Convergence of $Ax:=\sum_{n=1}^{\infty}\alpha_nP_nx$

analysisfunctional-analysishilbert-spacesreal-analysissequences-and-series

Let $H$ be a Hilbert space and let $(P_n)_{n\in \mathbb N}$ be a sequence of orthogonal projections ($P_n \neq 0$) with $$\mathrm{ran}\ P_n\perp \mathrm{ran}\ P_m\ \mathrm{for}\ n \neq m.$$

Furthermore let $\alpha=(\alpha_n)_{n\in \mathbb N} \in l^\infty$ with $$\alpha_n \neq 0\ \forall n \in \mathbb N.$$

Show that the series $$Ax:=\sum_{n=1}^{\infty}\alpha_nP_nx$$ converges for every $x \in H$, $\|Ax\|=\|\alpha\|_\infty$. And show that it coverges with respect to the operator norm if $\alpha_n \to 0$.

I tried:

If $P_1$ and $P_2$ are orthogonal projections with $ran\ P_1 \perp ran\ P_2$ then $P_1+P_2$ is an orthogonal projection. So every finite sum of orthogonal projections is an orthogonal projection. Orthogonal projections are continuous. Therefore we have $$\|P_1x+\ldots P_n x\|^2\leq \|x\|^2$$ And because they are mutually orthogonal, we have: $$\|P_1x\|^2+\ldots+\|P_nx\|^2=\|P_1x+\ldots P_n x\|^2\leq \|x\|^2$$

So, $$\|Ax\|^2=\|\sum_{n=1}^{\infty}\alpha_nP_nx\|^2\leq\sum_{n=1}^{\infty}\|\alpha_nP_nx\|^2\leq\|\alpha\|_\infty^2\|x\|^2<\infty$$

Therefore $Ax$ converges and the line above also implies that for the operator norm of $Ax$ we have $$\|Ax\|\leq \|\alpha\|_\infty$$

For showing $\|Ax\|\geq \|\alpha\|_\infty$:

$$\|Ax\|=\sup_{x \in H\\\|x\|=1}\|Ax\|\geq\sup_{x \in S\\\|x\|=1}\|Ax\|=\sup_{\|x\|=1}\|\sum_{n=1}^\infty\alpha_nx\|\geq \sup_{\|x\|=1}\|x\|\sum_{n=1}^{\infty}(|\alpha_n|)\geq\|\alpha\|_\infty$$

Is this correct so far?

I don't know how to show that $Ax$ converges in operator norm

Best Answer

Your attempt looks fine, except for the last part (note though, that $\|Ax\|^2=\sum_{n=1} |\alpha_n| \|P_n x\|^2$ by our own argumentation - you don't just have inequality). It's not clear what $\mathcal{S}$ is, and there is no way that $Ax=\sum_{n=1}^{\infty} \alpha_n x$ for any $x$ since the $\alpha_n$ are pairwise orthogonal.

However, if $x\in Ran P_n$, then $A x= \alpha_n x$ so if you pick $x\in Ran P_n$ with $\|x\|=1$, you get that $\|A\|\geq \alpha_n$ for every $n$, which establishes the other inequality.

Finally, to prove norm convergence when $\alpha_n\to 0$, let $A_N:=\sum_{n=1}^N \alpha_n P_n$ and note that $$ \|(A-A_N) x\|^2=\sum_{n=N+1}^{\infty} |\alpha_n|^2 \|P_n x\|^2\leq \sup_{n>N} |\alpha|_n^2 \|x\|^2, $$ which implies that $\|A-A_N\|\to 0$ by our assumption of decay.

Related Question