Learning convex optimization control policies

This repository accompanies the paper Learning convex optimization control policies. It contains the source code for the examples therein as IPython notebooks.

Our examples make use the Python package cvxpylayers to differentiate through convex optimization problems.

Abstract

Many control policies used in various applications determine the input or action by solving a convex optimization problem that depends on the current state and some parameters. These types of control policies are tuned by varying the parameters in the optimization problem, such as the linear quadratic regulator weights, to obtain good performance, judged by application-specific metrics. Our paper introduces a method to automate this process, by adjusting the parameters using an approximate gradient of the performance metric with respect to the parameters. Our procedure relies on recently developed methods that can efficiently evaluate the derivative of the solution of a convex optimization problem with respect to its parameters.

Citing

@article{agrawal2019cocp,
    author       = {Agrawal, Akshay and Barratt, Shane and Boyd, Stephen and Stellato, Bartolomeo},
    title        = {Learning Convex Optimization Control Policies},
    journal      = {arXiv},
    archivePrefix = {arXiv},
    eprint = {1912.09529},
    primaryClass = {math.OC},
    year    = {2019},
}