Noncommutative Berezin transforms and multivariable operator model theory (Q2472870): Difference between revisions

Let \(\mathbb{F}_n^+\) be the unital free semigroup on \(n\) generators \(g_1,\dots,g_n\) and the identity \(g_0\). To each \(n\)-tuple \(X=(X_1,\dots,X_n)\) of bounded linear operators on the Hilbert space \(\mathcal H\) and each polynomial \(q=q(Z_1,\dots,Z_n)=\sum_\alpha c_\alpha Z_\alpha\) in noncommutative indeterminates \(Z_1,\dots,Z_n,\) where \(Z_\alpha:=Z_{i_1}\dots Z_{i_k}\) if \(\alpha=g_{i_1}\dots g_{i_k}\in\mathbb{F}_n^+\), \(i_1,\dots,i_k\in\{1,\dots,n\}\) and \(Z_{g_0}=I\), one can associate the operator \(q(X,X^*):=\sum_\alpha c_\alpha X_\alpha X^*_\alpha\). The author initiates a deep and complete study of the noncommutative domain \[ \mathbf{D}_p^m(\mathcal{H}):=\{X\in B(\mathcal{H})^n\mid (1-p)^k(X,X^*)\geq 0\text{ for }1\leq k\leq m\}, \] where \(m\) is a positive integer and \(p=p(Z_1,\dots,Z_n)=\sum_\alpha a_\alpha Z_\alpha\) is a positive regular polynomial (i.e., \(a_\alpha\geq 0\) for \(\alpha\in\mathbb{F}_n^+,\;a_{g_0}=0\) and \(a_{g_i}>0,\;i=1,\dots,n\)). The role of the universal model for the elements of \(\mathbf{D}_p^m(\mathcal{H})\) is played by a tuple \((W_1,\dots,W_n)\) of weighted shifts acting on the full Fock space \(F^2(H_n)\) with \(n\) generators. The main tool for the later developments in the paper is a noncommutative Berezin transform \(\mathbf{B}_T:B(F^2(H_n))\to B(\mathcal{H})\) which is introduced for each tuple \(T\in\mathbf{D}_p^m(\mathcal{H})\) having joint spectral radius \(r_p(T)<1\). It has an extension \(\tilde{\mathbf{B}}_T\) to any \(T\in\mathbf{D}_p^m(\mathcal{H})\). When \(T\) is a pure tuple of operators (i.e., \(p^k(T,T^*)\to 0\) strongly as \(k\to\infty\)), the restriction of \(\mathbf{B}_T\) to the operator system \(\overline{\text{span}}\,\{W_\alpha W_\beta^*: \alpha,\beta\in\mathbb{F}_n^+\}\) is proved to be a unital completely contractive linear map such that \[ \tilde{\mathbf{B}}_T[W_\alpha W_\beta^*]=T_\alpha T_\beta^*,\quad \alpha,\beta\in\mathbb{F}_n^+. \] A similar conclusion is obtained for tuples \(T=(T_1,\dots,T_n)\) with the radial property, i.e., \(rT=(rT_1,\dots,rT_n)\in\mathbf{D}_p^m(\mathcal{H})\) for any \(r\in(\delta,1]\) and some \(\delta\in (0,1)\). The Hardy algebra \(F_n^\infty(\mathbf{D}_p^m)\) associated with the noncommutative domain \(\mathbf{D}_p^m\) is introduced and an \(F_n^\infty(\mathbf{D}_p^m)\)-functional calculus for completely non-coisometric \(n\)-tuples of operators in \(\mathbf{D}_p^m\) is developed. The eigenvectors for \(W_1^*,\dots,W_n^*\) are precisely determined. This result is used to describe the \(w^*\)-continuous linear functionals on \(F_n^\infty(\mathbf{D}_p^m)\) and to compute the right joint spectrum of \((W_1,\dots,W_n)\). The symmetric weighted Fock space \(F_s^2(\mathbf{D}_p^m)\) is then defined and identified with a reproducing kernel Hilbert space \(H^2(\mathbf{D}_{p,\circ}^1(\mathbb{C}))\). The algebra \(H^\infty(\mathbf{D}_{p,\circ}^1(\mathbb{C}))\) of all multipliers of \(H^2(\mathbf{D}_{p,\circ}^1(\mathbb{C}))\) is shown to be reflexive. Noncommutative varieties \(\mathcal{V}_{p,\mathcal{Q}}^m(\mathcal{H})\subset\mathbf{D}_p^m(\mathcal{H})\) determined by sets \(\mathcal{Q}\) of noncommutative polynomials are considered. To each such a variety, the author associates a universal model \((B_1,\dots,B_n)\in \mathcal{V}_{p,\mathcal{Q}}^m(\mathcal{N}_{\mathcal Q})\) which is the compression of \((W_1,\dots,W_n)\) to an appropriate subspace \(\mathcal{N}_{\mathcal Q}\) of the full Fock space. A constrained version of the noncommutative Berezin transform is introduced and used to find analogous of some of the results above, for subvarieties. Certain dilation and model theorems for the elements of the noncommutative variety \(\mathcal{V}_{p,\mathcal{Q}}^m(\mathcal{H})\) are finally presented. By this general approach, it is possible to recover several results concerning model theory for commuting \(n\)-tuples of operators (in this case, \(\mathcal Q\) consists of the polynomials \(W_iW_j-W_jW_i,\;i,j\in1,\dots,n\)) such as the ones obtained by \textit{V.\,Müller} [J.~Oper.\ Theory 20, No.\,1, 3--20 (1988; Zbl 0695.47008)], \textit{V.\,Müller} and \textit{F.--H.\thinspace Vasilescu} [Proc.\ Am.\ Math.\ Soc.\ 117, No.\,4, 979--989 (1993; Zbl 0777.47009)], \textit{F.--H.\thinspace Vasilescu} [J.~Funct.\ Anal.\ 110, No.\,1, 47--72 (1992; Zbl 0776.32011)], \textit{R.\,E.\thinspace Curto} and \textit{F.--H.\thinspace Vasilescu} [Acta Sci.\ Math.\ 57, No.\,1--4, 65--78 (1993; Zbl 0801.47002)] or \textit{S.\,Pott} [J.~Oper.\ Theory 41, No.\,2, 365--389 (1999; Zbl 0995.47003)].