FIRST-ORDER OPTIMIZATION METHODS FOR STOCHASTIC DYNAMIC PROGRAMMING
Abstract
This project aims to study the foundation of numerical optimization algorithms for solving challenging stochastic dynamic programming problems. Stochastic dynamic programming enables sequential decision making under uncertainty, and thus supports critical DoD missions in search-and-rescue, unmanned aerial vehicle (UAV) control, and resource allocation etc. Mathematical optimization problems arising in this area are notoriously difficult due to high dimensionality, inherent nonconvexity, and unknown underlying transition models.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Apr 20, 2023
- Source ID
- FA95502210447
Entities
People
- Guanghui Lan
Organizations
- Air Force Office of Scientific Research
- Georgia Tech Research Corporation
- United States Air Force