4

Suppose that $f\colon [0,\infty)\to \mathbb{R}$ is a continuous strictly increasing function with $f(0)=0$. Prove that for $a,b \gt 0$ we have Young's inequality

$$ ab \le \int_0^af(x)dx+\int_0^bf^{-1}(x)dx$$ and that equality holds if and only if $b=f(a)$.

In this post the following proof is given by kobe:

Let $C$ be the graph of $v = f(u)$ over the interval $[0,f^{-1}(b)]$. If $f(a) > b$, then $f^{-1}(b) < a$, in which case

\begin{align}\int_0^a f(x)\, dx + \int_0^b f^{-1}(x)\, dx &= \int_0^{f^{-1}(b)} f(x)\, dx + \int_0^b f^{-1}(x)\, dx + \int_{f^{-1}(b)}^a f(x)\, dx\\ &= \int_C u\, dv + v\, du + \int_{f^{-1}(b)}^a f(x)\, dx\\ &= \int_C d(uv) + \int_{f^{-1}(b)}^a f(x)\, dx\\ &= bf^{-1}(b) + \int_{f^{-1}(b)}^a f(x)\, dx\\ &> bf^{-1}(b) + b(a - f^{-1}(b))\\ &= ab \end{align}

Similarly if $f(a) < b$, then $\int_0^a f(x)\, dx + \int_0^b f(x)\, dx > ab$. If $f(a) = b$, then $$\int_0^a f(x)\, dx + \int_0^b f^{-1}(x)\, dx = \int_C u\, dv + v\, du = \int_C d(uv) = af(a) = ab.$$

I would like to focus on the following equality:

$$\int_0^{f^{-1}(b)} f(x)\; dx=\int_C v\;du$$

We parameterize $C$ using the curve $$\gamma(x)=(u(x), v(x)):=(x, f(x)), \quad x\in [0, f^{-1}(b)]$$ then for definition of line integral we have that $$\int_0^{f^{-1}(b)}f(x)\; dx=\int_\gamma v\; du.$$

Question 1. Is $\gamma$ piecwise regular or regular?

For me the answers is yes, because the function $f$ is monotone and then $f'(x)$ exists finite for almost every $x\in [0, f^{-1}(b)]$.

Thus exists a null set on which $f'$ may be not exists.

Question 2 Do the hypotheses we have allow us to say something more about this set of zero measure, such as that it is finite?

Thanks!

Jack J.
  • 1,140
  • 1
    According to Frigyes Riesz, for every set $E$ of measure 0, one can build a non-decreasing function $f$ which does not have a finite derivative on any point of $E$. – Gribouillis Jan 13 '25 at 16:34
  • I think it is a the beginning of the book 'Functional Analysis' by Riesz and Nagy – Gribouillis Jan 13 '25 at 16:55
  • @GribouillisThanks for your attention, but how does this solve my question? – Jack J. Jan 13 '25 at 17:11
  • I think it answers Question 2 in the sense that the set of zero measure can be virtually any set of zero measure. – Gribouillis Jan 13 '25 at 17:56
  • Using line integrals seems unnecessarily complicating the question. One could prove in this order for $f$ being: $$C^1\implies\text{piecewise linear}\implies C.$$ – Ѕᴀᴀᴅ Jan 14 '25 at 01:32
  • @ЅᴀᴀᴅI don't understand why the line integral is a complication. How can your suggestion help me solve the question? Could you be more detailed? Thanks for your attention. – Jack J. Jan 14 '25 at 07:41
  • @JackJ. I didn't post an answer since it wouldn't answer the two questions in the OP. – Ѕᴀᴀᴅ Jan 14 '25 at 08:26
  • 1
    I don't see how that line integral makes sense as it stands. In general, the path, as parameterized, need not be rectifiable -- e.g. if $f$ is a Cantor-Vitali type function. You may argue that one can re-parameterize it as a Lipschitz curve b/c it is of bounded variation... but, what is the point then: we were supposed to prove a deep but not too difficult to prove inequality. – Behnam Esmayli Jan 16 '25 at 02:11
  • @BehnamEsmayliBut then is kobe's answer wrong? What is the difference between kobe's answer and this revisitation I made regarding the parameterizations? – Jack J. Jan 19 '25 at 11:39

0 Answers0