1
$\begingroup$

Let $f: \mathbb{R} \to \mathbb{R}$ be increasing, right-continuous and nonnegative. Define the Lebesgue-Stieltjes measure $\mu_f$ by $$ \mu_f(a,b] = f(b) - f(a). $$

When does $\mu_f\{b\} = 0$?

I think it is true for linear such $f$, since $f(0) = 0$. Proof: Let $f$ be linear and let $\epsilon > 0$. Then $$ \mu_f\{b\} \leq \mu_f(b-\epsilon, b] = f(b) - f(b - \epsilon) = f(\epsilon). $$ Since this holds $\forall \epsilon > 0$, $$ \mu\{b\} = \lim_{\epsilon \to 0^+} \mu_f\{b\} \leq \lim_{\epsilon \to 0^+} f(\epsilon) = f(0) = 0, $$ where I used right-continuity of $f$ in the second-to-last equality and linearity of $f$ for the last.

Hence $\mu_f\{b\} = 0$ for linear such $f$

It looks like it's also true for affine such $f$. Is there a more general result?

Update Thanks to Cain's comment, we also get $\mu_f\{b\} = 0$ for continuous such $f$.

$\endgroup$
  • 5
    $\begingroup$ If you'll inspect your claim you will notice you've only used the fact that a linear function is continous. Try to generalize this idea. $\endgroup$ – Cain Sep 18 '15 at 20:03
  • $\begingroup$ @Cain Ah, good eye. I see it. Thanks for the insight! $\endgroup$ – bcf Sep 18 '15 at 20:09
  • 1
    $\begingroup$ even more, notice that you just need right continuity. $\endgroup$ – L.F. Cavenaghi Sep 18 '15 at 20:17
  • $\begingroup$ @LeonardoFranciscoCavenaghi hmm, okay. Since every Lebesgue-Stieltjes function is right-continuous, we get $\mu_f\{b\} = 0$ for any such $f$?? I seemed to have thought it was possible that $\mu_f\{b\} > 0$ for some $f$. $\endgroup$ – bcf Sep 18 '15 at 20:22
  • 1
    $\begingroup$ @L.F.Cavenaghi is mistaken here. Points can have positive measure even when $f$ is right continuous; consider $f = 1_{[0,\infty)}$ which is a point mass at 0. They will have zero measure iff $f$ is also left continuous. Measurability is not relevant; indeed, every Lebesgue-Stieltjes function is measurable. $\endgroup$ – Nate Eldredge Oct 31 '17 at 15:53
3
$\begingroup$

Hopefully someone finds this later answer useful. What is actually true (see exercise 1.28 in Folland's Real Analysis) is that $$\mu_F(\{a\})=F(a)-F(a-),$$ where $F$ is increasing and right-continuous. It need not, of course, be true for such $F$ that $F(a-)=F(a)$. Consider $$ F(x)=\begin{cases} 0&\text{if}~x<1\\ 1&\text{if}~x\geq 1 \end{cases}. $$ Then $F$ is increasing and right continuous, but $$\mu_F(\{1\})=F(1)-F(1-)=F(1)-\lim_{x\nearrow 1}F(x)=1-0=1>0.$$

$\endgroup$

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.