English

Reversed Hardy-Littewood-Sobolev inequality

Analysis of PDEs 2014-07-11 v3

Abstract

The classical sharp Hardy-Littlewood-Sobolev inequality states that, for 1<p,t<1<p, t<\infty and 0<λ=nα<n0<\lambda=n-\alpha <n with 1/p+1/t+λ/n=2 1/p +1 /t+ \lambda /n=2, there is a best constant N(n,λ,p)>0N(n,\lambda,p)>0, such that RnRnf(x)xyλg(y)dxdyN(n,λ,p)fLp(Rn)gLt(Rn) |\int_{\mathbb{R}^n} \int_{\mathbb{R}^n} f(x)|x-y|^{-\lambda} g(y) dx dy|\le N(n,\lambda,p)||f||_{L^p(\mathbb{R}^n)}||g||_{L^t(\mathbb{R}^n)} holds for all fLp(Rn),gLt(Rn).f\in L^p(\mathbb{R}^n), g\in L^t(\mathbb{R}^n). The sharp form is due to Lieb, who proved the existence of the extremal functions to the inequality with sharp constant, and computed the best constant in the case of p=tp=t (or one of them is 2). Except that the case for p((n1)/n,n/α)p\in ((n-1)/n, n/\alpha) (thus α\alpha may be greater than nn) was considered by Stein and Weiss in 1960, there is no other result for α>n\alpha>n. In this paper, we prove that the reversed Hardy-Littlewood-Sobolev inequality for 0<p,t<10<p, t<1, λ<0\lambda<0 holds for all nonnegative fLp(Rn),gLt(Rn).f\in L^p(\mathbb{R}^n), g\in L^t(\mathbb{R}^n). For p=tp=t, the existence of extremal functions is proved, all extremal functions are classified via the method of moving sphere, and the best constant is computed.

Keywords

Cite

@article{arxiv.1309.1974,
  title  = {Reversed Hardy-Littewood-Sobolev inequality},
  author = {Jingbo Dou and Meijun Zhu},
  journal= {arXiv preprint arXiv:1309.1974},
  year   = {2014}
}

Comments

Comment on recent development on the application of the reversed HLS inequality is added (in introduction section), new references ([7] and [22]) are added. Some typoes are corrected