8
$\begingroup$

Motivation: I am doing functional analysis on locally convex spaces for the first time and I would like to know when I am allowed to characterise limit points and continuity sequentially. (This may well be a silly question.)

I phrase my question abstractly, but I mostly care about $\mathcal{C}_0^\infty(\mathbb{R}^n)$.

Let $(X_n)$ be an increasing sequence of first-countable locally convex topological vector spaces and let $X=\cup_n X_n$.

If we topologise $X$ with the finest locally convex topology such that the inclusions $X_n\rightarrow X$ are all continuous, is $X$ necessarily a first-countable space?

(A local base for such a topology is given by the the collection of all balanced, convex, absorbent sets whose intersections with every $X_j$ is open in $X_j$.)

I am currently reading from Reed & Simon Methods of Mathematical Physics, I cannot seem to find too many modern and systematic treatments of locally convex vector spaces!

$\endgroup$
  • $\begingroup$ Related to this $\endgroup$ – Noix07 Mar 18 at 13:23
9
+50
$\begingroup$

One can sharpen Alex Ravsky's answer:

Recall that a topological vector space is metrizable if and only if it is first countable.

Let $X_n \subsetneqq X_{n+1} \subsetneqq \cdots$ be a strictly increasing sequence of Fréchet spaces such that each $X_n$ carries the topology induced by $X_{n+1}$. Then $X = \varinjlim X_n$ is not metrizable.

The point is that $X$ is complete (see e.g. Schaefer-Wolf, Proposition 6.6 and its corollary). If $X$ were metrizable, then $X = \bigcup_{n=1}^\infty X_n$ would be a union of countably many nowhere dense sets, contradicting the Baire category theorem.

$\endgroup$
  • $\begingroup$ Thankyou, I found and convinced myself of this result today. Fortunately we can often characterise continuity sequentially for such spaces in spite of the lack of countability, thanks to the following result: A linear map $T:X\rightarrow Y$ (where $X=\lim_{\rightarrow}X_n$ is given the inductive limit topology and $Y$ is locally convex), is continuous if and only if its restriction to every $X_n$ is continuous. Main point: If the spaces $X_n$ are first-countable (as they are in the case of the test functions), we can use sequences to characterise continuity of these restrictions. $\endgroup$ – Goonfiend Apr 13 '13 at 9:48
  • $\begingroup$ @SeanGomes why are the $X_n$'s nowhere dense? $\endgroup$ – user153312 Jul 28 '15 at 7:58
2
$\begingroup$

It seems that usually $X$ is not first countable and even the direct limit $\mathbb{R}^\infty$ of $\mathbb{R^n}$ has the uncountable character, equal to the small cardinal $\mathfrak d$. If you need references, I shall ask my scientific consultant for them.

$\endgroup$

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.