
Claim: |f(z_n)| \rightarrow 1 whenever |z_n| \rightarrow 1.
Proof of claim: Let \{z_n\} \subseteq A(1, 1 + \delta). Note that \{f(z_n)\} does not have a limit point in A_2 since otherwise \{z_n\} would have a limit point in A_1 by continuity of f^{-1}, contradicting |z_n| \rightarrow 1. Hence |f(z_n)| \rightarrow 1 or |f(z_n)| \rightarrow R_2 (it must converge by continuity). The latter case is ruled out as f(z_n) \in V \subseteq A(1, r).
Similarly, we also have:
Claim:  |f(z_n)| \rightarrow R_2 whenever |z_n| \rightarrow  R_1.
Now set \alpha = \log R_2/ \log R_1. Define g: A_1 \rightarrow \mathbb{R} by g(z)  = \log |f(z)|^2 - \alpha \log |z|^2 = 2(\log |f(z)| - \alpha \log |z|). We know that \log |h| is harmonic whenever h is holomorphic and nonzero, so g is a harmonic function. By the two claims, g extends to a continuous function on \overline{A_1} vanishing on \partial A_1. This forces g to vanish identically on A_1. In particular, 0 = \frac{\partial g}{\partial z} = \frac{f'(z)}{f(z)} - \frac{\alpha}{z}. Take some 1 < c < R_1 and let \gamma(t) = ce^{it}, t \in [0, 2\pi]. We have \alpha = \frac{1}{2 \pi i}\int_\gamma \frac{\alpha}{z} dz = \frac{1}{2 \pi i}\int_\gamma \frac{f'(z)}{f(z)} dz = \mathrm{Ind}_{f \circ \gamma}(0), so \alpha > 0 is an integer. Observe that \frac{d}{dz}(z^{-\alpha}f(z)) = z^{-\alpha - 1}(-\alpha f(z) + zf'(z)) = 0, thus f(z) = Kz^{\alpha} for some nonzero constant K. Since f is injective, \alpha = 1. Therefore R_2 = R_1.

Reference:
Rudin - Real and Complex Analysis (3rd Ed.)

Saturday, April 21, 2012

Koebe's 1/4-Theorem

Notation: B(a, r) = \{z \in \mathbb{C}: |z - a| < r \}, B = B(0, 1), \mathbb{S} is the Riemann sphere.

Let \mathcal{S} be the collection of all injective analytic function f on the unit disk with  f(0) = 0 and f'(0) = 1. We know from the Open Mapping Theorem that f(B) must contain a disk B(0, r_f). For the class \mathcal{S}, there is a universal r > 0 such that f(B) contains B(0, r) for all f \in S.

Theorem (Koebe's 1/4-Theorem). If f \in \mathcal{S} then B(0, 1/4) \subseteq f(B).

We need to the following result, which is of independent interest.

Theorem (The Area Theorem). Let g:  B \backslash \{0\} \rightarrow  \mathbb{C} be an injective analytic function with Laurent series expansion at 0 g(z) = \frac{1}{z} + c_0 + c_1 z + c_2 z^2 + \cdots. Then \sum_{n = 1}^\infty n|c_n|^2 \leq 1.
Proof: Note that g is a conformal equivalence between B and g(B) \subseteq \mathbb{S} and \infty \in g(B). Fix any 0 < r < 1. Consider U_r = \mathbb{S} \backslash \overline{B(0, r)}. It is an open set in \mathbb{C} with boundary \gamma parametrized by \theta \mapsto h(re^{i\theta}) where \theta runs from 2\pi to 0. The area of U_r is given by \begin{align*} \mathrm{Area}(U_r) &= \frac{1}{2i}\int_\gamma \bar{w} dw = -\frac{1}{2i} \int_0^{2\pi} \overline{h(re^{i\theta})}h'(re^{i\theta})ire^{i\theta}d\theta \\ &= - \frac{1}{2i} \int_0^{2\pi}  [ \left(-\frac{1}{r^2} + |c_1|^2 + 2|c_2|^2 + 3|c_3|^2 + \cdots \right)r^2 \\ &{\,\,} + \textrm{ terms with nonzero integral powers of } e^{i\theta} ]  \\ &= \pi \left(\frac{1}{r^2} - \sum_{n = 1}^\infty n|c_n|^2 r^2 \right). \end{align*}
Since area is always non-negative, let r \rightarrow 1 and we are done.

Proof of Koebe's Theorem: Let f \in \mathcal{S}. Then f omits some w_0 \in \mathbb{C} (or otherwise f^{-1} is a bounded entire function, which is then constant by Liouville's Theorem, contradiction), so h = 1/f omits 0 and 1/w_0. Observe that the Taylor series of f at 0 takes the form f(z) = z + a_2 z^2 + a_3 z^3 + \cdots, so h(z) = \frac{1}{z + a_2 z^2 + \cdots} = \frac{1}{z} + c_0 + c_1 z + c_2 z^2 + \cdots.  Let z_0 = 0 \textrm{ or } 1/w_0. We have g(z) = \frac{1}{h(z) - z_0} =  \frac{z}{1 + (c_0 - z_0)z + c_1 z^2 + \cdots} = z + (z_0 - c_0)z^2 + \cdots. In particular, g(0) = 0 and g'(0) = 1. Also note that g is injective. Thus g \in \mathcal{S}.
We claim that there exists some u \in \mathcal{S} such that u(z)^2 = g(z^2). Since g is injective and g(0) = 0, the analytic function g(z)/z omits 0 on B, we can find some analytic function \varphi on B such that \varphi(z)^2 = g(z)/z and \varphi(0) = 1. Let u(z) = z\varphi(z^2). Then g(z^2) = z^2 \cdot g(z^2)/z^2 = z^2\varphi(z^2)^2 = u(z)^2. Since \varphi(0) = 1 and \varphi'(0) = \frac{1}{2}(g(z)/z)^{-1/2}\frac{d}{dz}(g(z)/z)|_{z = 0} = \frac{1}{2} (z_0 - c_0), u(z) = z\varphi(z^2) = z + \frac{1}{2}(z_0 - c_0)z^3 + \cdots. Hence u(0) = 0 and u'(0) = 1. To show that u \in \mathcal{S} it remains to prove that u is injective. Suppose u(z) = u(w). Then g(z^2) = g(w^2). By injectivity of g, z^2 = w^2 or z = \pm w. If z = -w, then u(w) = u(z) = u(-w) = -u(w), so g(w^2) = u(w)^2 = 0 and w^2 = 0 by injectivity of g again, i.e. w = z = 0. In any case we have z = w.
Now consider the Laurent expansion of 1/u at 0, we have \frac{1}{u(z)} = \frac{1}{z + \frac{1}{2}(z_0 - c_0)z^3 + \cdots} = \frac{1}{z}(1 - \frac{1}{2}(z_0 - c_0)z^2 + \cdots) = \frac{1}{z} - \frac{1}{2}(z_0 - c_0)z + \cdots. By the Area Theorem, |-\frac{1}{2}(z_0 - c_0)| \leq 1, or |z_0 - c_0| \leq 2. Therefore |1/w_0| \leq |1/w_0 - c_0| + |0 - c_0| \leq 2 + 2 = 4, or |w_0| \geq 1/4. It follows that \mathbb{C} \backslash f(B) \subseteq \{w \in \mathbb{C}: |w| \geq 1/4\}, so B(0, 1/4) \subseteq f(B). QED.


Reference:
Andersson - Topics in Complex Analysis

Friday, April 20, 2012

Weak*-closed subspaces as dual spaces

Let X be a Banach space and M be a subspace of X^* closed under the weak* topology. We show that M can be identified as the dual space of some Banach space.

Let M_\perp = \{x \in X: f(x) = 0 \,\, \forall f \in M\} = \bigcap_{f \in M} \ker(f). This is a (weakly) closed subspace of X. Then we can form the quotient Banach space X/M_\perp with the norm \|x + M_\perp\| = \inf_{m \in M_\perp} \|x - m\|.

Theorem. M is isometrically isomorphic to (X/M_\perp)^*.
Proof: Define a map \Phi: M \rightarrow (X/M_\perp)^* by [\Phi(f)](x + M_\perp) = f(x) for all f \in M. Firstly, \Phi(f) is a well-defined linear functional on X/M_\perp because f vanishes on M_\perp whenever f \in M. Secondly, \Phi(f) is bounded as |\Phi(f)(x + M_\perp)| = |f(x)| = |f(x - m)| \leq \|f\|\|x - m\| \,\, \forall m \in M_\perp,  so |\Phi(f)(x + M_\perp)| \leq \|f\|\|x + M_\perp\|. Hence \Phi is well-defined.
Clearly \Phi is linear. From the above we have \|\Phi(f)\| \leq \|f\|. Let \epsilon > 0. Choose some x \in X with \|x\| \leq 1 and |f(x)| > \|f\| - \epsilon. Then \|x + M_\perp\| \leq 1 and |[\Phi(f)](x + M_\perp)| = |f(x)|  > \|f\| - \epsilon. Thus \|\Phi(f)\| = \|f\|. It follows that \Phi is an isometry.
It remains to show that \Phi is surjective. Let \ell \in (X/M_\perp)^* be given. Define f \in X^* by f(x) = \ell(x + M_\perp), i.e. f = \ell \circ \pi where \pi is the quotient map. Then f vanishes on M_\perp by definition, hence f \in M by the following lemma and it is clear that \Phi(f) = \ell.

Lemma. If f \in X^* vanishes on M_\perp, then f \in M.
Proof: Suppose f \in X^* \backslash M. By Hahn-Banach Theorem and weak*-closedness we can find some x \in X such that f(x) = 1 and g(x) = 0 whenever g \in M. But then x \in M_\perp and so f(x) = 0, contradiction.

This result implies the following important observation in the theory of operator algebras.

Theorem. Every von Neumann algebra is the dual space of some Banach space.
Proof: Let M be a von Neumann algebra on a Hilbert space \mathcal{H}. Then M is a \sigma-weakly closed subalgebra of B(\mathcal{H}), i.e. it is a weak*-closed subalgebra of B(\mathcal{H}) where the weak* topology of B(\mathcal{H}) comes from the fact that the dual space of the trace-class operators L^1(\mathcal{H}) on \mathcal{H} can be identified with B(\mathcal{H}). By our main theorem, M \cong (L^1(\mathcal{H})/M_\perp)^*.

A predual of a von Neumann algebra M is a Banach space X such that M = X^*. The above result says that every von Neumann algebra has a predual. In fact, the predual is unique by a result of Sakai. Furthermore, the property that having a predual characterizes von Neumann algebras among the class of C*-algebras.

Theorem (Sakai).  Suppose A is a W*-algebra, i.e. A is a C*-algebra and the dual space of some Banach space. Then A is *-isomorphic to a von Neumann algebra over some Hilbert space.

Open mapping theorem in terms of norm estimates

Here we present a simple but useful corollary of the Open Mapping Theorem in functional analysis.

Notation: B_X(a, r) = \{x \in X: \|x - a\| < r\}.

Proposition. Suppose X and Y are Banach spaces. Let f: X \rightarrow Y be a surjective bounded linear map. Then there exists some \delta > 0 such that \|f(x)\| \geq \delta\|x\| for all x \in X.

Proof: By the Open Mapping Theorem, f is an open map. Hence we can find some \delta > 0 such that B_Y(0, \delta) \subseteq f(B_X(0, 1)), so f(X \backslash B(0, 1)) \subseteq Y \backslash B(0, \delta), i.e. \|f(x)\| \geq \delta whenever \|x\| \geq 1. Therefore \|f(x)\| = \|x\|\|f(x/\|x\|)\| \geq \delta\|x\|.

Here is an application of the above result in the theory of Fourier series.

Theorem. The Fourier transform \mathcal{F}: L^1(\mathbb{T}) \rightarrow c_0(\mathbb{Z}) is not surjective.
Proof: Suppose not. By the Proposition, there exists \delta > 0 such that \|\hat{f}\|_\infty \geq \delta\|f\|_1. But if \{D_N\}_{N = 0}^\infty is the Dirichlet kernel, i.e.
D_N(x) = \sum_{n = -N}^N e^{inx}
then we have \|D_N\|_1 \rightarrow \infty and \|\widehat{D_N}\|_\infty \geq \delta\|D_N\|_1 \rightarrow \infty, which is absurd.

Tuesday, April 17, 2012

Boundedness of weakly convergent sequences

Let X be a Banach space. A net \{x_\alpha\} in X is said to weakly converge to x if \ell(x_\alpha) \rightarrow \ell(x) for all \ell \in X^*. Clearly, every norm convergent net is weakly convergent (with the same limit). We know that a norm convergent sequence is necessarily bounded. It is natural to ask boundedness still hold for weakly convergent sequences.

Theorem. Every weakly convergent sequence in X is bounded.
Proof: Let \{x_n\} be a weakly convergent sequence in X. Let T_n \in X^{**} be defined by T_n(\ell) = \ell(x_n) for all \ell \in X^*. Fix an \ell \in X^*. For any n \in \mathbb{N}, since the sequence \{\ell(x_n)\} is convergent, \{T_n(\ell)\} is a bounded set. By Uniform Boundedness Principle,
\sup_{n \in \mathbb{N}} \|x_n\| = \sup_{n \in \mathbb{N}} \|T_n\| < \infty,  i.e. \{x_n\} is bounded.

However, the above result is no longer true when sequences are replaced by nets. To give a counterexample, let X = L^1(\mathbb{R}). Denote the characteristic function of a subset A by \chi_A. Let \mathcal{O} be the collection of all open neighborhoods of 0 in \mathbb{R} of finite Lebesgue measure with the ordering given by U \prec V iff U \supseteq V. Consider the net \{\chi_U: U \in \mathcal{O}\} in X. It is not bounded since \|\chi_{(-n, n)}\|_1 = 2n \rightarrow \infty as n \rightarrow \infty. On the other hand, it converges weakly to 0. Fix any f \in X^* = L^\infty(\mathbb{R}). Choose M > 0 such that |f| \leq M almost everywhere. For every \epsilon > 0, we can choose some V \in \mathcal{O} with m(V) < \epsilon/M, then |\int f\chi_U dm| \leq M m(U) \leq M m(V) < \epsilon whenever U \succ V.