\(\gamma \)-Radonifying operators and UMD-valued Littlewood-Paley-Stein functions in the Hermite setting on BMO and Hardy spaces (Q1932177)

The authors study the boundedness of the square function associated to the Hermite operator from the corresponding bounded mean oscillation (\(\mathrm{BMO}\)) and \(H^1\) defined in the Hermite setting for vector-valued functions and its interplay with \(\gamma\)-Radonifying operators and UMD spaces. Harmonic analysis associated to the harmonic oscillator \[ L=-\Delta + |x|^2 \] has been developed in the last years by many authors. In the paper, \(\mathcal L\) stands for \[ \mathcal L (f)=\sum_{k} (2|k|+n)\langle f, h_k\rangle h_k \] for \(f\) belonging to a suitable domain, where \(h_k\) are multidimensional Hermite functions. Starting with such an operator, the authors consider the heat semigroup \(W_t^{\mathcal L}\) generated by \(-\mathcal L\) in \(L^2(\mathbb R^n)\) which is defined by \[ W_t^{\mathcal L}(f)=\sum_{k} e^{-i(2|k|+n)}\langle f,h_k\rangle h_k \] and the Poisson semigroup \(P_t^{\mathcal L}\) generated by \(-\sqrt {\mathcal L}\) which is given using the classical subordination formula. \textit{K. Stempak} and \textit{J. L. Torrea} [Acta Math. Hung. 109, No. 1--2, 99--125 (2005; Zbl 1150.42326)] considered the Littlewood-Paley \(g\)-function \[ g^{\mathcal L}(f)=\left(\int_0^\infty |t\partial_t P_t^{\mathcal L}f(x)|^2 \frac{dt}{t}\right)^{1/2} \] and showed that \[ \|f\|_p\approx \|g^{\mathcal L}(f)\|_p\quad\text{for }1<p<\infty. \] The vector-valued analogues \[ \mathcal G_{\mathcal L, B}(f)(x,t)= t\partial_t P_t^{\mathcal L}f(x) \] for \(f\in L^p(\mathbb R^n, B)\) where \(B\) is a Banach space and \[ g_B^{\mathcal L}(f)=\left(\int_0^\infty \|\mathcal G_{\mathcal L, B}(f)(x,t)\|\frac{dt}{t}\right)^{1/2} \] were also analyzed in [the authors, Isr. J. Math. 182, 1--30 (2011; Zbl 1222.46011)], where it was shown, as a consequence of other results, that \[ \|f\|_p\approx \|g_B^{\mathcal L}(f)\|_p\quad\text{for some }1<p<\infty \] if and only if \(B\) is isomorphic to a Hilbert space. It has recently been shown that we can replace the Littlewood-Paley \(g\)-function by a different one using \(\gamma\)-Radonifying operators by considering, for each \(x\in \mathbb R^n\), the operator \[ g\to \int_0^\infty g(t)\mathcal G_{\mathcal L, B}(f)(x,t)\frac{dt}{t} \] as an operator from \(L^2((0,\infty),\frac{dt}{t})\) into \(B\) and defining the \(g\)-function using its \(\gamma\)-Radonifying norm \[ \|\mathcal G_{\mathcal L, B}(f)(x,\cdot)\|_{\gamma(L^2((0,\infty),\frac{dt}{t}), B)}. \] With this and using ideas from \textit{C. Kaiser} and \textit{L. Weis} [Stud. Math. 186, No. 2, 101--126 (2008; Zbl 1213.42145)], it has been recently shown by the authors [``Square functions in the Hermite setting for functions with values in UMD spaces'', preprint, \url{arXiv:1203.1480}] that, for UMD spaces \(B\), it holds that \[ \|f\|_{L^p(\mathbb R^n, B)}\approx \|\mathcal G_{\mathcal L, B}(f)\|_{L^p(\mathbb R^n, \gamma(L^2((0,\infty),\frac{dt}{t}), B)}. \] In the present paper, equivalent inequalities for the \(g\)-function associated to \(\gamma\)-Radonifying operators are analyzed for both functions in \(H^1_{\mathcal L}(\mathbb R^n, B)\) and \(\mathrm{BMO}_{\mathcal L}(\mathbb R^n, B)\) where \(H^1_\mathcal L(\mathbb R^n, B)\) consists of those functions in \(L^1(\mathbb R^n, B)\) for which the maximal function \(\sup_{t>0} \|W^{\mathcal L}_t(f)\|\in L^1(\mathbb R^n)\) and \(\mathrm{BMO}_{\mathcal L}(\mathbb R^n, B)\) is the vector-valued version of the dual of \(H^1_\mathcal L(\mathbb R^n)\), where the absolute value is replaced by a norm (see [\textit{J. Dziubański} et al., Math. Z. 249, No. 2, 329--356 (2005; Zbl 1136.35018)]).