Polynomials in c c -free random variables with applications to free denoising

Adrián Celestino, Franz Lehner, Kamil Szpojankowski · 2026

Read Polynomials in c c -free random variables with applications to free denoising by Adrián Celestino, Franz Lehner, Kamil Szpojankowski (2026) 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-2603-21372.txt.

Introduction

Celestino and co-authors study polynomials in c-free random variables. C-freeness is a form of independence for two-state noncommutative probability spaces, due to Bozejko, Leinert, and Speicher. They set up recursions between the joint Boolean cumulants of such variables. These give a way to compute the laws of self-adjoint polynomials in them. As an application, they find certain conditional expectations for free variables. They also define projections that play the same role in the c-free setting.

Catalog record

CREATORAdrián Celestino, Franz Lehner, Kamil Szpojankowski
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE192243
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.21372

Browse the full library · About the Tower of Babel

Full text

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

Polynomials in c c -free random variables with applications to free denoising Thanks: Supported by the Austrian Federal Ministry of Education, Science and Research and the Polish Ministry of Science and Higher Education, grants N os {}^{\textrm{os}} PL 08/2016 and PL 06/2018. AC: Funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany’s Excellence Strategy EXC 2044/2 –390685587, Mathematics Münster: Dynamics–Geometry–Structure. FL: This research was funded in part by the Austrian Science Fund (FWF) grant I 6232-N BOOMER (WEAVE). KSz: This research was funded in part by National Science Centre, Poland WEAVE-UNISONO grant BOOMER 2022/04/Y/ST1/00008. Adrián Celestino, Franz Lehner, Kamil Szpojankowski Abstract . We study distributions of polynomials in conditionally free ( c c -free) random variables, a notion of independence for two-state noncommutative probability spaces ( 𝒜 , φ , ψ ) (\mathcal{A},\varphi,\psi) introduced by Bożejko, Leinert and Speicher. To this end we establish recursive relations between the joint Boolean cumulants of c c -free random variables, analogous to previously found recursions for Boolean cumulants of free random variables. The algebraic reformulation of these recursions on the free associative algebra provides an effective formal machinery for the computation of the moment generating functions and thus the distributions of arbitrary self-adjoint polynomials in c c -free random variables. As an application of a recent observation, our approach can be used to determine conditional expectations of the form E ⁡ [ a | P ⁡ ( a , b ) ] E\left[a|P(a,b)\right] , where P P is a self-adjoint polynomial in free (in the sense of Voiculescu) random variables a , b a,b . We illustrate this with an example where P ⁡ ( a , b ) = i ⁡ [ a , b ] P(a,b)=i[a,b] . Finally we define orthogonal projections that formally play the role of conditional expectations in the framework of c c -freeness and share some properties with the conditional expectations of free variables. In particular they can be used to re-derive by purely algebraic methods the formula of Popa and Wang for the Σ \Sigma -transform for the c c -free multiplicative convolution. Dedicated to professor Marek Bożejko on the occasion of his 80th birthday. 1. Introduction The notion of c -freeness (or conditional freeness) was introduced by Bożejko, Leinert, and Speicher in [ 10 , 9 ] . It is a noncommutative independence in the framework of a two-state noncommutative probability space (\mathcal{M},\varphi,\psi) . This concept generalizes freeness in the following sense: random variables a,b\in\mathcal{M} are c -free with respect to the pair (\varphi,\psi) , if they are free in the sense of Voiculescu with respect to the second state \psi and satisfy a factorization condition with respect to the first state \varphi . For precise definitions of both types of independence see Section 2 below. The distribution of a self-adjoint element a in a two-state probability space is described by a pair of measures \mu_{a}^{\varphi} and \mu_{a}^{\psi} , corresponding to the distributions with respect to \varphi and \psi , respectively. In this paper, we present a method for determining the distribution of polynomials in c -free variables. More precisely, for c -free variables a,b and a noncommutative polynomial P\in{\mathbb{C}}\langle X,Y\rangle , we provide a method for determining the (ordinary) moment generating function of P(a,b) with respect to \varphi (see Theorem 4.16 ). For simplicity, our presentation focuses on the case of two variables, but this generalizes easily to the case of n variables. Our main tools are the same as in [ 18 ] where we studied polynomials in free variables: • derivations on the free algebra which translate combinatorial recursions for Boolean cumulants into effective algebraic identities between their generating functions; • linearization, which allows us to express the resolvent (1-zP(X,Y))^{-1} in the form u^{t}(I-zAX-zBY)^{-1}v for vectors u,v\in{\mathbb{C}}^{n} and n\times n matrices A and B . On the combinatorial side of this development we prove a new characterization of c -freeness in terms of Boolean cumulants (see Theorem 3.5 ). It turns out to be very similar to the characterization of freeness from [ 12 ] , and involves the so-called Vertical-No-Repeat property which identifies maximal elements among coloured noncrossing partitions with respect to a partial order denoted by \ll , which arises naturally in the context of free probability (see [ 6 ] ). Our motivation to study this problem comes from the recent developments in [ 13 ] where a surprising connection between c -freeness and so called free denoising was established, which we briefly describe below. Let (\mathcal{M},\psi) be a tracial W^{*} -probability space, i.e., \mathcal{M} is a von Neumann algebra with a faithful, normal trace \psi . Assume further that two random variables a,b are free with respect to \psi . We interpret the random variable a as a “signal” of interest and b as a “noise”, and we view P(a,b) as a noisy version of a , where P\in{\mathbb{C}}\langle X,Y\rangle . We are interested in the conditional expectation E[a|P(a,b)] , which is the best approximation of a by a function of P(a,b) with respect to the L^{2} -norm with respect to \psi . It was shown in [ 13 ] that, under the assumptions a\geq 0 and \psi(a)=1 , if we define a new state \varphi(c)=\psi(ac) then a,b are c -free with respect to the pair (\varphi,\psi) , and moreover we have E[a|P(a,b)]=h(P(a,b)) where h=d\mu_{P(a,b)}^{\varphi}/d\mu_{P(a,b)}^{\psi} is the Radon-Nikodym derivative. Methods of determining distributions of polynomials in free variables are known (see [ 5 , 18 ] ), and thus we fill this gap and provide a method to compute the Radon-Nikodym derivative above, and consequently the conditional expectation. In order to illustrate our results we present one example, which is fully discussed in Section 5 . Suppose a is a standard semicircular element and b has distribution \tfrac{1}{2}\left(\delta_{-1}+\delta_{1}\right) . Let c=i[a,b] be their commutator, then for s<1/2 we have E\bigl[(1-sa)^{-1}|c\bigr]=h(c) with h(t)=\frac{2(2+\tilde{\eta}_{a}^{2}(s))}{(1-s\tilde{\eta}_{a}^{2}(s))\left((2+\tilde{\eta}_{a}(s))^{2}-\tilde{\eta}_{a}^{2}(s)t^{2}\right)}, where \tilde{\eta}_{a}(s)=\frac{1-\sqrt{1-4s^{2}}}{2s} is the shifted Boolean transform of the standard semicircle distribution. Expanding the above power series at s=0 we obtain \displaystyle E[a^{2}|c] \displaystyle=\frac{c^{2}+2}{4}, \displaystyle E[a^{4}|c] \displaystyle=\frac{c^{4}+6c^{2}+12}{16}. The same formulas hold more generally for the anticommutator and in fact any element of the form c=\theta ab+\bar{\theta}ba , with \theta\in{\mathbb{C}} arbitrary such that \lvert\theta\rvert=1 . In the framework of general c -free random variables in a two-state W^{*} -noncommutative probability space (\mathcal{M},\varphi,\psi) it is natural to ask whether conditional expectations onto c -free subalgebras with respect to \varphi exist. This question is non-trivial because in general c -freeness is incompatible with a tracial state. Indeed the examples in Section 6 below show that in the c -free case conditional expectations in the strict sense do not exist, however the projection in L^{2}(\mathcal{M}) with respect to the inner product \langle a,b\rangle=\varphi(ab^{*}) exists and serves as a formal analog of the conditional expectation. We call this projection a right quasi-conditional expectation , as it has only the right module property and is not positive. We denote it by \rE_{b}^{\varphi} , when we compute the projection on subalgebra generated by b . We show that this projection shares some properties with the conditional expectations with respect to \varphi . In particular for any polynomial P in c -free variables \rE_{b}^{\varphi}[P(a,b)] is a polynomial in b of degree at

Read the complete text