Searcharxiv⌕ Search

arXiv subjects

Charles-Philippe Diez

Publications and source records attributed to Charles-Philippe Diez.

8 recordsLinked to original sources

Obata's rigidity theorem in free probability

We establish a free analogue of Obata's rigidity theorem. More precisely, Cheng and Zhou (2017) proved that on a weighted Riemannian manifold, the sharp spectral gap (Poincaré constant) is achieved only when the space splits isometrically off a one-dimensional Gaussian factor, providing an infinite-dimensional counterpart of Obata's rigidity theorem. We obtain the corresponding phenomenon in free probability, extending it beyond the setting of analytic self-adjoint potentials: Assume a self-adjoint $n$-tuple $X=(X_1,\dots,X_n)$ admits Lipschitz conjugate variables in the sense of Dabrowski (2014). Under a suitable non-commutative curvature-dimension condition, we show that any non-zero saturator of Voiculescu's free Poincaré inequality must be an affine function of the generators. Consequently, we deduce that the von Neumann algebra $M=W^*(X_1,\dots,X_n)$ necessarily splits off a freely complemented semicircular component $W^*(Y_1)\simeq L^{\infty}([-2,2],μ_{\rm sc})$, which is also maximal amenable in $M$. More generally, whenever the first eigenspace of the free Laplacian $Δ=\partial^*\bar\partial$ is finite-dimensional of rank $r\ge 1$, our rigidity argument shows that these $r$ extremal directions form a free semicircular family, yielding a free product decomposition with an $L(\mathbb{F}_r)$ factor. This provides a free-probability analogue of the classical Gaussian splitting phenomenon and reveals a rigidity mechanism under non-commutative curvature.

math.OA↗

Mathematical research with GPT-5: a Malliavin-Stein experiment

On August 20, 2025, GPT-5 was reported to have solved an open problem in convex optimization. Motivated by this episode, we conducted a controlled experiment in the Malliavin--Stein framework for central limit theorems. Our objective was to assess whether GPT-5 could go beyond known results by extending a \emph{qualitative} fourth-moment theorem to a \emph{quantitative} formulation with explicit convergence rates, both in the Gaussian and in the Poisson settings. To the best of our knowledge, the derivation of such quantitative rates had remained an open problem, in the sense that it had never been addressed in the existing literature. The present paper documents this experiment, presents the results obtained, and discusses their broader implications.

math.PR↗

A sharp symmetrized free transport-entropy inequality for the semicircular law

In this paper, following the recent work of Fathi (2018) in the classical case, we provide by two different methods a sharp symmetrized free Talagrand inequality for the semicircular law, which improves the free TCI of Biane and Voiculescu (2000). The first proof holds only in the one-dimensional case and has the advantage of providing a connection with the machinery of free moment maps introduced by Bahr and Boschert (2023) and a free reverse Log-Sobolev inequality. This case also and sheds light on a dual formulation via the free version of the functional Blaschke-Santalo inequality. The second proof gives the result in a multidimensional setting and relies on a random matrix approximation approach developed by Biane (2003), Hiai, Petz and Ueda (2004) combined with Fathi's inequality on Euclidean spaces.

math.OA↗

Free Stein Kernel and Moments maps

In this paper, we propose a free analogue to Fathi's construction of Stein kernels using moment maps (2019). This is possible for a class of measures called free moment measures that was introduced in the free case by Bahr and Boschert (2021), and by using the notion of free moment maps which are convex functions, solutions of a variant of the free Monge-Ampère equation discovered by Guionnet and Shlyakhtenko (2012). We then show how regularity estimates in some weighted non-commutative Sobolev spaces on these maps control the transport distances to the semicircular law. We also prove in the one dimensional case a free analogue of the moment map version of the Cafarelli contraction theorem (2001), discovered by Klartag in the classical case (2014), and which leads to a uniform bound on the free moment Stein kernel. Finally, we discuss the applications of these results: we prove a stability result characterizing the semicircular distribution among a certain subclass of free Gibbs measures, the probabilistic interpretation of this free moment Stein kernel in terms of free diffusion processes, its connections with the theory of non-commutative Dirichlet forms, and a possible notion of a non-commutative (free) Hessian manifold associated with a free Gibbs measure, conjecturally having free analogue properties of a classical Hessian manifold associated with a log-concave measure and considered by Kolesnikov (2012), which has the striking property of having a Ricci curvature bounded from below by $1/2$.

math.OA↗

Free Stein kernels, free moment maps, and higher order derivatives

In this work, we describe new constructions of free Stein kernels. Firstly, in dimension one, we propose a free analog to the construction of Stein kernels using moment maps as the one proposed by Fathi. This will be possible for a class of measures called the free moment measures via the notion of free moment map (convex functions), introduced in the free case by Bahr and Boschert. In a second time, we introduce the notion of higher-order free Stein kernels relative to a potential, which can be thought as the free counterpart of a recent and powerful idea introduced in the classical case by Fathi, and which generalize the notion of free Stein kernels by introducing higher-order derivatives of test functions (in our context noncommutative polynomials). We then focus our attention to the case of homothetic semicircular potentials. We prove as in the classical case, that their existence implies moments constraints. Finally, we relate these discrepancies to various metrics: the free (quadratic) Wasserstein distance, the relative free Fisher information along the Ornstein-Uhlenbeck flow or the relative non-microstates free entropy. Finally, as an important application, we provide new rates of convergences in the entropic free CLT under higher moments constraints.

math.PR↗

Sobolev-Wigner spaces

In this paper, we provide new results about the free Malliavin calculus on the Wigner space first developed in the breakthrough work of Biane and Speicher. We define in this way the higher-order Malliavin derivatives, and we study their associated Sobolev-Wigner spaces. Using these definitions, we are able to obtain a free counterpart of the of the Stroock formula and various variances identities. As a consequence, we obtain a sophisticated proof a la Ustunel, Nourdin and Peccati of the product formula between two multiple Wigner integrals. We also study the commutation relations (of different significations) on the Wigner space, and we show for example the absence of non-trivial bounded central Malliavin differentiable functionals and the absence of non-trivial Malliavin differentiable projections.

math.PR↗

Free Malliavin-Stein-Dirichlet method: multidimensional semicircular approximations and chaos of a quantum Markov operator

We combine the notion of free Stein kernel and the free Malliavin calculus to provide quantitative bounds under the free (quadratic) Wasserstein distance in the multivariate semicircular approximations for self-adjoint vector-valued multiple Wigner integrals. On the way, we deduce an HSI inequality for a modified non-microstates free entropy with respect to the potential associated with these semicircular families in the case of non-degeneracy of the covariance matrix. The strategy of the proofs is based on functional inequalities involving the free Stein discrepancy. We obtain a bound which depends on the second and fourth free cumulant of each component. We then apply these results to some examples such as the convergence of marginals in the free functional Breuer-Major CLT for the non commutative fractional Brownian motion, and we provide a bound for the free Stein discrepancy with respect to semicircular potentials for $q$-semicirculars operators}. Lastly, we develop an abstract setting on where it is possible to construct a free Stein Kernel with respect to the semicircular potential: the quantum chaos associated to a quantum Markov semigroup whose $L^2$ generator $Δ$ can be written as the square of a real closable derivation $δ$ valued into the square integrable bi-processes or into a direct sum of the coarse correspondence.

math.PR↗

Limiting behavior of large correlated Wishart matrices with chaotic entries

We study the fluctuations, as $d,n\to \infty$, of the Wishart matrix $\mathcal{W}_{n,d}= \frac{1}{d} \mathcal{X}_{n,d} \mathcal{X}_{n,d}^{T} $ associated to a $n\times d$ random matrix $\mathcal{X}_{n,d}$ with non-Gaussian entries. We analyze the limiting behavior in distribution of $\mathcal{W}_{n,d}$ in two situations: when the entries of $\mathcal{X}_{n,d}$ are independent elements of a Wiener chaos of arbitrary order and when the entries are partially correlated and belong to the second Wiener chaos. In the first case, we show that the (suitably normalized) Wishart matrix converges in distribution to a Gaussian matrix while in the correlated case, we obtain its convergence in law to a diagonal non-Gaussian matrix. In both cases, we derive the rate of convergence in the Wasserstein distance via Malliavin calculus and analysis on Wiener space.

math.PR↗