0
$\begingroup$

This is a follow-up to this question.

Let $M$ be manifold (Hausdorff, countable base, finite dimensional, if it simplifies anything embedded in $\mathbb{R}^n$).

I'm interested in the topological vector space $\mathcal{C}_c^{\infty}(M)$ of smooth functions with compact support equipped with the Whitney-$\mathcal{C}^\infty$-topology (The strong one induced by the strong topologies on $\mathcal{C}^{k}(M)$ for finite $k$. Another description for this topology can be found in 4. at the linked question.)

If $M$ is noncompact $\mathcal{C}^{\infty}(M)$ is in general not first countable and restricting to $\mathcal{C}_c^{\infty}(M)$ doesn't make anything better.

Therefore, since I want $\mathcal{C}_c^{\infty}(M)$ to be first countable, I have to make it even smaller. The next attempt is the following: Let $K\subseteq M$ be a compact subset and consider the subspace of functions which vanish in $K$ $$V\colon=\{f\in\mathcal{C}_c^{\infty}(M)|f|_K=0\}.$$ Since $K$ is compact, this should be the "bigger part" of the whole $\mathcal{C}_c^{\infty}(M)$. Now take the quotient vectorspace $\mathcal{C}_c^{\infty}(M)/V$ with the quotient topology.

I reckon that $\mathcal{C}_c^{\infty}(M)/V$ is now first countable, but I did not succeed to prove it. Since it's a topological vector space, it suffices to construct a countable neighbourhood basis of $0$. Can someone point me in the right direction? Thanks.

$\endgroup$
1
+50
$\begingroup$

Here is a proof:

Let $K\subset U\subset L\subset M$ where $U$ is open in $M$ and $L$ is compact. Then $C^\infty_c(M) = C^\infty_c(M\setminus K) + C^\infty_c(U)$ by a partition of unit element. Moreover $C^\infty_c(U)\subset C^\infty_L(M):=\{f\in C^\infty_c(M): \text{supp}(f)\subseteq L\}$, and the latter space is a Frechet space and is closed in $C^\infty_c(M)$. Thus $C^\infty_c(M) = C^\infty_c(M\setminus K) + C^\infty_L(M)$, and $C^\infty_c(M\setminus K)\subseteq V$. So finally $$ C^\infty_C(M)/V = C^\infty_L(M)/(V\cap C^\infty_L(M)) $$ is a quotient of the Frechet space $C^\infty_L(M)$ modulo a closed subspace, and is thus Frechet itself.

$\endgroup$

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.