BMO results for operators associated to Hermite expansions (Q819227): Difference between revisions

In the \(d\)-dimensional Euclidean space \(\mathbb R^d\), \(d\geq 1\), consider the system of multi-dimensional Hermite functions \[ h_\alpha(x)=h_{\alpha_1}(x_1)\cdots h_{\alpha_d}(x_d), \] where \(\alpha=(\alpha_1,\dots,\alpha_d)\), \(\alpha_i\in \mathbb Z_+\), \(x=(x_1,\dots,x_d)\), and \[ h_k(s)=(\pi^{1/2}2^kk!)^{-1/2}H_k(s)\exp(-s^2/2),\quad k\in \mathbb Z_+, \] are the one-dimensional Hermite functions, and \(H_k(s)\) denotes the \(k\)th Hermite polynomial. Let \({\mathcal P}_nf=\sum_{| \alpha| =n}\langle f,h_\alpha\rangle h_\alpha\) and \(\mathcal L=\sum_{n=0}^\infty(2n+d){\mathcal P}_n.\) Denote the heat-diffusion semigroup and the Poisson semigroup associated to \(\mathcal L\), respectively, by \(\{T_t\}_{t>0}\) and \(\{P_t\}_{t>0}\). It is known that for any \(f\in L^q(\omega)\), \(1\leq q<\infty\), \(\omega\in A_q\), or \(f\in L^\infty\), \(T_tf(x)\) and \(P_tf(x)\) are, respectively, equal to the heat-diffusion and the Poisson integral of \(f\), defined by \[ g(t,x)=\int_{\mathbb R^d}G_t(x,y)f(y)\,dy, \quad f(t,x)= \int_{\mathbb R^d}P_t(x,y)f(y)\,dy, \] where \[ G_t(x,y)=(2\pi \sinh(2t))^{-d/2} \exp\Big(-\frac1{4}(\tanh(t)| x+y| ^2+\coth(t)| x-y| ^2)\Big) \] and \[ P_t(x,y)=\frac{t}{\sqrt{4\pi}}\int_0^\infty G_s(x,y)s^{-3/2}e^{-t^2/{4s}}\,ds \] are the heat-diffusion and Poisson kernels. Define also \[ T^\ast f(x)=\sup_{t>0}| g(t,x)| ,\quad P^*f(x)=\sup_{t>0}| f(t,x)| , \] \[ \widetilde{g}(f)(x)=\left(\int_0^\infty\Big| \frac{\partial}{\partial t}g(t,x)\Big| ^2 t\,dt\right)^{1/2},\quad g(f)(x)=\left(\int_0^\infty\Big| \frac{\partial}{\partial t}f(t,x)\Big| ^2 t\,dt\right)^{1/2} \] and \[ g_{\nabla}f(x)=\left(\int_0^\infty| \nabla f(t,x)| ^2 t\,dt\right)^{1/2}, \] where \(\nabla=\nabla_H=(\delta_1^-, \cdots, \delta_d^-, \frac{\partial}{\partial t}, \delta_1^+, \cdots, \delta_d^+)\) and \(\delta_j^\pm=\partial_{x_j}\pm x_j\). The essential aim of the present paper is to investigate the action of the aforementioned operators: \(T_t\) and \(P_t\), \(t>0\) fixed, the maximal operators \(T^*\) and \(P^*\), the square functions \(\widetilde{g},\;g\) and \(g_\nabla\), on the spaces \(L^\infty\) and BMO, by using a technique of vector-valued Calderón-Zygmund operators. A characteristic feature of these operators is that they do not map BMO into BMO with a control of the BMO seminorm, which is a major difference between the situation here and the classic (Euclidean) situation.