2
$\begingroup$

In proving :

If $f$ is bounded variation on $[0,1]$, then $$F(x) = (1/x) \int_0^x f(t) dt$$ is of bounded variation on $[0,1]$.

I am attempting to prove: If $f$ is increasing, then $F$ is increasing.

I attempted to bypass the problem in numerous ways but I was unsuccessful. If I have the claim, I solve the problem. Can anyone give me a hint? Any help would be appreciated!

$\endgroup$
1
  • $\begingroup$ On top of my head: if $f$ were continuous, then it would hold $$F'(x)=\frac{f(x)-F(x)}{x}$$ and, since $f$ is increasing and $F(x)$ is the average of $f$ in the interval $[0,x]$, by continuity it follows $F(x)<f(x)$. Perhaps you can recycle this idea. $\endgroup$
    – user228113
    Commented Nov 5, 2016 at 0:11

3 Answers 3

7
$\begingroup$

We want to show that $x < y \Rightarrow F(x) \leq F(y)$, i.e. $$\frac{1}{x}\int_0^x f(t)\,\mathrm{d}t \leq \frac{1}{y}\int_0^y f(t)\,\mathrm{d}t$$ We note that $$\frac{1}{y}\int_0^y f(t)\,\mathrm{d}t = \frac{1}{y}\left[\int_0^x f(t)\,\mathrm{d}t+\int_x^y f(t)\,\mathrm{d}t\right]$$ Therefore, we can rewrite our inequality as $$\left(\frac{y}{x}-1\right)\int_0^x f(t)\,\mathrm{d}t \leq \int_x^y f(t)\,\mathrm{d}t$$ We multiply both sides by $\frac{1}{y-x}$ to get $$\frac{\frac{y}{x}-1}{y-x}\int_0^x f(t)\,\mathrm{d}t = \frac{1}{x}\int_0^x f(t)\,\mathrm{d}t \leq \frac{1}{y-x}\int_x^y f(t)\,\mathrm{d}t$$ The left-hand side is the average value of $f(t)$ on $(0, x)$, while the right-hand side is the average value of $f(t)$ on $(x, y)$. The left-hand side is therefore bounded above by $\sup_{t\in (0, x)} f(t)$, while the right-hand side is bounded below by $\inf_{t\in (x, y)} f(t)$. As $f$ is increasing, we therefore have that $$\frac{1}{x}\int_0^x f(t)\,\mathrm{d}t \leq \sup_{t\in (0, x)} f(t) \leq \inf_{t\in (x, y)} f(t) \leq \frac{1}{y-x}\int_x^y f(t)\,\mathrm{d}t$$ so we are done.

$\endgroup$
0
4
$\begingroup$

Calculate the derivative of $F$: $F'(x) = \frac{\int_0^x (f(x) - f(t)) dt}{x^2}$ almost everywhere. We see that if $f$ is non-decreasing then $F' \geq 0$.

$\endgroup$
9
  • $\begingroup$ If $f$ is discontinuous, then $F$ may be nondifferentiable. $\endgroup$
    – Michael L.
    Commented Nov 5, 2016 at 0:34
  • 1
    $\begingroup$ terrytao.wordpress.com/2010/10/16/… Theorem 6. $\endgroup$
    – mcd
    Commented Nov 5, 2016 at 1:03
  • 1
    $\begingroup$ Elegant solution; +1. $\endgroup$ Commented Nov 5, 2016 at 1:06
  • $\begingroup$ But i have to admit that there are still some details missing, but i'm certain one can make this argument work. Michael's argument is better if one is not familiar with measure theory. $\endgroup$
    – mcd
    Commented Nov 5, 2016 at 1:27
  • $\begingroup$ I'm not sure how much measure theory is required besides the intuition that a.e. nonnegativity of $F'$ implies that $F$ is increasing. Other than that, considering its brevity, it is a very nice proof. $\endgroup$
    – Michael L.
    Commented Nov 5, 2016 at 1:31
0
$\begingroup$

A simple proof, using a linear substitution: For $0 < x < y$ is $$ \begin{align} \boxed{F(x)} &= \frac 1x \int_0^x f(t) \, dt \\ &= \frac 1y \int_0^y f(\frac xy s) \,ds \quad \text{(substitution $t = \frac xy s$)} \\ &\boxed{\le} \frac 1y \int_0^y f(s) \, ds \quad \text{(because $f$ is increasing)}\\ &= \boxed{F(y)} \, . \end{align} $$

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .