Subalgebras of \(C^*\)-algebras. III: Multivariable operator theory (Q1282108)

The paper is an extensive treatment of multivariable operator theory associated with the unit ball \(B_d\) in \(d\)-dimensional complex space \({\mathbb C}^d\) and its applications. The starting point is the definition of the Hilbert space \(H^2_d\) as a completion of the set \(P\) of all holomorphic polynomials on \(B_d\) with respect to a suitable Hilbertian norm. \(H_d^2\) is a natural generalization of the Hardy space \(H^2\). The basic theory of the spaces \(H^2_d\) and their multiplier algebras is worked out. It turns out that, unlike ``standard'' Hardy spaces, there is a function in \(H^\infty(B_d)\) which does not belong to \(H_d^2\) and that the space \(H_d^2\) is not induced by any measure. A special \(d\)-tuple \(S_d=(S_1,S_2,\ldots, S_d)\) of operators acting on \(H_d^2\), called \(d\)-shift, is introduced as a generalization of the unilateral shift in the one dimensional case. The \(d\)-shift is always a \(d\)-contraction and constitutes a basic example of a \(d\)-contraction for which the most natural generalization of the von Neumann inequality to more dimensions fails. It is also shown that the \(H_d^2\)-norm is the largest Hilbertian norm on \(P\) which is suitable for operator theory. After developing the theory of Hardy spaces in Part I the connection between \(d\)-contraction and \(A\)-morphisms is established in part II. At first the structure of the Toeplitz algebra associated with the \(d\)-shift is explored. It is proved that every \(d\)-contraction on a Hilbert space \(H\) is uniquely given by a \(A\)-morphism of this Toeplitz algebra into \(B(H)\), where \(A\) is the subalgebra generated by all polynomials in the \(d\)-shift \((S_1,S_2,\ldots, S_d)\). Further, on arbitrary operator space \(S\) a sequence of numerical invariants \(E_n(S)\) is defined. The operator space generated by the \(d\)-shift, \(d\geq 2\), is then characterized among all \(d\)-dimensional, commutative, irreducible operator spaces consisting of essentially normal operators by the property of having maximal possible sequence \(E_n(S)\). Interestingly enough, there is more uniqueness in dimension \(d\geq 2\) than in a one-dimensional case. Many applications of the theory developed to function theory and multivariable operator theory is given. These are a version of the von Neumann inequality for arbitrary \(d\)-contractions, a model theory of \(d\)-contractions based on the description of the \(d\)-contraction by the \(d\)-shift and the spherical operators, etc. The paper contains interesting deep results combining methods of \(C^\ast\)-algebras, operator theory, and function theory and it is transparently written. [For part II see ibid. 128, 271-308 (1972; Zbl 0245.46098)].