Let $\{\omega_i\}_{i\in I}$ be a non-empty set of increasing (not necessarily strictly) continuous functions preserving $0$. Then, for each $i \in I$ define the space $$ C_{\omega_i}(\mathbb{R}^n,\mathbb{R}^d):= \left\{ f \in (\mathbb{R}^n,\mathbb{R}^d):\, \|f\|_{\omega_i,\infty}<\infty \right\} \mbox{ where } \|f\|_{\omega_i,\infty}:= \sup_{x \in \mathbb{R}^n} (\omega_i(\|x\|)+1)^{-1}\|f(x)\|. $$ Then this is a Banach space since the map $f \mapsto f (\omega_i(\|x\|)+1)$ is clearly an isometry with $C_0(\mathbb{R}^n,\mathbb{R}^d)$ for the sup-norm, and the maps between $C_{\omega_i}(\mathbb{R}^n,\mathbb{R}^d)$ and $C_{\omega_j}(\mathbb{R}^n,\mathbb{R}^d)$ can be defined similarly by rescaling analogously. This makes $I$ into a poset with $$ i\leq j \mbox{ if and only if } \sup_{x \in \mathbb{R}^n}\omega_i(\|x\|) \leq \sup_{x \in \mathbb{R}^n}\omega_j(\|x\|). $$ Thus, we can define the LCS colimit of this co-cone $\left\{C_{\omega_i}(\mathbb{R}^n,\mathbb{R}^d)\right\}$.

Now for the question, if $\{\omega_i\}_i$ is taken to be the collection of all monotonically increasing and continuous functions identifying $0$ (i.e.: $\omega(0)=0$) then does $\operatorname{co-lim}_i C_{\omega_i}(\mathbb{R}^n,\mathbb{R}^d)$ contain all uniformly continuous functions?

Note: Here the colimit is in the LCS sense.


Maybe, I miss something, but the answer seems to be easy: If $f:\mathbb R^n\to\mathbb R^d$ is continuous with $f(0)=0$ you can define the weight function $$\omega(r)=\sup\{\|f(x)\|: \|x\|\le r\}$$ which is obviously increasing with $\omega(0)=0$. Moreover, it is continuous at $r\ge 0$ because of the uniform continuity of $f$ on the compact set $\{x\in\mathbb R^n:\|x\|\le r+1\}$. Then $f\in C_\omega(\mathbb R^n, \mathbb R^d)$ holds. The extra assumption $f(0)=0$ can be removed by applying this to $\tilde f(x)=f(x)-f(0)$.

This shows that $C(\mathbb R^n, \mathbb R^d)$ is the co-limit of $\{C_\omega(\mathbb R^n, \mathbb R^d)\}$ in the category of vector spaces. Whether this holds in LCS is a different question.

  • $\begingroup$ You mean in the category of topological vector spaces? In that case $C(\mathbb{R}^n,\mathbb{R}^d)$ has compact-convergence topology or some (non-metrizable) final topology? Also, are these examples where, $I$ is say countable and fully ordered (besides the space of functions of moderate growth?) $\endgroup$ – AIM_BLB Jul 31 '20 at 8:57
  • $\begingroup$ You wrote: Does the co-limit contain all uniformly continuous functions? -- This sounds very much like an algebraic question (where the co-limit is the union). $\endgroup$ – Jochen Wengenroth Jul 31 '20 at 10:31
  • $\begingroup$ Ah, I see the issue; my error. But I'm mostly looking in the LCS category (I made an edit to help clarify), I'm sorry abou the confusion $\endgroup$ – AIM_BLB Jul 31 '20 at 10:46
  • 1
    $\begingroup$ Note that a uniformly continuous function on a normed space has a linear growth, $\|f(x)\|\le a\|x\|+b$. So you don't need such a wide class of weights $\{\omega_i\}$. 0n the other hand, adding $1$ in the denominator in the definition of the norms means you do not care to much about being uniformly continuous or just continuous. $\endgroup$ – Pietro Majer Jul 31 '20 at 14:38

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.