Let $(X,B,\mu)$ and $(Y,C,\nu)$ be probability spaces, and let $m$ be the product measure. Let $f:X \times Y \rightarrow [0,\infty )$ be a $B \otimes C $ measurable function, $1 < p < \infty$, and define $f_x(y):=f(x,y)$.

What is the most general condition on $f$ to make sure that: $||f||_{L_p(m)}=\int{||f_x||_{L_p(\nu)}}d\mu(x)$ ?

I know that if $\int{|f_x(y)|^p}d\nu(y)$ is constant there is equality. But is it the only case?


yes, it is the only case. Denote $g(x)=(\int{|f(x,y)|^pd\nu(y))}^{1/p}$, then your condition is $||g||_1=||g||_p$ which is known to imply that $g$ is a constant function, see here.

  • $\begingroup$ Thank you. I wonder if under the same conditions as in the original question, is it always true that $||f||_{L_p(m)} \gt ∫||f_x||_{L_p(ν)}dμ(x)$? I think it can be proven using Tonelli's theorem and concave Jensen's inequality, but I'm not sure they are applicable here. $\endgroup$
    – Shlomi
    May 2 '11 at 14:56

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.