Can we have simple algorithms to escape saddle points in high-dim functions with better convergence rate? In my #NeurIPS2021 paper with Chenyi Zhang, we proposed a GD-based algorithm with a poly-speedup in log n. Join our poster session at spot A3 at Dec 7, 11:30-13:00 EST!

With Weiyuan Gong and Tongyang Li, we study the robustness of quantum algorithms for d-dim nonconvex optimization with noisy inputs and characterize the domains where they can find an approximate local min with polylog, poly, or exp number of queries in d. arxiv.org/abs/2212.02548

With Tongyang Li, we study quantum lower bounds on finding stationary points of nonconvex functions, and proved that there’s no quantum speedup in the following two settings: having access to 1) p-th order derivatives, or 2) stochastic gradients. arxiv.org/abs/2212.03906

We just posted a brand new version of our paper on "Quantum Pseudoentanglement," with two new authors and many new results! 😀 See scirate: scirate.com/arxiv/2211.007… 🧶 1/13 #QuantumComputing

With Hao Wang and Chenyi Zhang, we study quantum algorithms for minimizing the maximum of convex functions, and also prove quantum lower bounds showing near-optimality. The paper was accepted by ICLR 2024 with arXiv link arxiv.org/abs/2402.12745 See you in the conference!

I'm honored that this work with David Gosset and Chenyi Zhang was selected as one of three long plenary talks at QIP 2026!