EconPDEs.jl
EconPDEs.jl solves nonlinear ODEs and PDEs that arise in economic models, especially Hamilton–Jacobi–Bellman equations. You write the local equation; the package supplies the finite-difference derivatives, upwinding, sparse Jacobians, and pseudo-transient Newton iteration.
Repository: github.com/matthieugomez/EconPDEs.jl
Installation
The package is registered in the Julia General registry:
] add EconPDEsCurrent versions require Julia 1.10 or later.
Where to start
Start with Getting started, which solves a first model end to end. Writing the PDE function is the reference for the package's naming conventions and the upwinding patterns used in most models. Boundary conditions, Time-dependent problems, and Solver and troubleshooting cover what usually matters after the first model works. The pdesolve reference lists every solver option.
The Examples in the sidebar each solve a model, plot the solution, and explain what it means economically. The overview lists them all, flagging what each demonstrates beyond the baseline workflow. Start with Neoclassical growth. The InfinitesimalGenerators page explains when to combine EconPDEs with InfinitesimalGenerators.jl for stationary distributions or lower-level finite-difference residuals.
Why EconPDEs
Julia already has good PDE packages. EconPDEs.jl has a narrower goal: solve the continuous-time HJBs that come up in economic models.
These equations have a structure that is easy to lose in a generic PDE interface. The state drift and volatility are often chosen by the agent, so the differential operator depends on the current guess for the value function. At the same time, the equation is strongly forward-looking: today's value depends on future values, future policies, and the way the state distribution moves under those policies.
EconPDEs.jl keeps that feedback loop explicit. You write the local economic equation, including the policy rules and the implied drift or volatility. The package then builds the finite-difference derivatives, boundary treatment, sparse Jacobian, and nonlinear solve needed to turn that equation into a stationary or time-dependent solution.