Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Let $H=\ell_2$ be the Hilbert space of the square-summable sequences where $$ \langle x,y\rangle=\sum_{i=1}^{\infty}x_iy_i, \quad \|x\|=\sqrt{\langle x,x\rangle}. $$ Let $F: H\rightarrow H$ be an affine mapping, i.e. $$ F[\lambda u+(1-\lambda)v]=\lambda F(u)+(1-\lambda)F(v), \quad \forall u,v\in H, \lambda\in \mathbb{R}. $$ Let $\{u^k\}$ be a sequence given by $$ u^{k+1}=F(u^k) \quad k\in\mathbb{N}, $$ where $u^0$ is an any point in $H$. Find the conditions on $F$ and $u^0$ such that $\{u^k\}$ is weakly convergent but not strongly convergent.

Example. If $u^0=(1,0,0,\ldots,0,\ldots)$ and $F(u)$ is given by $$ F(u)=(0,u_1, u_2, \ldots, u_n, \ldots) \quad \forall u=(u_1,u_2,\ldots, u_n, \ldots)\in H. $$ The sequence $\{u^k\}$ generated by the formula $u^{k+1}=F(u^k)$ is given by $$ u^0=(1,0,\ldots, 0,\ldots), \quad u^1=(0,1,0,\ldots, 0, \ldots), \ldots, u^n=(0,0,\ldots, 1, 0, \ldots),\ldots $$ is weakly convergent but not strongly convergent to $0\in H$.

share|cite|improve this question
$F$ cannot be a compact opeator. – Tomás Oct 9 '12 at 11:25
Is the mapping $F$ arbitrary? In this case, the question is strange, for example we can choose, once $x_0$ is fixed, $F$ who map $x_0$ to $e_1$ (the first vector of the classical Hilbert basis), $e_1$ to $e_2$ etc... (if $x_0$ is not one of these vectors) and anything you want for the non concerned vectors. – Davide Giraudo Oct 9 '12 at 12:22
@Davide Giraudo: Dear Sir. Thank you for your comments. I would like to find some sufficent conditions on the mapping $F$ and $u^0$ such that $\{u^k\}$ is weakly convergent but not strongly convergent. – blindman Oct 9 '12 at 23:27
The problem is that if $F$ is non-linear, we allow a lot of things. At least, we can restrict us to linear $F$ (as linearity is quite compatible with weak convergence). – Davide Giraudo Oct 10 '12 at 7:44
@Davide Giraudo: I reduced my question to the case $F$ is an affine mapping. – blindman Oct 11 '12 at 0:19

I think the question is too broad to find a simple answer. In any case, let us restrict to linear, bounded $F$. It is possible that $||F|| > 1$ and $u^0$ can be found according to your description, but I think that would be difficult to find criteria in that case (generically, I would imagine that $||F^nx||$ runs off to $+\infty$ so $F^nx$ does not converge weakly by a classical theorem, and the only way to help this is to restrict $F$ to a subspace of which it is a contraction, but then we might as well start with that)

A result you might find relevant is Jacobs–Glicksberg–de Leeuw decomposition (a great book that mentions this theorem is Eisner's Stability of Operators and Operator Semigroups), which says that $H$ splits into a direct sum of $H_r = \operatorname{span}\{x \in H \ : \ F x = \gamma x \text{ for some $\gamma \in \mathbb{C}$ with $|\gamma| = 1$ } \}$ and $H_s = \{x \in H \ : \ 0 \text{ is a weak accumulation point of } \{F^nx\}_n \}$. This would mean that if $F$ has no eigenvalues on the unit circle (and we probably don't want that in any case), then $F^n x$ has $0$ as an accumulation point, for any choice of $x$.

It is not quite what you want - accumulation point is not yet the limit - but I hope it will be helpful.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.