MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

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 E be a normed space, and let $T$:E * $\rightarrow$ E * be a nonlinear operator.

Suppose that :

1) $T$ is continuous from (E *, ||.||) to itself (i.e., it is norm-continuous).


2) $T$ is continuous from (E *, w *) to itself (i.e., it is weakly-star continuous).

Then does it follow that

3) $T$ is continuous from (E *, w) to itself (i.e., it is weakly continuous) ?

I guess the case E = ℓ∞ would be particularly interesting.

share|cite|improve this question
Ady, is your interest in $\ell_\infty$ because it is a Grothendieck space whose dual unit ball is not weak* sequentially compact? – Philip Brooker Jan 23 '10 at 0:51
Very true. :-) Indeed. Mainly because it is Grothendieck (and with Dunford-Pettis). I just cannot see a counterexample here. – Ady Jan 23 '10 at 1:02

I think Ady's question has a negative answer.

Recall that the Mazur map, $T$, from the unit ball of $\ell_2$ onto the unit ball of $\ell_1$, is defined by $T(\sum a_i e_i)= \sum a_i^2 b_i e_i$, where $b_i$ is the sign of $a_i$. It is a uniform homeomorphism in the norm topologies and obviously is coordinatewise continuous, which means that it is weak ($=$ weak$*$ in $\ell_2$) to weak$^*$ continuous (since on the unit ball of $\ell_p$ with $p$ finite the weak* topology is the topology of coordinatewise convergence). $T$ is not weak to weak continuous since the unit vector basis converges weakly to zero in $\ell_2$ but not in $\ell_1$.

The problem is to extend $T$ to all of $\ell_2$. I think that the easiest way to do this is to show that there is a retraction $R$ from $\ell_2$ onto its unit ball which is both norm to norm continuous and weak to weak continuous. Define $R$ on the complement of the unit ball by

$$ R(\sum a_i e_i) = \sum_{i=1}^n a_i e_i + t e_{n+1}, $$ where $\sum_{i=1}^n a_i^2 \le 1 < \sum_{i=1}^{n+1} a_i^2$ and $t$ is chosen to have the same sign as $a_{n+1}$ and to make the image vector have norm one. $R$ is obviously continuous (even Lipschitz) in the norm topology and is continuous in the topology of coordinatewise convergence, hence is weak to weak continuous (since in all of $\ell_2$, the weak topology is stronger than the topology of coordinatewise convergence).

To get a counterexample that maps one dual space to itself, work with the space $\ell_2 \oplus \ell_1$.

share|cite|improve this answer
Thank you very much ! It seems that your construction heavily relies on Schur. Therefore, may I ask: do you think that the answer is negative for all nonreflexive spaces, including the quasi-reflexive ones ? E.g., what about E = J, the James space ? – Ady Jan 13 '10 at 16:21
Any non reflexive space with a basis has a normalized basis that does not converge weakly to zero, but I don't know a replacement for the Mazur map even for J. My guess is that the answer is negative for all non reflexive dual spaces, but my guesses have a probability of only around 1/2 of being correct. – Bill Johnson Jan 13 '10 at 16:49

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.