In terms of mappings on topological spaces, are preserving limits and preserving cluster points equivalent?

25 Views Asked by At

This question comes from one of my exercises, which asks to prove that a mapping on a topological space is continuous if and only if it preserves cluster points. That is to say, for a topological space $X$, $Y$, and a mapping $f:X\to Y$, suppose $\xi:D\to X$ is a net on $X$ and $x$ is a cluster point of this net, then $f(x)$ is a cluster point of the net $f\circ \xi$ on $Y$.

Now that we know a mapping on a topological space is a continuous mapping if and only if for any net $\xi:D\to X$ on $X$, if $\xi\to x$, then the net $f\circ \xi$ on $Y$ converges to $f(x)$, and if $x$ is a cluster point of the net $\xi$ on $X$, then there exists a subnet $\xi\circ\eta$ of $\xi$ such that $\xi\circ\eta\to x$, it is clear that preserving cluster points follows from preserving limits.

But I have no idea how to prove the other direction, and I even began to doubt its correctness because intuitively, the property of preserving continuity is obviously stronger than the property of preserving cluster points. Are they really equivalent? Or have I made a logical mistake in the relationship between these three propositions?

1

There are 1 best solutions below

1
On BEST ANSWER

The key idea you seem to be missing is that if a net fails to converge to a point $x$, then there is a subnet such that $x$ is not a cluster point (since there must be some neighborhood $U$ of $x$ which the net is frequently not in so there is a subnet that is never in $U$). So suppose $f:X\to Y$ fails to preserve limits, so there is some net $\xi:D\to X$ which converges to $x\in X$ such that $f\circ\xi$ does not converge to $f(x)$. Then there is a subnet $f\circ\xi\circ\eta$ of $f\circ\xi$ which does not have $f(x)$ as a cluster point. Since $\xi\circ\eta$ still converges to $x$ (and in particular has $x$ as a cluster point), this shows that $f$ fails to preseve clulster points.