Uploaded October 2026 | Updated November 2026, 3 days ago
Robert Wang (University of Waterloo)
https://simons.berkeley.edu/talks/robert-wang-university-waterloo-2026-10-13
Probabilistic Phenomena on Graphs, Groups, and Manifolds
Recently, sharp matrix concentration inequalities were developed using the theory of free probability. The main observation underlying these results is that certain random matrix models that are "highly non-commutative" are well-approximated by deterministic operators. This phenomenon was dubbed "intrinsic freeness". In this talk, we will show how the intrinsic freeness property can also be used for efficient computation by derandomizing these new concentration inequalities. Using basic tools from free probability, we develop deterministic algorithms to construct matrix outcomes that satisfy the high-probability eigenvalue bounds. As direct consequences, we obtain polynomial time deterministic algorithms for construction of near-Ramanujan expander graphs.
Robert Wang (University of Waterloo)
https://simons.berkeley.edu/talks/robert-wang-university-waterloo-2026-10-13
Probabilistic Phenomena on Graphs, Groups, and Manifolds
Recently, sharp matrix concentration inequalities were developed using the theory of free probability. The main observation underlying these results is that certain random matrix models that are "highly non-commutative" are well-approximated by deterministic operators. This phenomenon was dubbed "intrinsic freeness". In this talk, we will show how the intrinsic freeness property can also be used for efficient computation by derandomizing these new concentration inequalities. Using basic tools from free probability, we develop deterministic algorithms to construct matrix outcomes that satisfy the high-probability eigenvalue bounds. As direct consequences, we obtain polynomial time deterministic algorithms for construction of near-Ramanujan expander graphs.










