Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Explore quantum work extraction from an agential perspective in this 48-minute quantum lunch seminar that examines how agents with limited classical memory can design protocols to maximize extracted work from unknown quantum states. Delve into two distinct scenarios: first, learn about adaptive strategies for extracting work from finite sequences of identical but unknown pure qubit states, where the proposed approach achieves cumulative work dissipation scaling polylogarithmically with sample number—an exponential improvement over conventional measure-then-extract protocols. Second, investigate sequences of quantum states with temporal correlations modeled by classical Hidden Markov Models, discovering how agents can leverage structural knowledge for more effective work extraction. Understand the phase transition in parameter space where adaptivity enhances performance, and examine upper bounds on adaptive agent performance using computational mechanics methods and reinforcement learning tools including dynamic programming.
