How Free-Free-Boolean Independence Arises in Bi-Free Probability

Daniel Pepper · 2025

Read How Free-Free-Boolean Independence Arises in Bi-Free Probability by Daniel Pepper (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-18274.txt.

Introduction

Daniel Pepper studies a layered notion of independence in probability in this 2025 paper. Liu introduced independence types for triples of algebras that mix free, Boolean, and other rules. Bi-free probability studies pairs of faces of variables at once. Pepper shows the free free Boolean independence for triples can be embedded inside the bi-free setting.

Catalog record

CREATORDaniel Pepper
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE84 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-18274

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

How Free-Free-Boolean Independence Arises in Bi-Free Probability Thanks: This research was completed with the support of NSERC (Canada) grant RGPIN-2024-06269. Daniel Pepper Abstract . This work concerns notions of multi-algebra independence introduced by Liu and how they can be studied in the context of bi-free probability. In particular, we show how the free-free-Boolean independence for triples of algebras can be embedded into and therefore studied from a lens of bi-free probability. It is also shown how its cumulants can be constructed from the bi-free cumulants. 2020 Mathematics Subject Classification 46L53, 46L54 1. Introduction The advent of free probability in the 1980’s, owing its existence to Voiculescu, brought with it a fertile ground for many rich inquiries into the depths of non-commutative probability. One such question that naturally arose was ‘ What other kinds of non-commutative probability are there and how are they related to each other? ’ It is an extended branch of this question that this work concerns. A hallmark of probability is in its notion of independence. In non-commutative probability theory, independence shows how joint moments in independent algebras is built from the individual algebras. Speicher showed in [ 6 ] that there are only five types of this independence. Classical independence, free independence, and Boolean independence are the three symmetric types, while there are two antisymmetric independences called monotone and anti-monotone independence. Each of these can be seen via an action of operators on a corresponding product space. For example, the classical independence arises from an action on a tensor product space, while the free independence arises from an action on a free product space. Voiculescu originally considered only one-sided actions on the free product in his development of free probability, but later he noticed that he could simultaneously consider both left and right sided actions, giving birth to bi-free independence for pairs of algebras in [ 7 ] . He noticed that both classical independence and free independence could be studied through bi-free probability, and conjectured that Boolean independence could also be modeled therein. This conjecture was answered affirmatively by Skoufranis in [ 5 ] . Skoufranis also showed that the monotone independence for a pair of algebras could be modeled in a similar fashion. The success of bi-free probability opened up questions of what other types of independences exist for tuples of algebras. In [ 2 ] , Liu defined several such notions, with emphasis on a free-Boolean independence for pairs of algebras, and a free-free-Boolean independence for triples of algebras in [ 3 ] . Voiculescu’s question then extends to this setting. Can these multi-algebra independences be studied through bi-free probability as well? This work answers this question in the affirmative for the free-free-Boolean (and consequently for the free-Boolean) independence with amalgamation over an algebra B . We extend the ideas of Skoufranis in [ 5 ] to see how. Apart from the introduction, this paper is split into five sections. Section 2 introduces the basic structures of non-commutative probability with amalgamation. Abstractly, these take the form of B - B -non-commutative probability spaces which act on B - B -bimodules with specified B -projections. Free product modules are recalled to make sense out of bi-free independence. We define the Boolean projections on the free product, which form a key part of the definition of Liu’s free-free-Boolean independence. Section 3 discusses a very useful calculus arising from the incidence algebra on the so-called bi-non-crossing partitions. The LR diagram calculus from [ 1 ] is called upon to handle vector components of actions, and combined with our Boolean projections. These combinatorial tools are a staple of many non-commutative probability techniques. Section 4 serves as a very brief introduction to free-free-Boolean independence with amalgamation. Originally defined in the non-amalgamated setting in [ 3 ] , we extend the definition to the amalgamated case and mention a couple of quick facts. In section 5 we tackle the main result of this paper, which shows that free-free-Boolean independent families of triples of algebras can be embedded into, and therefore studied through, bi-free probability. To do so, a bi-free family with a few extra moment conditions are leveraged to house free-free-Boolean independent families. An explicit representation is discussed, and we demonstrate that the bi-free cumulants can be used to define free-free-Boolean cumulants. Section 6 concludes this work with a brief comment on Liu’s other independences. 2. Preliminaries on B -valued probability and Bi-Free Independence 2.1. Non-commutative Probability Spaces and Bimodules Throughout this work, we denote by B , a unital algebra over {\mathbb{C}} . In B -valued probability theory, we concern ourselves with algebras of random variables and B -valued distributions. More formally, we can define a B -valued probability space as follows. Definition 2.1 . A B -valued probability space is a pair ({\mathcal{A}},E) where {\mathcal{A}} is an algebra over B and E\mathrel{\mathop{\mathchar 58\relax}}{\mathcal{A}}\to B is a linear map such that E(b_{1}ab_{2})=b_{1}E(a)b_{2} for all a\in{\mathcal{A}} and b_{1},b_{2}\in B . As the notation might suggest, we often call elements of {\mathcal{A}} non-commutative random variables and E an expectation onto the algebra B . We can consider elements of {\mathcal{A}} as operators, most simply, if we think of these elements as acting on {\mathcal{A}} by left (or right) multiplication. This turns {\mathcal{A}} into a B - B -bimodule, and embeds {\mathcal{A}} into the space of linear operators on {\mathcal{A}} . The expectation returns to us in the form of a projection onto B . Definition 2.2 . A B - B -bimodule with a specified B -projection is a triple ({\mathcal{X}},\accentset{\circ}{{\mathcal{X}}},p) where {\mathcal{X}} is a direct sum of B - B -bimodules {\mathcal{X}}=B\oplus\accentset{\circ}{{\mathcal{X}}} and p\mathrel{\mathop{\mathchar 58\relax}}{\mathcal{X}}\to B is the projection (with kernel \accentset{\circ}{{\mathcal{X}}} ) p(b\oplus\eta)=b. Let {\mathcal{L}}({\mathcal{X}}) denote the algebra of {\mathbb{C}} -linear maps on {\mathcal{X}} . We have an expectation , E_{{\mathcal{L}}({\mathcal{X}})}\mathrel{\mathop{\mathchar 58\relax}}{\mathcal{L}}({\mathcal{X}})\to B defined by E_{{\mathcal{L}}({\mathcal{X}})}(a)=p(a(1_{B}\oplus 0))=pa1_{B}. One can check that starting with a B - B -bimodule with specified projection ({\mathcal{X}},\accentset{\circ}{{\mathcal{X}}},p) , we get for free a B -valued probability space ({\mathcal{L}}({\mathcal{X}}),E_{{\mathcal{L}}({\mathcal{X}})}) . As alluded to above, the same is true in reverse; starting with a B -valued probability space ({\mathcal{A}},E) , we can create a B - B -bimodule with specified projection, ({\mathcal{A}},\ker(E),E) . This correspondence between B -valued probability and B - B -bimodules lies at the heart of many of our investigations that follow. Now, in order to talk about pairs of faces and bi-free probability, we need to be able to discuss left and right acting operators. These behave slightly differently when we consider the algebra B over which we are amalgamating. For example, if b\in B , let L_{b} and R_{b} denote the action on a B - B -bimodule {\mathcal{X}} by left and right multiplication by b\in B respectively, we can consider the left algebra of {\mathcal{L}}({\mathcal{X}}) to be {\mathcal{L}}_{\ell}({\mathcal{X}})\mathrel{\mathop{\mathchar 58\relax}}=\{T\in{\mathcal{L}}({\mathcal{X}})\ |\ TR_{b}=R_{b}T\} and the right algebra of L({\mathcal{X}}) to be {\mathcal{L}}_{r}({\mathcal{X}})\mathrel{\mathop{\mathchar 58\relax}}=\{T\in{\mathcal{L}}(X)\ |\ TL_{b}=L_{b}T\}. In other words, left-acting operators should commute with right multiplication by B , and right acting operators should commute with left-multiplication by B . Motivated by this, we look at the following abstraction, that of a B - B -non-commutative probability space. Definition 2.3 . A B - B -non-commutative probability space is a triple ({\mathcal{A}},E_{\mathcal{A}},\varepsilon)

Read the complete text