Quantum Speedups for Nonconvex Optimization via Quantum Tunneling Walks

Explore quantum speedups for nonconvex optimization problems through the innovative quantum tunneling walk (QTW) algorithm in this hour-long lecture. Delve into the challenges of classical algorithms in solving nonconvex optimization problems with high barriers between local minima. Learn how QTW leverages quantum tunneling's global effect to achieve quantum speedups over classical stochastic gradient descents (SGD) in specific scenarios. Examine a constructed double-well landscape demonstrating QTW's efficiency in hitting target wells compared to classical algorithms. Gain insights into the conditions where QTW outperforms classical methods, particularly when barriers are high but thin and minima are flat. Discover the collaborative research findings, supported by numerical experiments, and access the full paper for an in-depth understanding of this quantum optimization approach.

Tongyang Li: On Quantum Speedups for Nonconvex Optimization via Quantum Tunneling Walks