Aspects of free analysis (Q1043877)

The author investigates the analysis around the free difference quotient derivation, which is the natural derivation for variables with the highest degree of non-commutativity. An analogue of the Fourier transformation is then bialgebra duality for the bialgebra with derivation-comultiplication to which the free difference quotient gives rise and which involves fully matricial analytic functions. This article under review is based on the 2nd Takagi Lectures that the author delivered at the University of Tokyo in May 2007. It surveys aspects of analysis at the highest degree of non-commutativity [\textit{D.-V.\thinspace Voiculescu}, Int.\ Math.\ Res.\ Not.\ 2000, No.\,2, 79--106 (2000; Zbl 0952.46038), Int.\ Math.\ Res.\ Not.\ 2004, No.\,16, 793--822 (2004; Zbl 1084.46053) and (2009; arXiv/0806.0361v3)]. This work originates from free probability theory, especially the study of free entropy and of its connections to random multi-matrix systems and to von Neumann algebras. A basic fact underlying commutative analysis is that the exponential functions are eigenfunctions of the partial derivatives. As is well-known, this is related to the Fourier transformation, or, more generally, to duality of abelian groups or, even more generally, to Hopf algebra duality. There is a natural question: Does there exist a replacement for this in the established background? Underlying the author's considerations is the fact that the free difference quotient derivation, appearing in free probability theory, is a co-associative co-multiplication. The co-representations of this co-algebra with derivation co-multiplication are then generalized matricial resolvents of the variable. The duality transformation is a common generalization of the Cauchy-Stieltjes transformation. With this realization about the role of generalized resolvents, the theory begins to take the shape of a non-commutative generalization of the spectral theory of one operator, in which the field of scalars \(\mathbb{C}\) is replaced by an algebra \(B\) of operators, which is not assumed to commute with the operator. All this indicates that the bialgebras with derivation-comultiplication (infinitesimal bialgebras, which at a time were called generalized difference quotient rings) are the underlying structure for the spectral study of operators via resolvents and for its non-commutative generalization. The bialgebra duality and spectral theory of resolvents naturally fit with a non-commutative generalization of the holomorphic functions of one complex variable and of the difference quotient derivation on such functions. The generalization involves matricial hierarchies of functions, which will be called fully matricial analytic functions. The last part of this paper outlines the role of these bialgebra considerations in the free probability analogue of the Fisher information and of the analytic subordination results related to Markovianity in free probability. The author points out further connections to free probability and random matrix theory. In particular, in the case of \(B\)-valued non-commutative random variables it is natural to use a fully matricial version of the R-transform. Also some comments about his current work in progress aiming at quantum relativistic uses of the non-commutative extension of spectral theory are included.