Mathematical Finance Colloquium: Convergence of Policy Iteration for Deterministic Control
Monday, November 20, 2023 2pm to 3pm
Monday, November 20, 2023 2pm to 3pm
About this Event
3620 South Vermont Avenue, Los Angeles, CA 90089
Yumin Zhang, Auburn University
[in-person]
or
Join Zoom Meeting:
https://usc.zoom.us/j/94973619069?pwd=VnU5bVlMc1pzVTlEYUVaZUYyNSt6UT09
Meeting ID: 949 7361 9069
Passcode: 925028
Abstract: We study the convergence of policy iteration for (deterministic) optimal control problems. To overcome the problem of ill-posedness due to lack of regularity, we consider both discrete and semi-discrete schemes by adding a viscosity term via finite differences in space. We prove that PI for the schemes converges exponentially fast, and provide a bound on the error induced by the schemes. If time permits, I will also discuss the convergence of exploratory Hamilton--Jacobi--Bellman (HJB) equations arising from the entropy-regularized exploratory control problem. These are joint works with Wenpin Tang, Hung Tran and Xunyu Zhou.
This program is open to all eligible individuals. USC operates all of its programs and activities consistent with the university’s Notice of Non-Discrimination. Eligibility is not determined based on race, sex, ethnicity, sexual orientation or any other prohibited factor.