发表机构
Central South University(中南大学)
机构由 AI 辅助整理,请以论文原文为准。AI 中文总结
本文在超立方体上证明了尖锐的分数阶 Riesz 估计,解决了开放问题,并推广到高阶情形,简化了相关应用证明。
AI 中文摘要
设 $\Omega_{n}=\{-1,1\}^n$ 为配备归一化均匀测度的 $n$ 维超立方体,设 $\nabla$ 为 Walsh 梯度,$\Delta$ 为 Walsh 拉普拉斯算子。对每个 $1<p\leq 2$,我们证明如下估计 \\[ \\|\nabla f\\|_{L_p(\Omega_n;\ell_2^n)} \leq c_{\rm abs}(p-1)^{-2}\\|\Delta^{1/p}f\\|_{L_p(\Omega_n)}. \\] 指数 $\frac1p$ 是最优的,因此这解决了 Efraim 和 Lust-Piquard \cite{E-LP2008} 提出的尖锐分数阶 Riesz 估计的开放问题,该问题随后被 Ivanisvili 和 Volberg \cite{I-V2022} 强调。我们还建立了高阶对应版本。作为我们结果的应用,我们获得了 $\nabla e^{-t\Delta}$ 的最优短时估计以及 $d$-有界次数函数的 Bernstein-Markov 型不等式的更简单证明。
英文摘要
Let $Ω_{n}=\{-1,1\}^n$ be the $n$-dimensional hypercube equipped with the normalized uniform measure, let $\nabla$ be the Walsh gradient and let $Δ$ be the Walsh Laplacian. For every $1<p\leq 2$ we prove the following estimate \[ \|\nabla f\|_{L_p(Ω_n;\ell_2^n)} \leq c_{\rm abs}(p-1)^{-2}\|Δ^{1/p}f\|_{L_p(Ω_n)}. \] The exponent $\frac1p$ is optimal, thus this settles the open problem on the sharp fractional Riesz estimate by Efraim and Lust-Piquard \cite{E-LP2008} which was subsequently highlighted by Ivanisvili and Volberg \cite{I-V2022}. We also establish the higher-order counterpart. As applications of our results, we obtain simpler proofs of the optimal short-time estimate for $\nabla e^{-tΔ}$, and the Bernstein-Markov type inequality for $d$-bounded degree functions.