Table of Contents
Fetching ...

Free cumulants and freeness for unitarily invariant random tensors

Benoit Collins, Razvan Gurau, Luca Lionni

TL;DR

This work extends free probability to LU-invariant random tensors by defining trace-invariants encoded by D-tuples of permutations, and by introducing tensorial free cumulants that describe first-order asymptotics under two scaling regimes: pure Gaussian-like tensors and mixed Wishart-like tensors. The authors show that tensorial free cumulants are additive under sums of independent tensors and formulate tensor freeness via the vanishing of mixed first-order cumulants, with rigorous finite-$N$ and asymptotic moment-cumulant relations using Weingarten calculus. They develop tensorial probability spaces that generalize non-commutative probability spaces, where limiting tensorial variables generate algebras whose traces are governed by melonic (dominant) invariants; pure and mixed cases share the same combinatorics but differ in the defining invariants at first order. The paper lays foundational machinery to study asymptotic tensor freeness and paves the way for applications in quantum information and random tensor models. Overall, it provides a coherent framework to analyze how LU-invariant random tensors converge in distribution and how tensor freeness arises from their combinatorial melonic structure.

Abstract

We address the question of the asymptotic description of random tensors that are local-unitary invariant, that is, invariant by conjugation by tensor products of independent unitary matrices. We consider both the mixed case of a tensor with $D$ inputs and $D$ outputs, and the case where there is a factorization between the inputs and outputs, called pure, which includes the random tensor models extensively studied in the physics literature. The finite size and asymptotic moments are defined using correlations of certain invariant polynomials encoded by $D$-tuples of permutations, up to relabeling equivalence. Finite size free cumulants associated to the expectations of these invariants are defined through invertible finite size moment-cumulants formulas. Two important cases are considered asymptotically: pure random tensors that scale like a complex Gaussian, and mixed random tensors that scale like a Wishart tensor. In both cases, we derive a notion of tensorial free cumulants associated to first order invariants, through moment-cumulant formulas involving summations over non-crossing permutations. The pure and mixed cases involve the same combinatorics, but differ by the invariants that define the distribution at first order. In both cases, the tensorial free-cumulants of a sum of two independent tensors are shown to be additive. A preliminary discussion of higher orders is provided. Tensor freeness is then defined as the vanishing of mixed first order tensorial free cumulants. The equivalent formulation at the level of asymptotic moments is derived in the pure and mixed cases, and we provide an algebraic construction of tensorial probability spaces, which generalize non-commutative probability spaces: random tensors converge in distribution to elements of these spaces, and tensor freeness of random variables corresponds to tensor freeness of the subspaces they generate.

Free cumulants and freeness for unitarily invariant random tensors

TL;DR

This work extends free probability to LU-invariant random tensors by defining trace-invariants encoded by D-tuples of permutations, and by introducing tensorial free cumulants that describe first-order asymptotics under two scaling regimes: pure Gaussian-like tensors and mixed Wishart-like tensors. The authors show that tensorial free cumulants are additive under sums of independent tensors and formulate tensor freeness via the vanishing of mixed first-order cumulants, with rigorous finite- and asymptotic moment-cumulant relations using Weingarten calculus. They develop tensorial probability spaces that generalize non-commutative probability spaces, where limiting tensorial variables generate algebras whose traces are governed by melonic (dominant) invariants; pure and mixed cases share the same combinatorics but differ in the defining invariants at first order. The paper lays foundational machinery to study asymptotic tensor freeness and paves the way for applications in quantum information and random tensor models. Overall, it provides a coherent framework to analyze how LU-invariant random tensors converge in distribution and how tensor freeness arises from their combinatorial melonic structure.

Abstract

We address the question of the asymptotic description of random tensors that are local-unitary invariant, that is, invariant by conjugation by tensor products of independent unitary matrices. We consider both the mixed case of a tensor with inputs and outputs, and the case where there is a factorization between the inputs and outputs, called pure, which includes the random tensor models extensively studied in the physics literature. The finite size and asymptotic moments are defined using correlations of certain invariant polynomials encoded by -tuples of permutations, up to relabeling equivalence. Finite size free cumulants associated to the expectations of these invariants are defined through invertible finite size moment-cumulants formulas. Two important cases are considered asymptotically: pure random tensors that scale like a complex Gaussian, and mixed random tensors that scale like a Wishart tensor. In both cases, we derive a notion of tensorial free cumulants associated to first order invariants, through moment-cumulant formulas involving summations over non-crossing permutations. The pure and mixed cases involve the same combinatorics, but differ by the invariants that define the distribution at first order. In both cases, the tensorial free-cumulants of a sum of two independent tensors are shown to be additive. A preliminary discussion of higher orders is provided. Tensor freeness is then defined as the vanishing of mixed first order tensorial free cumulants. The equivalent formulation at the level of asymptotic moments is derived in the pure and mixed cases, and we provide an algebraic construction of tensorial probability spaces, which generalize non-commutative probability spaces: random tensors converge in distribution to elements of these spaces, and tensor freeness of random variables corresponds to tensor freeness of the subspaces they generate.
Paper Structure (137 sections, 37 theorems, 295 equations, 23 figures)

This paper contains 137 sections, 37 theorems, 295 equations, 23 figures.

Key Result

Lemma 4.2

The functions $d_\mathrm{m}( [ {\bm{\sigma}} ]_{\mathrm{m}}, [{\bm{\tau}}]_{\mathrm{m}})\in \mathbb{N}$ and $d_\mathrm{p}([ {\bm{\sigma}}]_{\mathrm{p}}, [{\bm{\tau}}]_{\mathrm{p}})\in \mathbb{N}$ are distance functions between the equivalence classes, in particular: and analogously for $d_\mathrm{p}$ and $S_n^D /{\sim_\mathrm{p}}$.

Figures (23)

  • Figure 1: Left: The labeled planar bipartite map encoded by $\sigma=(12345)(876)$ and $\tau=(15)(3)(624)(78)$. Right: Two permutations satisfying $\tau\preceq \sigma$: $\sigma=(12345)(6789)$ and $\tau=(123)(45)(68)(7)(9)$. For each cycle of $\sigma$, the restriction of $\tau$ induces a non-crossing permutation
  • Figure 2: Left: a mixed tensor $A_{i^1\ldots i^D ; j^1 \ldots j^D}$ with input indices $j^c$ (resp. output $i^c$) represented as half-edges of color $c$ attached to the black (resp. white) vertex. Right: the pure case with $T_{i^1\dots i^D}$ and $\bar{T}_{j^1 \dots j^D}$.
  • Figure 3: Left: a trace-invariant for a mixed tensor $A$ with $D=3$ inputs and 3 outputs. Right: a similar invariant for the pure case with two tensors, $T$ with $D=3$ outputs and $\bar{T}$ with $3$ inputs.
  • Figure 4: Convention adopted for encoding index summations with permutations, mixed on the left and pure on the right.
  • Figure 5: Left: the only $D$-colored graph (here $D=3$) with two vertices is melonic. Right: a pair of vertices linked by $D-1$ edges is replaced by an edge.
  • ...and 18 more figures

Theorems & Definitions (54)

  • Remark 4.1
  • Lemma 4.2
  • Lemma 4.3
  • Remark 4.4
  • Theorem 4.5
  • proof
  • Proposition 4.6: Mixed case
  • Proposition 4.7: Pure case
  • Remark 4.8
  • Proposition 4.9
  • ...and 44 more