For a vector topological space, a sequence $(x_n)$ is Cauchy when $\forall$ neighbourhood $U$ of $0$, there exists $N$ such that $m,n>N$ implies $x_m-x_n\in U$.

For a metric topological space $(E,d)$, a sequence $(x_n)$ is Cauchy when $\forall \epsilon>0$, there exists $N$ such that $m,n>N$ implies $d(x_m,x_n)<\epsilon$.

QUESTION : are these two definitions equivalent in the case of a metric topological vector space ?

If the distance is invariant by translation, I can prove it. But in general no.

QUESTION : if the previous question has answer "no", what is the "good" definition in the intersection ?


The answer is : yes if the vector topological space is metrizable, that is IFF it has a countable basis of topology on every point.

See theorem 3.21 here : http://ma.huji.ac.il/~razk/iWeb/My_Site/Teaching_files/TVS.pdf (difficult)

And also the fact that a metrizable topological vector space has always a countable local basis of topology. This one is easy : if $A$ is closed in $V$, then consider the open sets $V_n=\{x\in X\,st\, d(A,x)<1/n\}$. We have $A=\bigcap_{n\geq 1}V_n$.

If you read French, I'll write the answer with some details here : http://laurent.claessens-donadello.eu/pdf/mazhe.pdf in the chapter "Topologie générale"


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.