Life Cycle Models

This package aims to generalize and facilitate the specification, solution, and simulation of finite-horizon discrete-continuous dynamic choice models.

Installation

PyLCM can be installed via PyPI or via GitHub. To do so, type the following in a terminal (or install via uv):

$ pip install pylcm

or, for the latest development version, type:

$ pip install git+https://github.com/OpenSourceEconomics/pylcm.git

GPU Support

By default, the installation of PyLCM comes with the CPU version of jax. If you aim to run PyLCM on a GPU, you need to install a jaxlib version with GPU support. For the installation of jaxlib, please consult the jax docs.

[!NOTE] GPU support is currently only tested on Linux with CUDA 12 and 13.

Developing

We use pixi for our local development environment. If you want to work with or extend the PyLCM code base you can run the tests using

$ git clone https://github.com/OpenSourceEconomics/pylcm.git
$ pixi run tests

This will install the development environment and run the tests. You can run ty using

$ prek run ty --all-files

Before committing, install the pre-commit hooks using

$ pixi global install prek
$ prek install

Questions

If you have any questions, feel free to ask them on the PyLCM Zulip chat.

Acknowledgments

PyLCM builds on the endogenous grid method (Carroll, 2006), its discrete-continuous extension (Iskhakov, Jørgensen, Rust & Schjerning, 2017), the Fast Upper-Envelope Scan (Dobrescu & Shanker, 2022), and the broader open-source ecosystem for dynamic programming — including OpenSourceEconomics, NumEconCopenhagen, and QuantEcon. See the Credits & Acknowledgments page for the full list of methods, replicated models, and software we are grateful to.

License

This project is licensed under the Apache License, Version 2.0. See the LICENSE file for details.

Copyright (c) 2023- The PyLCM Authors