Dice Question Streamline Icon: https://streamlinehq.com

Maximal matrix-coefficient dimension allowing strong convergence

Determine the sharp growth regime of the matrix-coefficient dimension D_N for which strong convergence holds for noncommutative polynomials with M_{D_N}(C)-valued coefficients applied to N×N GUE or Haar unitary ensembles; specifically, close the gap between the current positive range D_N = e^{o(N)} and known counterexamples when D_N ≥ e^{C N^2}.

Information Square Streamline Icon: https://streamlinehq.com

Background

Strong convergence with matrix coefficients is known to hold for subexponential dimensions D_N = e{o(N)}, but can fail at dimensions e{C N2}. The optimal threshold between these regimes is currently unknown.

This problem connects to questions in operator space theory and to quantitative refinements of strong convergence techniques (e.g., polynomial and interpolation methods).

References

While only the case $D_N=N$ is needed for that purpose, the optimal range of $D_N$ for which strong convergence holds remains open: for both Gaussian and Haar distributed matrices, it is known that strong convergence holds when $D_N=e{o(N)}$ and can fail when $D_N\ge e{CN2}$. Understanding what lies in between is related to questions in operator space theory \S 4.

The strong convergence phenomenon (2507.00346 - Handel, 1 Jul 2025) in Section 6.6 (The optimal dimension of matrix coefficients)