Inverse of a Sum of Random Matrices: A Dynamical Mean-Field Approach

Burak \c{C}akmak; Manfred Opper

In the authors' words

We study $(O^{⊤} AO + B)^{- 1}$ , where $A, B \in R^{N \times N}$ are symmetric positive definite, with limiting eigenvalue distributions $F_{A}, F_{B}$ and spectra bounded and bounded away from zero, and $O$ is Haar orthogonal. We analyze the deviation of deterministic-equivalent-type approximation $Δ ≐ (O^{⊤} AO + B)^{- 1} - S^{- 1}$ where $S = B - z^{⋆} I$ . First, let $O$ be independent of $(A, B, u)$ , $∥ u ∥ = 1$ . There is sequence of a standard Gaussian vectors $g_{N}$ independent of $(A, B, u)$ with $∥ S Δ Su - \frac{τ}{N} g_{N} ∥ \to 0$ a.s. Here $z^{⋆} = - R (- χ)$ and $τ = R^{'} (- χ) (1 + η R^{'} (- χ))$ , where $R$ is the R-transform of $F_{A}$ and $(χ, η)$ are the first two inverse moments of the free additive convolution of $F_{A}, F_{B}$ . Let $K = \sum_{i \leq ⌊ N^{q} ⌋} λ_{i} u_{i} v_{i}^{⊤}$ , $q \in [0, C]$ , with unit-norm, not necessarily orthogonal $u_{i}, v_{i}$ and $lim sup_{N} max_{i} ∣ λ_{i} ∣ < \infty$ a.s. If $O$ is independent of $(A, B, {u_{i}, v_{i}})$ , then $N^{- q} tr (K Δ) \to 0$ a.s. The proofs use a dynamical mean-field approach: the generating functions of the free Appell polynomials give a linear dynamics approximating the quantity of interest in the iterated large- $N$ , large-time limit, whose fluctuations we analyze.

Main resultThe abstract does not state a limitation.

Inverse of a Sum of Random Matrices: A Dynamical Mean-Field Approach

In the authors' words

More on Probability that day

SpaceDiffusion: Over-the-Orbit Diffusion for Space Generate-and-Forward Communications

A Probabilistic Modeling Framework for Transient Debris Outcomes in Low Lunar Orbits

Total Variation Distance between Product Distributions: an Analytic Proxy

The joint exit time and exit location law of a planar Ornstein-Uhlenbeck process on annular sectors and principal axis rectangles

Sharp High-Entropy Bounds for Sums of Independent Discrete Random Variables