Rough Stochastic Pontryagin Maximum Principle and an Indirect Shooting Method

About

Code for reproducing results in (T. Lew, "Rough Stochastic Pontryagin Maximum Principle and an Indirect Shooting Method", available at https://arxiv.org/abs/2502.06726, 2025).

Using this code

To reproduce results, move to the scripts folder cd scripts and run:

• Example for open-loop optimal control:

python example_openloop.py

• Example for feedback optimal control:

python example_feedback.py

• Compare solutions to open-loop and feedback problems:

python compare_openloop_feedback.py --solve --plot

• Solve the open-loop for different hyperparameters via the direct and indirect methods:

python compare_hyperparameters.py  --solve-comparison --sample-sizes-sweep

Setup

This code was tested with Python 3.10.12 on Ubuntu 22.04.5.

We recommend installing the package in a virtual environment. First, run

python -m venv ./venv
source venv/bin/activate

Upgrade pip:

python -m pip install --upgrade pip

Then, install all dependencies (numpy, scipy, jax, osqp, matplotlib, pytest, tqdm) by running

python -m pip install -r requirements.txt

and the package can be installed by running

python -m pip install -e .

Testing

The following command should run successfully:

python -m pytest