4
$\begingroup$

I was reading the solution to a functional inequality in an article when the author made the following remark without giving any proof: let $f(x): [0, \infty]\to[0, \infty]$ be locally integrable and such that $$\left(\int_0^t f(x)dx\right)^2 \ge \int_0^t f(x)^3dx$$ for all $t>0$. Then, the following statement is true:

$\int_0^t f(x)^\gamma dx \le \frac{1}{\gamma +1}\left(2\int_0^t f(x)dx\right)^{(\gamma + 1)/2}$ for all positive $t$ and $\gamma \in [1,3]$.

Again, there is no proof in the article, so I don't know if this is fairly easy to prove or very involved. One thing that might be worth mentioning is that the inequalities above become exact when $f(x)=x$. I am wondering if anyone has an idea or have seen these before.

$\endgroup$
1
  • 1
    $\begingroup$ Could you post the paper with the improved constant? $\endgroup$ Commented May 25, 2020 at 11:02

1 Answer 1

2
$\begingroup$

I don't know about the precise multiplicative prefactor $\frac{2^{\frac{\gamma+1}2}}{\gamma+1}$, but I can tell you how to get $$ \int _0^t f^\gamma\leq \left(\int_0^t f\right)^{\frac{\gamma+1}{2}}. $$ This is a simple interpolation inequality: for any fixed $\gamma\in (1,3)$ (the boudnary cases $\gamma=1,3$ are immediate) there is a unique $\theta=\theta(\gamma)\in(0,1)$ such that $$ \frac 1\gamma=\theta\frac{1}{1}+(1-\theta)\frac{1}{3}. $$ By standard interpolation you get $$ \|f\|_{\gamma}\leq \|f\|_{1}^{\theta} \|f\|_{3}^{1-\theta}. $$ Your assumption is equivalent to the "$L^3$ by $L^1$ control" $\|f\|_3\leq \|f\|_1^{\frac{2}{3}}$, which then immediately implies $$ \|f\|_{\gamma}\leq \|f\|_{1}^{\theta+\frac 23(1-\theta)}. $$ Leveraging the explicit value of $\theta=\theta(\gamma)$, some straightforward algebra then finally gives the exact exponent $\theta+\frac 23(1-\theta)=\frac{\gamma+1}{2}$.

$\endgroup$
5
  • $\begingroup$ Could you post or link the paper with the assertion on the improved constant? Thank you $\endgroup$ Commented May 25, 2020 at 10:27
  • $\begingroup$ @Giorgio_Metafune: I guess this is a comment for the OP's question, rather than for my answer? $\endgroup$ Commented May 25, 2020 at 10:58
  • $\begingroup$ @leo_monsaingeon Yes, true! Sorry, I wrote in the wrong place $\endgroup$ Commented May 25, 2020 at 11:01
  • $\begingroup$ @leomonsaingeon, Yes, the general integral relationship is straightforward (directly applying Holder to the condition on $f(x)$). However, the constant is critical. $\endgroup$
    – Ivan
    Commented May 25, 2020 at 18:17
  • 1
    $\begingroup$ Perhaps you should make this clear by editing your question? Indeed your prefactor is $<1$ for $\gamma\in(1,3)$ so just Hölder-ing the problem is suboptimal, and I have no clue about how to get there... Is the paper focussed on small $t\to 0$, on large $t\to\infty$, or on any $t$? $\endgroup$ Commented May 25, 2020 at 19:29

You must log in to answer this question.

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