Get started¶
PYDGENS is a Python/JAX package for approximating equilibrium solutions in multi-player dynamic and differential games.
PYDGENS is a pre-1.0 release. The package is ready for early adopters, but the public API may continue to evolve as the modeling frontend, examples, and solver interfaces mature.
Installation¶
PYDGENS requires Python 3.12 or newer.
Solvers¶
PYDGENS currently supports three main solver paths:
LQ: linear-quadratic, unconstrained games solved for feedback Nash strategiesiLQ: nonlinear, unconstrained games solved for local feedback Nash strategiesAL: constrained nonlinear games solved with an augmented-Lagrangian workflow for local open-loop trajectories
Quick Start¶
Run a minimal linear-quadratic example:
Run the quick test suite:
Development¶
For local development:
For a locked contributor environment:
Project Layout¶
src/pydgens/frontend/: user-facing modeling helpers.src/pydgens/ir/: JAX-friendly intermediate representations used by solvers.src/pydgens/solvers/: numerical algorithms for equilibria, trajectories, and constrained solves.src/pydgens/examples/: runnable game examples.tests/: unit, regression, integration, and benchmark tests.