The fractional Young convolution inequality for regular hypergraphs

Let M2M\geq 2 and write [M]={1,2,,M}[M]=\{1,2,\ldots,M\}. Let G\mathcal{G} be a dd-regular hypergraph on [M][M], let rr and psp_{\mathbf{s}} for sG\mathbf{s}\in\mathcal{G} be real numbers in (1,)(1,\infty) satisfying

sG1ps=Gdr,\sum_{\mathbf{s}\in\mathcal{G}}\frac{1}{p_{\mathbf{s}}}=|\mathcal{G}|-\frac{d}{r'},

where rr' is the Hölder conjugate of rr. Let fjf_j, j[M]j\in[M], be probability density functions on Rn\mathbb{R}^n, and let CpC_p denote the sharp Hausdorff–Young constant. The fractional Young convolution conjecture. One should have

j[M]fjr1CrnsG[Cpsnjsfjps]1/d.\left\|\star_{j\in[M]}f_j\right\|_r\leq \frac{1}{C_r^n}\prod_{\mathbf{s}\in\mathcal{G}}\left[C_{p_{\mathbf{s}}}^n\left\|\star_{j\in\mathbf{s}}f_j\right\|_{p_{\mathbf{s}}}\right]^{1/d}.

Furthermore, the inequality should be reversed when {ps:sG}{r}(0,1)\{p_{\mathbf{s}}:\mathbf{s}\in\mathcal{G}\}\cup\{r\}\subset(0,1). This conjecture refines sharp Young's inequality for convolutions of more than two functions and connects convolution inequalities with hypergraph structure, information theory, and probability. The paper proves it when r2r\geq2 and every ps[1,2]p_{\mathbf{s}}\in[1,2], but the full range of parameters remains open.

Sources & referencesView supporting material

Primary source

Sergey Bobkov, Mokshay Madiman and Liyao Wang, “Fractional generalizations of Young and Brunn-Minkowski inequalities”, arXiv:1006.2884 (2011).

Progress summary

Never refreshed

Nothing recorded yet. Refresh searches the literature and the public web for attempts on this problem, and writes the first summary here.

Solutions 0

No solutions have been posted yet.