rusde

rusde (pronounced like "rusty") implements numerical solvers for stochastic differential equations (SDEs) and their variants, e.g. delayed SDEs and stochastic integro-differential equations.

Motivation

Solving stochastic differential equations numerically can be tricky business. The complexity increases greatly when simulating their delayed or integro-differential variants. This library implements a variety of numerical schemes for solving user-specified SDEs with the correctness and speed guaranteed by Rust.

Goal Functionality

• Both forward/backward Euler-Maruyama
• Milstein
• Delayed
• Integro-differential SDEs
• Above schemes with jumps (Poisson process)
• Multi-threaded simulations
• Multi-dimensional systems with correlated noise
• Standard example processes that everybody likes and uses
• Solve PDE by stochastic approximation
• Kolmogorov forward/backward equations
• Neural SDEs with backpropagation