Let $X$ and $Y$ be infinite dimensional normed linear spaces and let $S:Y^*\to X^*$ be a one-one linear operator. I want to show that $S$ can not be weak*-norm continuous.

My idea is to choose a sequence $(y_n^*)$ in $Y^*$ which weak* converges to some $y^*$ in $Y^*$ (for this Banach-Alaoglu theorem will have to be used) such that $S(y^*_n)\not\to S(y^*)$ in norm. But I am not getting how infinite dimension of $X$ and $Y$ (and of course $X^*,Y^*$) will help. Any suggestion will be appreciated.

  • $\begingroup$ $S$ is any one one linear operator $\endgroup$ – Anupam Oct 23 '18 at 9:38
  • $\begingroup$ @daw The identity map is norm-weak$^*$ continuous, not the other way around. $\endgroup$ – Theo Bendit Oct 23 '18 at 9:49

I don't think the result is true as stated.

Take $X=Y=H$, an infinite-dimensional separable Hilbert space. Then $X^*=Y^*=H$, and the weak$^*$-topology agrees with the weak topology. Now any injective compact operator contradicts the assertion, as compact operators map weak (hence weak$^*$)-convergent sequences into norm-convergent sequences.

As an example, fix an orthonormal basis $\{e_n\}$ and let $S$ be the linear operator induced by $Se_n=\frac1n\,e_n$.


The claim is true for bijective $S$, I have no idea how to extend to the general case.

Assume that $S:Y^*\to X^*$ is weak-star-to-strong continuous. By the Banach-Alaoglu theorem, $S$ maps bounded sets to relatively compact sets, hence it is compact.

If $S$ is bijective, then it is continuously invertible and compact, hence the space is finite-dimensional, a contradiction.

  • $\begingroup$ How can we say $S^{-1}$ is continuous? $\endgroup$ – Anupam Oct 24 '18 at 2:52
  • $\begingroup$ I read 'one-one' as bijective? Did you mean injective ? $\endgroup$ – daw Oct 24 '18 at 6:06
  • $\begingroup$ Yeah I meant injective. $\endgroup$ – Anupam Oct 24 '18 at 7:47
  • $\begingroup$ In the example how does compactness of the operator implies weak*-norm continuity? Please see An introduction to Banach Space Theory by R E Megginson exercise 3.44 (a) and (b). $\endgroup$ – Anupam Oct 24 '18 at 8:00

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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