2
$\begingroup$

I'm working my way through Billingsley's Probability and Measure, and I've got a question on Problem 26.5: Show by Theorem 26.1 (Riemann-Lebesgue Theorem) and integration by parts that if $\mu$ has a density $f$ with integrable derivative $f'$ then $\phi(t)=o(t^{-1})$ as $|t|\rightarrow\infty$. Here $\phi(t)$ is the characteristic function.

Integration by parts gives me $$\phi(t)=\int_{-\infty}^\infty f(x)e^{itx}dx=\left[f(x)e^{itx}/it\right]^\infty_{-\infty}-\int_{-\infty}^\infty f'(x)e^{itx}/it dx$$

Multiplying both sides by $t$ then gives $t\phi(t)=-\left[f(x)e^{itx}i\right]^\infty_{-\infty}+\int_{-\infty}^\infty f'(x)e^{itx}i dx$$

We want to show the right hand side goes to 0 as $t\rightarrow\infty$. The second term goes to 0 as $|t|\rightarrow\infty$ since $f'$ is integrable (for this we can use the same argument used to prove the Riemann-Lebesgue Lemma). But how about the first term? Do we know that $f(x)\rightarrow 0$ as $|x|$ goes to $\infty$? I know that generally for $f(x)$ to be integrable this is not required. Is it required if $f'$ is integrable? If so, can somebody please give a hint as to why this is so?

Thanks!

$\endgroup$
1
$\begingroup$

According to some form of the fundamental theorem of calculus (Theorem 7.21 of Rudin's Real and Complex analysis), if $f$ is differentiable and its derivative is integrable, then $f(x) = f(0) + \int_0^x f'(t)dt$ , which converges (by dominated convergence theorem for instance) to $f(0)+ \int_0^{\pm\infty} f'(t)dt \in \mathbb R$ when $x$ goes to $\pm\infty$. These two limits must be $0$ otherwise $f$ couldn't be a density.

$\endgroup$
  • $\begingroup$ I don't understand why these limits must be 0. See link for example. $\endgroup$ – gruyb Sep 15 '16 at 9:29
  • $\begingroup$ In your case, the fundamental theorem of calculus implies that there is a limit in $+\infty$ (unlike in the counterexample you mentioned). Now if this limit, which we call $\lambda$, is not 0, then there is no way $f$ could integrate to $1$. Hence $\lambda = 0$. Same reasoning for $-\infty$. $\endgroup$ – justt Sep 15 '16 at 9:34
  • $\begingroup$ You're welcome ! $\endgroup$ – justt Sep 15 '16 at 9:52

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.