\( \newcommand{\N}{\mathbb{N}} \newcommand{\R}{\mathbb{R}} \newcommand{\C}{\mathbb{C}} \newcommand{\Q}{\mathbb{Q}} \newcommand{\Z}{\mathbb{Z}} \newcommand{\P}{\mathcal P} \newcommand{\B}{\mathcal B} \newcommand{\F}{\mathbb{F}} \newcommand{\E}{\mathcal E} \newcommand{\brac}[1]{\left(#1\right)} \newcommand{\abs}[1]{\left|#1\right|} \newcommand{\matrixx}[1]{\begin{bmatrix}#1\end {bmatrix}} \newcommand{\vmatrixx}[1]{\begin{vmatrix} #1\end{vmatrix}} \newcommand{\lims}{\mathop{\overline{\lim}}} \newcommand{\limi}{\mathop{\underline{\lim}}} \newcommand{\limn}{\lim_{n\to\infty}} \newcommand{\limsn}{\lims_{n\to\infty}} \newcommand{\limin}{\limi_{n\to\infty}} \newcommand{\nul}{\mathop{\mathrm{Nul}}} \newcommand{\col}{\mathop{\mathrm{Col}}} \newcommand{\rank}{\mathop{\mathrm{Rank}}} \newcommand{\dis}{\displaystyle} \newcommand{\spann}{\mathop{\mathrm{span}}} \newcommand{\range}{\mathop{\mathrm{range}}} \newcommand{\inner}[1]{\langle #1 \rangle} \newcommand{\innerr}[1]{\left\langle #1 \right \rangle} \newcommand{\ol}[1]{\overline{#1}} \newcommand{\toto}{\rightrightarrows} \newcommand{\upto}{\nearrow} \newcommand{\downto}{\searrow} \newcommand{\qed}{\quad \blacksquare} \newcommand{\tr}{\mathop{\mathrm{tr}}} \newcommand{\bm}{\boldsymbol} \newcommand{\cupp}{\bigcup} \newcommand{\capp}{\bigcap} \newcommand{\sqcupp}{\bigsqcup} \newcommand{\re}{\mathop{\mathrm{Re}}} \newcommand{\im}{\mathop{\mathrm{Im}}} \newcommand{\comma}{\text{,}} \newcommand{\foot}{\text{。}} \)

Thursday, October 23, 2014

Record a problem

Let $H$ be a Hilbert space. It is well-known that if $Z$ is a convex and closed subset of $H$, then for every $x\in H$, there is a unique element $z\in Z$ such that \[
\|x-z\|=d(x,Z).
\] Let's denote this unique element $z$ by $P_Zx$.
Problem. Let $Z_1\supseteq Z_2\supseteq Z_3\supseteq \cdots$ be a chain of closed convex subsets of a Hilbert space $H$, show that:
(a) If $\capp_{n=1}^\infty Z_n \neq\emptyset$, then $\|x-P_{Z_n}x\|\to \|x-P_Zx\|$.
(b) If $\capp_{n=1}^\infty Z_n=\emptyset$, then $\|x-P_{Z_n}x\|\to \infty$.
Since $Z_n$'s are not subspace of $H$, $P_{Z_n}$ itself is not a linear map, and therefore standard tools from functional analysis cannot be used. Nevertheless, by assuming $Z_n$'s are further a closed subspace, it is instructive to see a weak-convergence argument to conclude the result in (a) --- a slightly complicated solution.

Some also asked me can $\capp_{n=1}^\infty Z_n=\emptyset$ happen? Below is an example to show there is a chain of closed convex subsets with empty intersection:

Example. Let $Z_n=\{(1,\frac{1}{\sqrt{2}},\dots,\frac{1}{\sqrt{n}},a_1,a_2,\dots):(a_1,a_2,\dots)\in \ell^2\}$, then $Z_n$ is convex, a closed subset of $\ell^2$, and descending in $n$. However, if there is $(x_1,x_2,\dots)\in \capp_{n=1}^\infty Z_n$, then necessarily $x_n=1/\sqrt{n}$ for every $n$, and this element is not in $\ell^2$, a contradiction. So $\capp_{n=1}^\infty Z_n=\emptyset$.
  Next what is $d(x,Z_n)$ in this particular example? By definition for $z\in Z_n$ we have \[
\|x-z\|^2 = \sum_{k=1}^n\left|x_k-\frac{1}{\sqrt{k}}\right|^2+\sum_{k>n} |x_k-z_k|^2,
\] hence for fixed $x\in \ell^2$ the smallest possible $\|x-z\|$ is \[
d(x,Z_n)=\sqrt{\sum_{k=1}^n \left|x_k-\frac{1}{\sqrt{k}}\right|^2},
\] of course this diverges to infinity as $n\to\infty$ for any fixed $x\in \ell^2$. What's more, \[P_{Z_n}x =\textstyle (1,\frac{1}{\sqrt{2}},\dots,\frac{1}{\sqrt{n}}, x_{n+1},x_{n+2},\dots).\qed\]

No comments:

Post a Comment