Replicator Solver

A Rust implementation of a replicator-style dynamical system solver, designed for deterministic and stochastic (noise-driven) simulations with reproducible JSON outputs suitable for downstream analysis workflows.

This crate targets models of the general form $$ \frac{d\nu_i}{dt}=\nu_i\left(f_i(\nu,\ldots)-\sum_j \nu_j f_j(\nu,\ldots)\right), $$ with optional noise processes and (optionally) spatial extensions, while preserving the simplex constraints $\nu_i \ge 0$ and $\sum_i \nu_i = 1$ via standardized sanitization and renormalization routines.

Highlights

• Deterministic and noisy dynamics (configurable noise models via Noise / NoiseKind)
• Simplex-safe state handling (sanitization, cutoff, renormalization)
• Structured outputs written as JSON tables for analysis pipelines
• Testable, reproducible simulations (fixed seeds where applicable)
• Designed for both unit-test driven development and parameter sweeps

Repository layout

Typical layout (adapt as needed):

• src/
  • solver/ or src/solver.rs: core time-stepping logic (e.g., ReplicatorSolver)
  • noise/ or src/noise.rs: noise models and stochastic update rules
  • ecosystem/ (optional): tracking tables and IO utilities
  • tests/ (optional, internal module tests): Rust unit tests compiled with the crate
• tests/
  • Rust integration tests (recommended if you want cargo test --test <name>)
• examples/
  • Small runnable examples for common configurations
• Cargo.toml