[Paper Review] The $S$-transform in arbitrary dimensions

R. Friedrich, John McKay|arXiv (Cornell University)|Aug 3, 2013

Random Matrices and Applications4 references3 citations

TL;DR

This paper resolves the long-standing problem of defining the S-transform in free probability theory for arbitrary dimensions by generalizing the framework to commutative unital rings and embedding it in algebraic geometry. The key contribution is the construction of a faithful, group-like S-transform as a minimal representation into Borel subgroups, which satisfies a multiplicative property under free independence, thereby solving the additive-to-multiplicative transition problem in higher dimensions.

ABSTRACT

In this note we report the solution of the problem of defining the $S$-transform in Free Probability Theory in arbitrary dimensions. This is achieved by generalising the theory and embedding it into an algebraic-geometric framework. Finally, we classify the groups arising as distributions of s-tuples of non-commutative random variables.

Motivation & Objective

• To resolve the unresolved problem of defining the S-transform in free probability theory beyond one dimension.
• To establish a consistent algebraic-geometric framework for higher-dimensional free probability by generalizing to arbitrary commutative unital rings.
• To classify the groups arising from s-tuples of non-commutative random variables under the boxed convolution.
• To construct a faithful representation of the S-transform that preserves the multiplicative structure under free independence.
• To close the gap between additive and multiplicative free probability by providing a coherent transition mechanism in arbitrary dimensions.

Proposed method

• Generalizing free probability theory from the complex numbers to arbitrary commutative unital rings, leveraging the combinatorial nature of non-crossing partitions.
• Using Witt vector theory to establish isomorphisms between the ring of probability measures and power series rings under Hadamard multiplication.
• Defining the S-transform as a minimal faithful representation into the projective limit of Borel subgroups of general linear groups.
• Constructing the S-transform via a recursive antipode formula for the boxed convolution algebra, using non-crossing partitions and Kreweras complements.
• Employing the R-transform and moment-cumulant duality to relate cumulants to moments in the non-commutative setting.
• Representing the S-transform as a morphism into the pro-algebraic group of upper-triangular matrices, ensuring faithfulness and compatibility with free independence.

Experimental results

Research questions

• RQ1 How can the S-transform be consistently defined in free probability theory for arbitrary dimensions beyond the one-dimensional case?
• RQ2 What algebraic structure underlies the set of s-tuples of non-commutative random variables under the boxed convolution?
• RQ3 Is there a faithful representation of the S-transform that preserves the multiplicative structure under free independence in higher dimensions?
• RQ4 Can the transition from additive to multiplicative free convolution be systematically described in arbitrary dimensions?
• RQ5 What is the group-theoretic classification of distributions arising from s-tuples of free non-commutative random variables?

Key findings

• The S-transform is constructed as a minimal faithful representation into the projective limit of Borel subgroups, establishing a concrete realization in higher dimensions.
• The S-transform satisfies the multiplicative property: S_k(a⋆b) = S_k(a) · S_k(b) for combinatorially free s-tuples, generalizing the one-dimensional case.
• The group (G^s_+, ⋆) is faithfully representable as a closed subgroup of unipotent groups U_N(k), and (G^s, ⋆) as a closed subgroup of Borel groups B_N(k).
• The antipode of the boxed convolution is given by a recursive formula involving non-crossing partitions and their Kreweras complements.
• The ring of probability measures with mean 1 is isomorphic to the ring of power series under Hadamard multiplication, via the LOG and EXP maps.
• The S-transform provides a solution to the open problem of transitioning from additive to multiplicative free convolution in arbitrary dimensions.