Spring 2010 - Problem 1

construction, Lp spaces
  1. Let 1p<1 \leq p < \infty. Show that if a sequence of real-valued functions {fn}n1\set{f_n}_{n\geq1} converges in Lp(R)L^p\p{\R}, then it contains a subsequence that converges almost everywhere.
  2. Give an example of a sequence of functions converging to zero in L2(R)L^2\p{\R} that does not converge almost everywhere.
Solution.
  1. By definition, there exists fLp(R)f \in L^p\p{\R} such that fnfLp0\norm{f_n - f}_{L^p} \to 0. Thus, there exists n1n_1 so that fn1fLp21/p\norm{f_{n_1} - f}_{L^p} \leq 2^{-1/p}. By induction, there exists a subsequence {nk}k1\set{n_k}_{k\geq1} such that fnkfLp2k/p\norm{f_{n_k} - f}_{L^p} \leq 2^{-k/p}. Now consider

    FN=k=1Nfnk+1fnkpandF=k=1fnk+1fnkp.F_N = \sum_{k=1}^N \abs{f_{n_{k+1}} - f_{n_k}}^p \quad\text{and}\quad F = \sum_{k=1}^\infty \abs{f_{n_{k+1}} - f_{n_k}}^p.

    Since the terms are non-negative, {FN}N1\set{F_N}_{N\geq1} is an increasing sequence of functions, so by the monotone convergence theorem,

    FL1=k=1fnkfLppk=12k=1.\norm{F}_{L^1} = \sum_{k=1}^\infty \norm{f_{n_k} - f}_{L^p}^p \leq \sum_{k=1}^\infty 2^{-k} = 1.

    Thus, F<\abs{F} < \infty almost everywhere, so by the divergence test for series, it follows that fnkf0\abs{f_{n_k} - f} \to 0 for almost every xRx \in \R, i.e., {fnk}k1\set{f_{n_k}}_{k\geq1} converges to ff almost everywhere.

  2. Consider the typewriter sequence: for n1n \geq 1 and 1k2n11 \leq k \leq 2^{n-1}, set

    In,k=[k12n1,k2n1]andfn,k=χIn,k.I_{n,k} = \br{\frac{k - 1}{2^{n-1}}, \frac{k}{2^{n-1}}} \quad\text{and}\quad f_{n,k} = \chi_{I_{n,k}}.

    For each n1n \geq 1, {In,k}k\set{I_{n,k}}_k partition the interval [0,1]\br{0, 1}. Thus, for any xx and n1n \geq 1, there exists kk so that fn,k(x)=1f_{n,k}\p{x} = 1, but fn,k1(x)=0f_{n,k-1}\p{x} = 0 if k>1k > 1 or fn,k+1(x)=0f_{n,k+1}\p{x} = 0 if k=1k = 1, so fn,kf_{n,k} converges nowhere pointwise. On the other hand,

    fn,kL22=12n1n0,\norm{f_{n,k}}_{L^2}^2 = \frac{1}{2^{n-1}} \xrightarrow{n\to\infty} 0,

    so fn,k0f_{n,k} \to 0 in L2(R)L^2\p{\R}.