Assuming $\left\lbrace f_n\right\rbrace$ is measurable for all n, I was able to show using the fact that sequences of real numbers converge iff they are Cauchy that:

$\left\lbrace x: \lim f_n(x) \text{ exists} \right\rbrace =\bigcap_{k=1}^{\infty} \bigcup_{N=1}^{\infty} \bigcap_{m=N}^{\infty} \bigcap_{n=N}^{\infty} \left\lbrace x: f_n(x)-f_m(x) < \frac{1}{k} \right\rbrace \cap \left\lbrace x: f_m(x)-f_n(x) < \frac{1}{k} \right\rbrace$

Since $f_n$ is measurble for all n, $\left\lbrace x: f_n(x)-f_m(x) < \frac{1}{k} \right\rbrace $ should be a measurable set for all $n,m,$ and $k$, and then since there's a bunch of countable intersections and unions of all these sets, I should have that $\left\lbrace x: \lim f_n(x) exists \right\rbrace$ is measurable as well. What I don't understand is why $f_n$ being measurable immediately implies that $\left\lbrace x: f_n(x)-f_m(x) < \frac{1}{k}\right\rbrace $ is measurable. We just started learning about measurable functions and I know that $f:X \rightarrow Y$ is $M,N-$measurable if $f^{-1}(E) \in M$ for all $E \in N$ where $M$ is the sigma-algebra on $X$ and $N$ the sigma algebra on $Y$, but I don't see how to apply that definition here.

  • $\begingroup$ The difference of two measurable functions is measurable; hence $f_n-f_m$ is measurable. This, in turn, implies that $$\{x; f_n(x)-f_m(x)< \frac{1}{k} \} = (f_n-f_m)^{-1}((-\infty,1/k))$$ is measurable. $\endgroup$ – saz Oct 18 '15 at 19:16
  • $\begingroup$ With notations $f_n=f$, $f_m=-g$ you may use this answer. $\endgroup$ – A.Γ. Oct 18 '15 at 19:16
  • $\begingroup$ Thanks guys, I appreciate the help! $\endgroup$ – Brenton Oct 18 '15 at 22:49

An easier way to do this so that you won't have to take unions/intersections over both $m$ and $n$:

Define $f(x) = \limsup_{n \to \infty} f_{n}(x)$. Now, if $f_{n}(x)$ converges at $x$, it must converge to $f(x)$ (recall that convergence implies that your limit is equal to your limsup and liminf). Define $$ A_{m,n} = \{x: |f_{n}(x) - f(x)| < \frac{1}{m}\} $$ And notice that $$ \{x: f_{n}(x) \text{ converges}\} = \bigcap_{m \in \mathbb{N}} \bigcup_{N \in \mathbb{N}}\bigcap_{n \geq N} A_{m,n} $$ Showing that $f$ and consequetly $|f_{n} - f|$ are measurable isn't too hard, and if you're doing this problem, you've probably already proved this in your class/textbook.


For sake of completness I shall give a sketch for another proof.

$1)$ Show that the maps $x \mapsto \lim \inf f_n(x)$ and $x \mapsto \lim \sup f_n(x)$ are measurables.

$2)$ Show that the set if $f, g$ are measurable functions, then the set $\{x : f(x) = g(x)\}$ is measurable. To do this, use that since $\overline{\mathbb{R}}$ is Hausdorff, then the diagonal $\Delta := \{(x,x) : x \in \mathbb{R}\}$ is closed. Further, use thtat $f\times g(x) := (f(x),g(x))$.

$3)$ Recall that a $\lim f_n(x)$ exists if, and only if, $\lim \sup f_n(x) = \lim \inf f_n(x).$

What is the advantage of such approach? It is because you can change the counterdomain to any second countable Hausdorff space with Borel sigma algebra.


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.