MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Let $0< \alpha< n$, $1 < p < q < \infty$ and $\frac{1}{q}=\frac{1}{p}-\frac{\alpha}{n}$. Then: $ \left \| \int_{\mathbb{R}^n} \frac{f(y)dy}{|x-y|^{n-\alpha} } \right\|_{L^q(\mathbb{R}^n)}\leq$ $C\left\| f\right\| _{L^p(\mathbb{R^n})}$.

share|cite|improve this question
Nice statement. Thank you for sharing it. What would you like to know about it? – András Bátkai Oct 19 '12 at 9:36
up vote 3 down vote accepted

This is the standard Hardy-Littlewood-Sobolev inequality(or the theorem of fractional integration).A more direct approach is write $$ \int{f(x-y)|y|^{\alpha-n}dy}=\int_{|y|<R}+\int_{|y|\ge R} $$ For the second term on the RHS,using Holder inequality,and easy to see that it's dominated by $\|f\|_{L^p}R^{-\frac{q}{n}}$. For the first term,one can use the majorizationgiven by the maximal function M,and to see that $$ |f\ast |y|^{\alpha-n}|(x)\leq C(M(f)(x)\cdot R^{\alpha}+\|f\|_{L^p}\cdot R^{-\frac{q}{n}}) $$ Choosing a proper constant R to make the two terms above be equal,and then the desired inequality hold by intergration(note that the maximal operator is bounded on $L^p$ for $1<p<\infty$).

share|cite|improve this answer
Thank you, that is what I was looking for! – Felice Oct 20 '12 at 9:44
Can you prove it without using maximal function? – timur Oct 20 '12 at 13:29
There is a proof of this inequality by the layer cake representation of functions and the Hölder inequality in Lieb and Loss, Analysis, 2001, section 4.3. – Jean Van Schaftingen Dec 4 '13 at 15:28
@timur There is a nice proof in the first book of Hörmander's The analysis of linear partial differential operators, section 4.5 "Basic $L^p$ estimates$. In his proof he only uses a dyadic Calderón-Zygmund decomposition. – Paul-Benjamin Feb 5 '15 at 21:27

The function $\vert x\vert^{\alpha-n}$ is radial homogeneous of degree $\alpha-n$, so its Fourier transform is radial homogeneous of degree $-(\alpha-n)-n=-\alpha$ (both locally integrable since $\alpha >0$ and $-\alpha>-n$ so both are distributions which are easily seen as temperate: Fourier transforms make sense), so your convolution operator is in fact the Fourier multiplier $\vert D_x\vert^{-\alpha}$. The question at hand is thus (with homogeneous spaces) $$ \Vert u\Vert_{W^{-\alpha,q}}\lesssim \Vert u\Vert_{W^{0,p}},\quad \text{i.e. }W^{0,p}\subset W^{-\alpha,q}, $$ which is a particular case of Sobolev injection since $$0>-\alpha,\quad p < q,\quad \frac{1}{p}-\frac{1}{q}=\frac{\alpha}{n}. $$

share|cite|improve this answer
Thank you! But I was looking for a more "cheep" proof. Anyway can you give me some references to Fourier multiplier? – Felice Oct 19 '12 at 20:55
@Felice I think that Harmonic Analysis (Real-variable methods, orthogonality, and oscillatory integrals) by Elias Stein is a great reference, it starts from scratch the entire theory, is beautifully written, and contains a great amount of informations. You can also look Singular Integrals and differentiability of functions by Elias Stein, to see a proof of the "easy" cases of the theorem stated by Bazin. – Paul-Benjamin Feb 5 '15 at 21:32
@Bazin. The easiest way I know of to get these inequalities uses the Hardy-Littlewood inequality, so if I get your comment correctly, Sobolev embeddings are more straightforward that Hardy-Littlewood, aren't they? – Paul-Benjamin Feb 5 '15 at 21:38

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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