4
$\begingroup$

A smooth function on $\mathbb{R}^n$ is called slowly increasing if each of its derivatives is polynomially bounded. It seems that the collection of such functions is denoted as $O_M$.

Obviously, $O_M$ is a vector subspace of $\mathcal{S}'(\mathbb{R}^n)$ and a multiplicative subalgebra of $C^\infty(\mathbb{R}^n)$.

I am aware that the Schwartz space $\mathcal{S}(\mathbb{R}^n)$ is a nuclear (tamed) Fréchet space, so that $\mathcal{S}'(\mathbb{R}^n)$ is also nuclear with the strong dual topology. Also, topology of uniform convergence of all derivatives on any compact set is a Fréchet topology on $C^\infty(\mathbb{R}^n)$ but I don't know much of the details. Moreover, I have difficulty finding a detailed description of a topology on $O_M$.

So, I would like to ask here two questions.

Is $C^\infty(\mathbb{R}^n)$ also nuclear as a Fréchet space? The Wiki page on Montel spaces says that $C^\infty(\mathbb{R}^n)$ is Montel, and I am aware that a nuclear Fréchet space is Montel. So, I guess that I am correct?

Exactly what topology is "natural" on $O_M$? Perhaps it can be give some LF topology which is hopefully nuclear and also Montel?

Could anyone please provide any information or reference? I will split this post into two if that makes more sense.

$\endgroup$
3
  • 1
    $\begingroup$ Note that if the k-th derivative has polynomial growth so do all the other lower order derivatives. This space looks like a projective limit of spaces. $\endgroup$ Commented Oct 25 at 19:43
  • $\begingroup$ @LiviuNicolaescu I was not aware of such implications, which is not explicitly stated in the textbook I have either. In any case, could you provide more details or any reference? $\endgroup$
    – Isaac
    Commented Oct 25 at 21:01
  • 1
    $\begingroup$ Iit is an intersection of various spaces. Note that if the $n+k$-the derivative grows at most linearly then the $n$.-th derivative grows at most like a polynomial of degree $k$ $\endgroup$ Commented Oct 26 at 14:47

1 Answer 1

4
$\begingroup$

I think $C^{\infty}(\mathbb{R}^n)=\mathscr{E}(\mathbb{R}^n)$ is indeed nuclear, if I remember well. The reference to check for that would be the book by Trèves "Topological Vector Spaces, Distributions and Kernels".

For $\mathscr{O}_M(\mathbb{R}^n)$, the standard topology is the locally convex one defined by the (uncountable) collection of seminorms $$ f\mapsto\rho(gf) $$ where $g$ ranges over Schwartz space $\mathscr{S}(\mathbb{R}^n)$, and where $\rho$ ranges over continuous seminorms on $\mathscr{S}(\mathbb{R}^n)$. The topology of $\mathscr{O}_M(\mathbb{R}^n)$ is rather subtle. This space was conjectured to be isomorphic to $s\widehat{\otimes}s'$ by Grothendieck. Here $s$ is the space of rapidly decreasing sequences and $s'$ is the dual space of sequences of at most polynomial growth. This conjecture was proved about thirty years later by Valdivia. A good reference for $\mathscr{O}_M(\mathbb{R}^n)$ is the book by Horváth "Topological Vector Spaces and Distributions".

Also, a good reference for the topologies of all these spaces is the relatively recent article by Christian Bargetz "Explicit representations of spaces of smooth functions and distributions" Journal of Mathematical Analysis and Applications, vol. 424, no. 2, 1491--1505, 2015.

$\endgroup$
1
  • 1
    $\begingroup$ Now I recall that I ran into the conjecture a few years ago. I did not know that it has been proven already. And I do know of the Treves book; I will take a close look again. Thank you so much! $\endgroup$
    – Isaac
    Commented Oct 25 at 21:49

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .