I've tried looking for this question on this site, but I can't seem to find it. But if anyone can direct me to it, that would be great. But I'll pose my question in the mean time.

As the title states, if $f^{-1}(\alpha, \infty)$ is open, then show that $f$ is lower-semicontinuous.

Starting with the basics. The definition according to my professor gave on lower-semicontinuous is the following:

Given a function $f: X \to \mathbb{R}$ on a topological space $X$, $f$ is lower-semicontinuous if, for any $x \in X$ and for any $\epsilon > 0$, there is a neighborhood $N$ of $x$ such that

$f(x) - \epsilon < f(x')$ for all $x' \in N$.

I proved the conversed of this statement, but for this direction I seem to be stuck, been thinking for it for an hour or two. To my understanding a neighborhood of a point $x$ is a subset of $X$ such that it contains an open set which has $x$.

To me, it seems that I must consider if $x \in f^{-1}(\alpha, \infty)$ or $x \in f^{-1}(-\infty, \alpha)$. I started with the consideration of $x \in f^{-1}(\alpha, \infty)$. So by definition of pre-image, $f(x) \in (\alpha, \infty)$. We have that $\alpha < f(x) < \infty$. So it seems to me that I want a neighborhood of $x$ such that it satisfies $f(x) - f(x') < \epsilon$ for all $x' \in$ nieghborhood of $x$. Fix $x \in X$. Then I went on the path of suppose that $x \in f^{-1}(\alpha, \infty)$ and using that using $f^{-1}(\alpha, b)$ where $b = f(x)$. Since $f^{-1}(\alpha, \infty)$ can be written as the union (index starting at $k$) of $f^{-1}(\alpha, k)$ for $k > \alpha$ and $k \in \mathbb{R}$, then $f^{-1}(\alpha, b)$ must be open. Then the problem is that I can't get a neighborhood of $x$ such that $f(x) - f(x') < \epsilon$. I get a neighborhood of $x$ but it has elements such that $f(x) - f(x') \not<\epsilon$. I'll currently will update the progress of my work. Hopefully I can progress somewhere. Thanks.


Suppose we have $x \in X$ and $\epsilon > 0$. Then $f^{-1}(f(x)-\epsilon,\infty)$ is our desired neighbourhood of $x$.

| cite | improve this answer | |
  • $\begingroup$ I did that awhile ago, but I came to some point where I was struck. Let me put what I have so far. I said that $f^{-1}(f(x)- \epsilon, \infty) = N_x$ (a neighborhood of $x$). Now, to show that $f$ is lower-semicontinuous, then we must find an existence of a neighborhood of $x$ such that $f(x) - f(x') < \epsilon$, but I'm thinking that $N_x$ can contain some $x'$ such that $f(x') > f(x)$. $\endgroup$ – MathNewbie Mar 22 '12 at 19:06
  • $\begingroup$ So what if $f(x')>f(x)$? In that case we have $f(x)-f(x')<0<\epsilon$. $\endgroup$ – Chris Eagle Mar 22 '12 at 19:26
  • $\begingroup$ I was focusing on the $x'$ so much that I forgot the objective of this proof. Thanks a bunch Chris. $\endgroup$ – MathNewbie Mar 22 '12 at 19:29
  • $\begingroup$ @Chris Eagle: I would like to ask your comments in the following question math.stackexchange.com/questions/482684/… Thank you for your kind help $\endgroup$ – blindman Sep 4 '13 at 5:00
  • $\begingroup$ @Chris Eagle If G = { $(x, y)$ belongs to $R^{2}$ such that $y = f(x)$} is closed in $R ^{2}$.then $f$ is lsc?? $\endgroup$ – sani May 6 '17 at 14:33

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.