# Why is the sequential criterion true in metric spaces? When does it fail in general?

I’ve read just the basics of some introductory analysis books and sometimes they show that we can characterize things like limits, continuity, compactness, etc. in terms of sequences.

I’ve heard that these sequential criteria hold for general metric spaces, but that in topology for example one encounters situations where sequences aren’t quite sufficient, or where it’s better to consider some other object.

My questions are:

1. Is there some intuition for why the sequential criterion holds in things like Euclidean space or general metric spaces, but not in some other spaces?
2. Does it simply have to do with the fact that we have a metric, and if so, why does the metric “induce” such sequential criteria (versus without a metric we may not)?
3. Is the notion of distance/metric captured in some way by sequences because approaching some value is equivalent to a sequence approaching that value?
4. Are there any ways by which we can determine whether a general space possesses these sequential criteria? They seem quite useful.
• Not every body learns the same and with the same names. You should write down clearly what does "sequential criterion" mean...though I suspect is Heine's definition of limit. – DonAntonio May 11 '20 at 23:13
• Oh ok, by "sequential criterion" I mean like in a metric space we have things like sequential compactness, the sequential criterion for continuity, and the sequential criterion for limits. – twosigma May 11 '20 at 23:16
• In a metric space each point has a countable local neighborhood base but this is not true in a general topological space. – Kavi Rama Murthy May 11 '20 at 23:28

Many of these hold for sequential spaces. These can be defined in a variety of equivalent ways. One simple way that uses no new terminology is that $$X$$ is sequential iff for each non-closed $$A\subseteq X$$ there is a sequence $$\langle x_n:n\in\Bbb N\rangle$$ in $$A$$ converging to a point of $$(\operatorname{cl}A)\setminus A$$. It turns out that this is also equivalent to the statement that continuity of functions on $$X$$ is determined by sequences: $$X$$ is sequential iff for every space $$Y$$ a function $$f:X\to Y$$ is continuous iff it preserves convergent sequences, i.e., iff $$\langle f(x_n):n\in\Bbb N\rangle$$ converges to $$f(x)$$ in $$Y$$ whenever $$\langle x_n:n\in\Bbb N\rangle$$ converges to $$x$$ in $$X$$.
If $$X$$ is second countable (i.e., has a countable base for the topology), then it is compact iff it is sequentially compact, as shown in the answer to this question, but that is more than is needed; for instance, the comments under the question show that they two are equivalent in Lindelöf Hausdorff spaces. (Every second countable space is sequential and Lindelöf, but a sequential Lindelöf space need not be second countable.) The comments also note, with a reference, that these types of compactness are equivalent for the weak topology on Banach spaces, which is a sequential only if the space is finite-dimensional.