Stochastic Systems

Beginning Fall 2024, I am organizing a working group in data-driven modeling and control of stochastic systems. The object is to study a general framework for systems whose systems are governed by stochastic dynamics. These systems may be discrete or continuous in both space and time. Moreover, we account for the occurrence and real-life ramifications of extreme events, and are thus interested in models that can accurately capture their probabilities and and consequences. This perspective gives way to a class of ``risk-sensitive'' stochastic optimal control problems.

Structure:

The group will begin as an (informal) course in stochastic optimal control and reinforcement learning. The course will build on the topics course ``Stochastic Control and Learning,'' taught by Misha Chertkov in Fall 2023. While our exposition will be mostly self-contained, topics will largely be selected based on the interests of the group, possibly resulting in the novel synthesis of eclectic topics. We plan to collaboratively develop notes to be appended to Prof. Chertkov's course notes.

Modality: 

The group will meet approximately once a month in a hybrid modality. The remainder of our formal interaction will be asynchronous, utilizing the platforms Piazza and Discord. This will allow the participants the flexibility to contribute on their own terms, and also encourage the careful documentation of detailed discussions.