0
$\begingroup$

In my professor's notes on Lebesgue Integration, the proof of the Theorem above says that we need to prove that for all measurable sets $E$ with finite measure, and for $\epsilon>0$ exists a function $g_\epsilon$, continuous with compact support, such that $\Vert\chi_E - g_\epsilon\Vert < \epsilon$. There exists a compact $K$ and open $G$ such that $K\subseteq E\subseteq G$ and $\lambda(G - K)<\frac{\epsilon}{2}$. If we find $g$ with $supp(g)\subseteq G$ and $\chi_K\leq g\leq 1$ then $$\Vert\chi_E - g\Vert = \int_K |\chi_E-g|d\lambda + \int_{G-K} |\chi_E-g|d\lambda + \int_{\mathbb{R}^n-G} |\chi_E-g|d\lambda$$ Because $\chi_E=1$ in $K$ and $1=\chi_K\leq g\leq 1$, then $g=1$, and the first integral is $0$, and because the support is contained in G, then $g$ is $0$ a.e. in $\mathbb{R}^n-G$ and evidently $\chi_E=0$ in this set, so the integral is 0 too. Then we get that $\Vert\chi_E-g\Vert\leq2\lambda(G-K)<\frac{\epsilon}{2}.$

My question: Where are the hypoteses of $g$ being continuous and with compact support used?

Thanks in advance.

$\endgroup$
2
0
$\begingroup$

The hypothesis isn't used anywhere.

Your proof only uses the hypothesis that $g$ is measurable, that $g|_K\equiv 1,$ $g|_{\mathbb{R}\setminus G}\equiv 0$ and $\|g\|_{\infty}\leq 1$.

Hence, the real question is: Does such a $g$ exist, if we also require $g$ to be continuous with compact support? I.e., given $K\subseteq G$ with $G$ open and $K$ compact, does there exist a continuous function with compact support such that $f|_K\equiv 1$ and $f|_{\mathbb{R}\setminus G}\equiv 0$? This is a special case of Urysohn's Lemma in locally compact spaces.

$\endgroup$
1
  • $\begingroup$ Oh, so i see that adding the restriction of a continous and with compact support $f$ still can be found, but these properties are anywhere used to complete the proof. I was having difficulties understanding it because then the proof proceeds to find such a function with those properties, but they aren't seem to be used. $\endgroup$
    – deiv
    Jan 22 '20 at 15:02

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.