I know the definition of an upper semicontinuous set valued function;

A function $f:X\rightarrow 2^Y$ is upper semicontinuous at a point $x \in X$ provided that if $V$ is an open set in Y containing $f(x)$ then there exists an open set $U$ of $X$ that contains $x$ such that if $t \in U$ then $f(t)\subset V$. The function $f$ is upper semicontinuous if $f$ is upper semiconintuous at all $x \in X$

I can use this definition for some proofs in my study of inverse limits however I have no intuitive idea of what this actually means. What are some functions (preferably in the form $f:[0,1]\rightarrow 2^{[0,1]}$) that are and are not upper semicontinuous and what makes them so?


Let $X,Y$ be topological spaces and $f:X\to 2^Y$ a mutlivalued function. Define the graph of $f$:

$$Gr(f)=\big\{(x,y)\subseteq X\times Y\ |\ y\in f(x)\big\}$$

With that the following is true:

Lemma. Assume that $f(x)$ is nonempty and closed for each $x\in X$. If $f$ is upper hemicontinuous then $Gr(f)$ is closed in $X\times Y$. If additionally $Y$ is compact then the converse holds as well: if $Gr(f)$ is a closed subset of $X\times Y$ then $f$ is upper hemicontinuous.

So in case $X=Y=[0,1]$ and each $f(x)$ is nonempty and closed then upper hemicontiunity is simply equivalent to $Gr(f)$ being closed. I'm pretty sure that with that you can find lots of examples and counterexamples.

One such counterexample would be:

$$f(x)=\begin{cases} \{0\} & x\in[0,\frac{1}{2}) \\ \{1\} & x\in[\frac{1}{2}, 1) \end{cases}$$

a counterexample pretty much copied from single-valued case. The graph is not closed, every value is closed, hence the function is not upper hemicontinuous.

  • $\begingroup$ OK, that helps, is this the intended meaning of the definition? Is it just a way to officially classify functions with closed graphs or is the fact that they have closed graphs an interesting result? So similar to I think of the $(\epsilon , \delta)$ definition of continuity to be a way of formalising "a small change in x gives a small change in y" is the definition of upper semi-continuity a way to formalise the idea of closed graphs? or is there a different meaning it's trying to convey? $\endgroup$ – Simon Goodwin Dec 20 '17 at 22:09
  • $\begingroup$ It's an interesting result. Note the strong assumption you have to make: each $f(x)$ has to be nonempty and closed. Also I believe that this sentence: "a small change in x gives a small change in y" is a good way of thinking about upper hemicontinuity (even though formally it is not clear what a small change between sets is). After all a multivalued function is upper hemicontinuous if and only if it is continuous in the upper Vietoris topology. $\endgroup$ – freakish Dec 20 '17 at 22:15
  • $\begingroup$ One last thing, you keep on saying hemicontinuous, I've always seen semicontinuous, are they both interchangeable? $\endgroup$ – Simon Goodwin Dec 21 '17 at 0:27
  • $\begingroup$ @SimonGoodwin yeah, that's the same thing. Although semicontinuity is also used in single-valued case. I'm using hemi to avoid potential ambiguity. $\endgroup$ – freakish Dec 21 '17 at 5:20

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.