MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Suppose we have a set of regular functions defined on a product of metric spaces, for instance the Banach space of the smooth functions from $\mathbb R^n$ to $\mathbb C$. We know, thanks to the Taylor series, that such a function can 'almost' be written as a linear combination of products $f_1 \cdot \ldots \cdot f_n$, each of the $f_i$ a continuous function from $\mathbb R$ to $\mathbb C$, where the almost means we are using a density argument in the sup norm.

Is there a way to generalize such a situation to a generic product of metric spaces? In particular I am interested in the space of locally constant, compactly supported functions on powers of the $p$-adic field $\mathbb Q_p$ but also the general question sounds interesting. Any reference is greatly appreciated, thanks.

share|cite|improve this question
I can not think of a proof for $\mathbb{R}\times \mathbb{R}$ which can not be generalized directly to product of two metric spaces. – Anton Petrunin Jul 2 '12 at 14:28
The proof of Taylor in $\mathbb R \times \mathbb R$ that I know goes like this: Let $f : \mathbb R \times \mathbb R \to \mathbb R$ be given. Let $a\in \mathbb R \times \mathbb R$, and suppose $f$ is sufficiently regular at $a$. Let $h \in \mathbb R \times \mathbb R$ be small enough. Consider the function of one variable $g(t) = f(a+th)$. Apply the one-dimensional Taylor theorem to $g$. – Gerald Edgar Jul 2 '12 at 15:12
The Taylor series of a function doesn't always converge to it in the sup norm even in one dimension (take $e^x$). What exactly is the precise statement you want? – Qiaochu Yuan Jul 2 '12 at 15:28
A reasonable construction (for Banach space-valued functions defined on a product of metric spaces) is via partitions of unity. Given a continuous function $f:X\times Y\rightarrow (E, \|\cdot\|)$, and $\epsilon >0$, there is a uniform approximation $\epsilon$-close to $f$ of the form $f_\epsilon(x,y):=\sum_{i,j} f(x_i,y_j)\phi_i(x)\psi_j(y)$. – Pietro Majer Jul 2 '12 at 17:59
You may look at topological tensor products which yield representations of function spaces on products like $C(K \times L) = C(K) \tilde{\otimes}_\epsilon C(L)$ or $C^\infty (\Omega_1 \times \Omega_2)= C^\infty (\Omega_1)\tilde{\otimes} C^\infty (\Omega_2)$ (in the latter case you can choose any tensor topology due to nuclearity). – Jochen Wengenroth Jul 3 '12 at 7:35
up vote 2 down vote accepted

Is this the statement you want: any locally constant, compactly supported function from ${\bf Q}_p^n$ to ${\bf C}$ is uniformly approximated by linear combinations of products $f_1\cdots f_n$ where each $f_i$ is a locally constant, compactly supported function from ${\bf Q}_p$ to ${\bf C}$? Yes, this follows from the Stone-Weierstrass theorem for locally compact Hausdorff spaces. The linear combinations in question constitute a self-adjoint algebra of functions which separate points and separate each point from infinity. Therefore they are uniformly dense in the space of continuous functions vanishing at infinity on ${\bf Q}_p^n$.

share|cite|improve this answer
This is exactly what I was looking for, thanks a lot. – Niccolo' Jul 3 '12 at 10:34
You're welcome! – Nik Weaver Jul 3 '12 at 13:49

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.