1
$\begingroup$

Dear All,

As we know that this following Miskowski's integral inequality is true for $1\leq p<\infty$

$ [\int_{S_1}|\int_{S_2}F(x,y)d\mu_1(x)|^pd\mu_2(y)dy]^{1/p} \leq \int_{S_2}[\int_{S_1}|F(x,y)|^pd\mu_2(y)]^{1/p}d\mu_1(x) $

I think that Minkowski's integral inequality is not right for the case $p=\infty$. And I am trying to find a counter-example when $p=\infty$ but I have not had luck. Does anyone know any counter example for that case OR if you know how to prove that it is still right. Thank you very much and I really appreciate your help.

Phi

$\endgroup$
7
  • $\begingroup$ The wikipedia says that it IS true for $p=\infty" (en.wikipedia.org/wiki/Minkowski_inequality), and who are we to argue? $\endgroup$
    – Igor Rivin
    Dec 15, 2011 at 21:30
  • 1
    $\begingroup$ I'm confused. I see expressions of the form $[\ldots]^p$ in you formula. How should I interpret them when $p=\infty$? $\endgroup$ Dec 15, 2011 at 22:00
  • 1
    $\begingroup$ The wikipedia says that the statement should be modified "in the obvious way" for $p=\infty.$ Presumably to replace the usual integrals by essential suprema (or garden variety suprema for continuous functions). $\endgroup$
    – Igor Rivin
    Dec 15, 2011 at 22:09
  • 1
    $\begingroup$ In any case, if you still have questions, I think they would belong better on math.stackexchange.com where you will find many people able and willing to clarify. $\endgroup$
    – Yemon Choi
    Dec 15, 2011 at 22:19
  • 1
    $\begingroup$ Phi, as Andre points out, what you wrote down doesn't make sense for $p=\infty$. There is an analogue for $p=\infty$: $$\bigg\|\int f(x,y)\; dy\bigg\|_\infty\le \int \big\|f(x,y)\big\| _\infty\; dy$$ This is not in every analysis book, but it is in Folland's Real Analysis (p. 194). $\endgroup$
    – B R
    Dec 15, 2011 at 23:54

1 Answer 1

1
$\begingroup$

Presumably we want: $$ \sup_{y \in S_1} \left|\int_{S_2} F(x,y) d\mu_1(x)\right| \le \int_{S_2}\sup_{y \in S_1} |F(x,y)| d\mu_1(x) $$ where $\sup$ is actually essential supremum with respect to $\mu_2$ on $S_1$.

Notation a bit confusing, since here (unlike the wikipedia version) $\mu_1$ is on $S_2$ and $\mu_2$ is on $S_1$.

$\endgroup$

Not the answer you're looking for? Browse other questions tagged or ask your own question.