0

How can I prove the following theorem with explanation. please

For any nonempty subset $M$ of a Hilbert space $H$, the span of $M$ is dense in $H$ if and only if $M^{\perp}=\{0\}$

I read the prove from Erwin kreyszig page 149 but I didn't understand anything

If $V$ is a subspace of a Hilbert space $H$ and $V^{\perp}=${$0$} so $\rightarrow$ $V=H$ is this correct or not ? Or $V$ has to be closed?

I want help

  • 1
    Hi and welcome to the site! Since this is a site that encourages and helps with learning, it is best if you show your own ideas and efforts in solving the question. Can you edit your question to add your thoughts and ideas about it? – 5xum May 04 '15 at 15:07
  • 4
    Welcome to Math.SE! I think you misunderstood a key aspect of this site. This is more about helping you find the right answer then giving you answers to arbitrary problems. For example, if you were to give us what you've tried, where you've looked, what you've read to answer the problem, and describe why you aren't satisfied/doubt what you've come across, I'm sure others would help. Also, choosing a descriptive title that described the content would help. For more, I recommend reading How to ask a good question? – davidlowryduda May 04 '15 at 15:10

1 Answers1

0

First let us assume that $Span(M)$ is dense in $H$. We need to show that $M^{\perp} = \{0\}$. So, let $x\in M^{\perp}$ be any arbitrary element. Since $Span(M)$ is dense in $H$ so there exists a sequence $(x_n)$ of elements of $Span(M)$ such that $x_n\longrightarrow x$. Now, by continuity of inner product it follows that $\langle x_n, x\rangle \longrightarrow \langle x,x\rangle$. But each $\langle x_n, x\rangle = 0$, so, $\langle x,x\rangle = \|x\|^2 = 0$ which implies $x=0$. Thus, $M^{\perp}=\{0\}$.

$\textbf{Note:}$ Notice that $x$ was in $M^{\perp}$, so, $\langle x,y\rangle = 0$ for all $y \in M^{\perp}$. But this also implies that $\langle x,z\rangle = 0$ for all $z \in Span(M)$ because any $z \in Span(M)$ is a finite linear combination of elements of $M$ and inner product is linear in first variable and conjugate linear in second variable. This is the reason why $\langle x_n, x\rangle = 0$ for all $n$ above.

Conversely, let us assume that $M^{\perp} = \{0\}$. Then notice that $(Span(M))^{\perp}$ is a closed subspace of $H$. Therefore, by lemma 3.3.4 on page 146 you can write \begin{equation} H = (Span(M))^{\perp} \oplus ((Span(M))^{\perp})^{\perp}. \end{equation} Since $((Span(M))^{\perp})^{\perp} = \overline{Span(M)}$See this link, it follows that \begin{equation} H = (Span(M))^{\perp} \oplus \overline{Span(M)}. \end{equation} So, it is enough to show that $(Span(M))^{\perp}= \{0\}$. But this is easy to see and this is also done in the book.

Urban PENDU
  • 2,499
  • If $V$ is a subspace of a Hilbert space $H$ and $V^{\perp}=${$0$} so $\rightarrow$ $V=H$ is this correct or not ? –  May 04 '15 at 18:43
  • Or $V$ has to be closed ???? –  May 04 '15 at 18:54
  • See $V^{\perp} = {0}$ will only give you that $\overline{V} = H$ and if $V$ is closed, then it will follow that $V = H$. – Urban PENDU May 04 '15 at 19:30