2
$\begingroup$

$\newcommand{\RR}{\mathbb R}\newcommand{\diff}{\, \mathrm d}$ We fix $T \in (0, \infty)$ and $p, q \in [1, \infty)$. Let $\mathbb T$ be the interval $[0, T]$.

  • Let $E$ be the space of all real-valued Lebesgue measurable functions on $\mathbb T \times \RR^d$ with the finite norm $$ \| g \|_E := \sup_{x \in \RR^d} \bigg ( \int_{\mathbb T} \bigg ( \int_{B(x, 1)} | g(s, y) |^p \diff y \bigg )^{\frac{q}{p}} \diff s \bigg )^{\frac{1}{q}}. $$

  • Let $F$ be the space of all real-valued Lebesgue measurable functions on $\mathbb T \times \RR^d$ with the finite norm $$ \| g \|_F := \bigg ( \int_{\mathbb T} \bigg ( \sup_{x \in \RR^d} \int_{B(x, 1)} | g(s, y) |^p \diff y \bigg )^{\frac{q}{p}} \diff s \bigg )^{\frac{1}{q}}. $$

Above, $B(x, 1)$ is the open ball centered at $x$ with radius $1$. Clearly, $\| g \|_E \le \| g \|_F$ and thus $F \subset E$.

Is there a constant $c >0$ such that $\| g \|_F \le c \| g \|_E$ for all $g \in E$?

Thank you so much for your elaboration!

$\endgroup$
0

1 Answer 1

4
$\begingroup$

The opposite inequality cannot be true. If that were true, then consider a positive function $g$ with the property such that for all $s\in \mathbb{T}$ it holds that $g(s,x) \leq C g(s,y)$ whenever $|x-y|<1$ then it would hold that \begin{equation} \int_\mathbb{T} \sup_{x\in\mathbb{R}^d} g(x,s)ds \leq C_1 \sup_{x\in \mathbb{R}^d} \int_{\mathbb{T}}g(s,x)ds, \end{equation} for some $C_1>0$. This cannot be true consider for example $g(s,x)= \frac{\psi'(s)}{((|x|-\psi(s))^2+1)^\frac{d+1}{2}}$, where $\psi:(0,T)\to\mathbb{R}$ is a strictly increasing smooth function such that $\psi(0^+)=-\infty, \psi(T^-)=+\infty)$. Then the left hand side becomes \begin{equation*} \text{L.H.S.}=\int_0^T\psi'(s)ds = \int_\mathbb{R}dt=+\infty. \end{equation*} While the right hand side is \begin{equation*} \text{R.H.S.}= \sup_{x\in \mathbb{R^d}} \int_\mathbb{R} \frac{dt}{((|x|-t)^2+1)^{\frac{d+1}{2}}} = \int_\mathbb{R} \frac{dt}{(t^2+1)^{\frac{d+1}{2}}}<+\infty. \end{equation*}

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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