Let $f: \mathbb{R}^d \to \mathbb{R}^*$ be measurable. Let $f^*$ denote the Hardy-Littlewood maximal function.

1) If $f$ is nonzero on a set of positive measure prove that there exists $0 \lt A \lt \infty$ such that $f^*(x) \geq A ||x||^{-d}$ whenever $||x||$ is sufficiently large.

2) If $f \in L^1(\mathbb{R}^d)$ is compactly supported, prove that there exists $0 \lt B \lt \infty$ such that $f^*(x) \leq B ||x||^{-d}$ holds whenever $||x||$ is sufficiently large.

Both of these parts make sense intuitively: for the first one, I want to show that if $f$ is nonzero on a set of positive measure, then I can bound the "supremum" below by a nonzero function that depends on $x$. For the second part, I have that $f$ is integrable, so it is finite almost everywhere, so $f^*$ should be finite almost everywhere. Hence $f^*$ can be bounded above by some finite function which depends on $x$. I'm not sure where the compact support comes in though. Is my intuition correct? I have no idea how to prove this rigorously. How do I even start with this kind of problem?


1) Without loss of generality we may assume $A\subset B_R(0)$, where $B_R(z)$ is the ball of radius $R>0$ centered at $z$. If $\|x\|\ge2\,R$ then $A\subset B_{\|x\|+R}(x)$ and $$ f^*(x)\ge\frac{1}{|B_{\|x\|+R}(x)|}\int_A|f|=C_d(\|x\|+R)^{-d}\int_A|f|\ge\Bigl(C_d\,(2/3)^d\int_A|f|\Bigr)\,\|x\|^{-d}. $$

2) Suppose that the support of $f$ is contained in $B_R(0)$. Take $x$ with $\|x\|>2\,R$. Then $B_R(0)\cap B_r(x)=\emptyset$ if $r\le\|x\|-R$. If $r>\|x\|-R$ then $$ \frac{1}{|B_{r}(x)|}\int_{B_r(x)}|f|\le C_d\,r^{-d}\|f\|_1\le C_d\,(\|x\|-R)^{-d}\|f\|_1\le\Bigl(C_d\,2^d\,\|f\|_1\Bigr)\,\|x\|^{-d}. $$

  • $\begingroup$ What happens to your proof of 1) if the set $A$ where $f$ in nonzero has infinite measure? Wouldn't that make $R$ and $||x||$ infinite as well? Is the proof still valid if that happens? $\endgroup$ – Johanna Nov 20 '14 at 15:53
  • $\begingroup$ There exists $A=\cup_{n=1}^\infty A\cap B_n$. Since $|A|>0$, there is $n$ such that $|A\cap B_n|>0$. $\endgroup$ – Julián Aguirre Nov 20 '14 at 15:56
  • $\begingroup$ I don't understand how that helps if $|A| = \infty$? It seems to me that we need to contain all of $A$ in a ball of finite radius, so we need $|A \cap B_n| = |A|$, right? $\endgroup$ – Johanna Nov 20 '14 at 15:59
  • $\begingroup$ No. $f^*(x)\ge(f\,\chi_{A\cap B_n})^*(x)$ for all $x$. $\endgroup$ – Julián Aguirre Nov 20 '14 at 16:02

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.