A characterization of the Gaussian Lipschitz space and sharp estimates for the Ornstein-Uhlenbeck Poisson kernel (Q515344)

The authors establish a new representation of the Gaussian Lipschitz space \[ \text{GLip}_{\alpha} = \bigl\{f:\mathbb{R}^n\rightarrow\mathbb{R}:f \text{ measurable, bounded and }\exists A>0\;\forall t>0:\|\partial_tP_tf\|_{L^{\infty}}\leqslant A\,t^{\alpha-1}\bigr\} \] with norm \[ \|f\|_{\text{GLip}_{\alpha}} = \|f\|_{L^{\infty}} + \inf\bigl\{A>0:\forall\;t>0:\|\partial_tP_tf\|_{L^{\infty}}\leqslant A\,t^{\alpha-1}\bigr\} \] that was defined by \textit{A. E. Gatto} et al. [in: Recent advances in harmonic analysis and applications. In honor of Konstantin Oskolkov on the occasion of his 65th birthday. Based on the AMS sectional meeting, Georgia Southern University, Statesboro, GA, USA, March 11--13, 2011. Berlin: Springer. 105--130 (2013; Zbl 1291.47036)] using the Ornstein-Uhlenbeck Poisson kernel \[ P_t(x,y)=\frac{1}{2\pi^{(n+1)/2}}\int_0^{\infty}\frac{t}{s^{3/2}}e^{\frac{-t^2}{4s}}\frac{e^{-\frac{|y-e^{-s}x|^2}{1-e^{-2s}}}}{(1-e^{-2s})^{n/2}}\,ds. \] The authors show in the present paper that a measurable and bounded function \(f:\mathbb{R}^n\rightarrow\mathbb{R}\) belongs to \(\text{GLip}_{\alpha}\) if and only if there exists \(K>0\) such that \[ |f(x)-f(y)|\leqslant K \text{min}\bigl\{|x-y|^{\alpha},\Bigl(\frac{|x-y|}{1+|x|+|y|}\Bigr)^{\alpha/2}+((|x|+|y|)\sin\theta)^{\alpha}\bigr\}\tag{*} \] holds for all \(x\), \(y\in\mathbb{R}^n\) after correction on a null set. Here, \(\theta\) denotes the angle between \(x\) and \(y\) if \(x\not=0\not=y\) and is zero otherwise. Moreover, they show that on \(\text{GLip}_{\alpha}\), the norm \(\|\cdot\|_{\text{GLip}_{\alpha}}\) is equivalent to the norm given by \[ \|f\|=|f(0)|+\text{inf}\bigl\{K>0:(\star) \text{ holds for all }x,\,y\in\mathbb{R}^n \text{ after correction on a null set}\bigr\}. \] The special case \(n=1\), i.e., \(\sin\theta\equiv0\), illustrates that the estimate in \((\star)\) is a combination of two Lipschitz conditions with exponent \(\alpha\) for distances shorter than \(1/(1+|x|+|y|)\) and exponent \(\alpha/2\) for larger distances between \(x\) and \(y\). The proof relies on sophisticated estimates of the Ornstein-Uhlenbeck Poisson kernel and its derivatives.