My question is: Suppose that $f:X\rightarrow \mathbb{R}$ is lower semicontinuous and M is a real number. Define $f_M:X\rightarrow\mathbb{R}$ by $$ f_M(x) = \min(f(x),M). $$ Prove that $f_M$ is lower semicontinuous.

The definition that I am using is: A function $f$ is lower semi continuous on X if for all $x\in$X and every sequence $x_n\rightarrow x$, we have $$\lim_{n\rightarrow\infty} \inf f(x_n) \geq f(x).$$ Thank you in advance for any help!

  • $\begingroup$ There are a couple of different (though equivalent) ways to define lower semicontinuity; which one are you using? $\endgroup$ – Brian M. Scott Sep 3 '13 at 22:00

A real-valued function is lower semicontinuous if the inverse image of every set of the form $(-\infty,a]$ is closed (this is actually the definition in many real analysis textbooks, in some others an equivalence). Note that if $a>M$ then $f_M^{-1}((-\infty,a])=X$ which is trivially closed. In the other case, if $a<M$ then $f_M^{-1}((-\infty,a])=f^{-1}((-\infty,a])$ which is closed since you are assuming the lower semicontinuity of $f$. Therefore $f_M$ is lower semicontinuous.

With your definition: let $x$ be any element in $X$ and $\{ x_n \}_{n=1}^\infty$ a sequence converging to $x$. Note first that $f(y) \geq f_M(y)$ for every $y\in X$. Now, since $f$ is LSC we have $$ \liminf_{n\to\infty} f(x_n) \geq f(x) \geq f_M(x) \tag{1} $$ and since we also know that $f_M(x)\leq M$ we have $$f_M(x) \leq \min\{M,\liminf_{n\to\infty} f(x_n) \} \leq \liminf_{n\to\infty} ( \min\{M,f(x_n)\} ) = \liminf_{n\to\infty} f_M(x_n) \tag{2} $$ and since this holds for every $x\in X$ we can conclude that $f_M$ is LSC.

Try to justify (2) using the definition of $\liminf$. $$ \liminf_{n\to\infty} f(x_n) = \inf_{n\geq 0} \sup_{k\geq n}f(x_n) $$.

  • $\begingroup$ This was my proof: Suppose $M > f(x)$ then by definition of $f_M(x)$, $f_M(x)$ is lower semi-continuous. Next, suppose f(x) > M, then we obtain $$\lim_{n\rightarrow\infty} \inf M \geq M = f_M(x).$$ Hence $f_M(x) is lower semi-continuous. $\endgroup$ – RDizzl3 Sep 3 '13 at 22:40
  • $\begingroup$ If $M>f(x)$ then $f_M(x)=f(x)$ at $x$ but this does not guarantee the function is continuous, you are just evaluating in a point. The definition states that the inequality must hold at every point x. $\endgroup$ – Mauricio Tec Sep 3 '13 at 22:55

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.