This is an implementation of the policy optimization algorithm introduced in my paper Dynamic Output-feedback Synthesis Orbit Geometry: Quotient Manifolds and LQG Direct Policy Optimization.
This repo holds the methods for conducting gradient descent (GD) on the Linear-Quadratic-Gaussian (LQG) cost. It also performs Riemannian gradient descent (RGD) with respect to the Krishnaprasad-Martin (KM) metric introduced in the above paper.
LQG_methods.py holds the methods container of everything needed to run policy optimization on your LQG problem setup.
main.py runs a simple GD and RGD on a randomly generated system
experiment.py runs the exact experiment I included in the paper above.
Feel free to ask any questions on confusing parts, or raise issues for hidden bugs! :)
-Spencer Kraisler