5

At the $0$ order of derivatives of Sobolev spaces, we find Besov spaces $\dot{B}^0_{p,q}$, Triebel Lizorkin spaces $\dot{F}^0_{p,q}$ and Lorentz spaces $L^{p,q}$, with in particular if $p≥ 2$ $$ \begin{align*} \dot{B}^0_{p,1} ⊂ \dot{B}^0_{p,2} ⊂ L^p &= \dot{F}^0_{p,2} \subset \dot{F}^0_{p,p} =\dot{B}^0_{p,p} ⊂ \dot{B}^0_{p,\infty} \\ L^{p,1} \subset L^{p,2} ⊂ L^{p} &= L^{p,p} \subset L^{p,\infty}. \end{align*} $$ So, the ordering is well understood when remaining either in the $B,F$ setting, either in the Lorentz setting. so my question is Is there any embedding from one setting into the other one?


The way of building these spaces is different (in one case one cuts in frequency and in the other one cuts in height). However, Sobolev embeddings tells us that we can trade a bit of local regularity for a bit of local integrability. Moreover the function $|x|^{-a}$ is in both $\dot{B}^{0}_{d/a,\infty}$ and $L^{d/a,\infty}$, but not in $L^{d/a}$.

Rushabh Mehta
  • 13,845
LL 3.14
  • 13,938

1 Answers1

3

(I came across this problem by accident, hopefully it's not outdated in the sense of, say, looking for some research problems.)

This turns to be a difficult problem. There are Besov-Lorentz spaces $B_{(p,r),q}^s$ and Triebel-Lorentz space $F_{(p,r),q}^s$ where the $L^p$ integral is replaced by the $L^{p,r}$ norm. In this setting we have $L^{p,r}=F_{(p,r),2}^0$ when $1<p<\infty$ and $0<r\le\infty$.

According to Seeger-Trebels (2019), when $1<p<\infty$, $B_{p,q}^0\hookrightarrow L^{p,r}$ holds iff one of the following:

  • $2\neq p\le r$ and $q\le\min(p,2)$;
  • $2=p\le r$ and $q\le \min(p,r)$.

And $L^{p,r}\hookrightarrow B_{p,q}^0$ holds iff one of the following:

  • $2\neq p\ge r$ and $q\ge\max(p,2)$;
  • $2=p\ge r$ and $q\ge \max(p,r)$.

$F_{(p,r_0),q_0}^0\hookrightarrow F_{(p,r_1),q_1}^0$ holds iff $r_0\le r_1$ and $q_0\le q_1$.

Liding Yao
  • 3,037
  • Oh, thank you about this answer, I didn't know about this paper and it seems indeed to answer the question. – LL 3.14 Mar 18 '24 at 16:09
  • @LL3.14 I believe there are not many research involving Lorentz variant on Triebel-Lizorkin spaces. Nowaday more likely people just care about Besov and Triebel-Lizorkin spaces on general Banach function spaces or ball Banach function spaces (if you are interested). – Liding Yao Mar 18 '24 at 20:36