Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. 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

There are two equivalent definitions of a locally convex topological vector space. Note that since the vector space is topological, addition by a fixed vector x and multiplication by a fixed scalar r are both homeomophisms, which means it is enough to give a neighborhood basis at the vector 0.

The first definition is that $0$ has a local base consisting of open sets that are convex (if $x$ and $y$ are in $S$, then $\lambda x+(1-\lambda)y$ is also in $S$ for every $\lambda\in[0,1]$), balanced, (if $x\in S$, then $r x\in S$ for every $r\in\mathbb{F}$ with $|r|=1$, where $\mathbb F$ is either the real or complex numbers), and absorbing (for any vector $y$ there exists a $\lambda\in (0,\infty)$ such that $\lambda S$ contains $y$, i.e. $S$ absorbs $y$).

The second definition is that $0$ has a local base given by the ''balls'' of radius $r$ for each $r$ of each semi-norm in some fixed collection of semi-norms (a semi-norm is just like a norm except that non-zero vectors can have a non-zero semi-norm). The two definitions are equivalent in one direction because "balls" of semi-norms are convex, balanced and absorbing, and in the other because we can define for any convex, balanced, absorbing set $S$ the semi-norm $f_S$ by assigning to $x$ the "least" (infimum) $\lambda$ for which $x$ is in $\lambda S$.

My question is: what are some examples of topological vector spaces for which $0$ does have a local base of convex open sets, but that no local base consists of convex sets that are also balanced and absorbing?

share|cite|improve this question
up vote 5 down vote accepted

Unless I am misunderstanding the situation, there is no such example. That is, if $V$ is a topological vector space (over $\mathbb R$ or $\mathbb C$) which has a basis of neighbourhoods of the origin consisting of convex sets, then it also has a basis of neighbourhoods of the origin consisting of balanced convex sets.

Also, any neighbourhood of $0$ is automatically absorbent (because of continuity of scalar multiplication).

My reference is Robertson and Roberston, Topological Vector Spaces (Cambridge Tracts in Mathematics and Mathematical Physics, 53), which is a kind of summary of Bourbaki's Topological Vector Spaces volume.

The proof is not hard: one first uses continuity of scalar multiplication to show that any neighbourhood of $0$ contains a balanced neighbourhood. Now if $U$ is a convex n.h of $0$, and $W$ is a balanced n.h. of $0$ contained in $U$, then $V:= \cap_{|x| \geq 1} x U$ contains $W$, so is a n.h., is convex (being the intersection of convex sets), is balanced (by construction), and lies in $U$ (again by construction --- consider $x = 1$). Thus $V$ is a convex balanced n.h. of $0$ contained in $U$.

share|cite|improve this answer

NB: As Matt notes in a comment below, the following does not really work!

Consider a vector space $V$ of dimension at least $2$, pick a non-zero proper subspace $W\subseteq V$, and consider the unique linear topology on $V$ for which $\{W\}$ is a basis of neighborhoods of $0\in V$.

If you want an Hausdorff example, let now $V$ be an infinite dimensional vector space, and consider a decreasing sequence $(W_n)_{n\geq1}$ of non-zero subspaces of $V$ such that $\bigcap_{n\geq1}W_n=0$. Now put on $V$ the unique linear topology which has $\{V_n\}_{n\geq1}$ as a basis of neighborhoods of $0\in V$.

share|cite|improve this answer
In your first example, let $v \in V$ not lie in $W$. Then $0 \times v = 0 \in W$, but there is no n.h. $U$ of $0$ such that $U \times v \subset W.$ So it seems that scalar multiplication is not continuous. Am I blundering again? – Matt E Nov 15 '10 at 4:29
Argh. You are of course right! A tvs is always connected, and my spaces aren't. – Mariano Suárez-Alvarez Nov 15 '10 at 4:38

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.