If $X$ and $Y$ are topological vector spaces over $\mathbb R$, then a map $f:X\to Y$ is called uniformly continuous if for each neighborhood $V\subseteq Y$ of $0\in Y$, there exists a neighborhood $W\subseteq X$ of $0\in X$ such that for any $x,y\in X$ satisfying $x-y\in W$, it follows that $f(x)-f(y)\in V$.

It is not difficult to show that addition $+:X\times X\to X$ is a uniformly continuous function when $X\times X$ is endowed with the product topology. Also, for a given $\alpha\in\mathbb R$, the map $x\mapsto \alpha x$ is a uniformly continuous function from $X$ to $X$.

However, I suspect that scalar multiplication $\cdot:\mathbb R\times X\to X$ is not necessarily uniformly continuous when $\mathbb R$ is endowed with the usual topology and $\mathbb R\times X$ is endowed with the product tvs structure. Indeed, if $X=\mathbb R$ under the usual topology, then the function $(\alpha,x)\mapsto \alpha x$ from $\mathbb R^2$ to $\mathbb R$ is clearly not uniformly continuous.

Can anybody confirm this? I'm asking because several textbook references on topological vector spaces claim that both addition and multiplication are uniformly continuous functions, but I suspect uniform continuity of scalar multiplication can only be established in a restricted sense, i.e., when it is interpreted for a given scalar and not when $\mathbb R$ is considered as a tvs in its own right to be multiplied with $X$ to form a product tvs. In particular, I'm confused as to which of these two notions of “scalar multiplication” is appropriate when one is talking about the uniform continuity of such an operation.

  • $\begingroup$ I am not familiar with topological vector spaces but I suspect that we don't have a sequential criterion on these spaces? Because in any metric space the scalar multiplication is not uniformly continuous. $\endgroup$ – user169373 Dec 28 '14 at 22:01

You're right, as can be seen in the example you mention, multiplication $\mathbb R\times\mathbb R \to\mathbb R$. And you say this is "clearly" not uniformly continuous, so I don't know if you have a question about proving it. You can take as $V$ the interval $(-1,1)$. Each open $W\subset \mathbb R\times \mathbb R$ containing $(0,0)$ also contains $(\delta,0)$ for some $\delta>0$, but $\left(\frac1\delta+\delta\right)\cdot\frac1\delta - \frac1\delta\cdot\frac1\delta = 1\not\in (-1,1)$ even though $\left(\frac1\delta+\delta,\frac1\delta\right)-\left(\frac1\delta,\frac1\delta\right)\in W$.

You might want to consider for which subsets $A\subset \mathbb R$ the restriction $A\times X\to X$ is uniformly continuous. It is necessary and sufficient that $A$ be bounded.


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.