pylcm specifies, solves, and simulates finite-horizon discrete-continuous dynamic choice models. You write states, actions, preferences, constraints, and transition laws as ordinary Python functions; pylcm turns that economic description into a backward-induction and simulation pipeline built on JAX.
Use the route that matches what you are trying to do:
Install pylcm, solve a tiny model, learn its vocabulary, and decide which solver family your model must be written for.
Define regimes, grids, shocks, transition laws, parameters, and outputs through task-oriented guides.
Start from the economic structure, then compare numerical scaling, memory, and target hardware.
Browse executable examples by economic feature, pylcm feature, and solver family, or follow links to the wider model zoo and benchmark suite.
Find signatures, declarations, capability tables, runtime controls, and every public export.
The Concepts & Methods chapter explains why the algorithms and representations work. The Development chapter is for people changing pylcm itself.
Scope¶
pylcm is for finite-horizon models with discrete and continuous states and actions, deterministic or stochastic transitions, and potentially multiple regimes. Brute-force grid search covers the broadest class. Endogenous-grid solvers cover narrower consumption-saving structures and require the model to declare those structures from the outset. GPU acceleration can make large models feasible, but the best solver still depends on candidate growth, memory, compilation, and hardware.
If finite-horizon dynamic programming is new to you, begin with the QuantEcon Dynamic Programming book before using the methods chapter.