Suppose that $(X,\tau)$ is already a locally convex TVS. Let us denote by $X'$, the space of all $\tau$-continuous linear functionals on $X$, the topological dual of $X$. For each $f\in X'$, define $p_f(x)=|f(x)|, x\in X$. Then $$\Gamma=\{p_f: f\in X'\}$$ is a separating family of seminorms on $X$. By a theorem of Rudin, there exists a locally convex topology on $X$, commonly denoted by $\sigma:=\sigma(X,X')$, with the property that every member of $\Gamma$ is $\sigma$-continuous. Specifically, $\sigma$ consists of arbitrary unions of the translates of finite intersections of the sets $\{x\in X: |f(x)|<\epsilon\}$, $f\in X'$ and $\epsilon>0$.

Question. Is $\tau=\sigma$? What is only clear to me (if Im correct) is that $\sigma\subset\tau$. I need some help on this.

  • $\begingroup$ Its already 12:30 am here. If somebody ask for clarification, my apology that I cannot entertain your question. I got to back @ this site by tomorrow. $\endgroup$ – Juniven Aug 17 '13 at 16:12
  • $\begingroup$ In general $\tau$ is strictly finer than $\sigma$. Think of normed spaces, then $\sigma$ is the weak topology. $\endgroup$ – timur Aug 18 '13 at 1:41

As timur pointed out this topologies in general are different. Indeed, take arbitrary infinite dimensional normable space $(X,\tau)$. We know that in such spaces there are neighbourhhods of zero which does not contain linear subspaces, e.g. unit ball. But as for the weak topology $\sigma(X,X')$ it is not true, because all neighbourhoods of zero contain infinnite dimensional subspace (see remark before theorem 3.12 in Rudin's Functional analysis)


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.