simsim is a framework for (sim)ple (sim)ulation of generalized, discrete-time dynamical systems written in pure Rust. A user implements a system, a state transition function, and a specification of the data they'd like to record, and simsim runs this simulation—and structures its results—in a straightforward manner.

1. Implement a system struct that inherits from the BaseSystem trait.
2. Implement a state struct that inherits from the BaseState trait.
3. Given the above, instantiate a Simulation object.
4. Call the Simulation's run method.

For example, for a Lotka-Volterra simulation, these steps are respectively implemented here:

1. System.
2. State.
3. Simulation.
4. Run.

To run the Lotka-Volterra simulation above, run the following commands:

git clone [email protected]:cavaunpeu/simsim.git
cargo run \
  --manifest-path examples/lotka_volterra/Cargo.toml \
  -- \
  --runs 10 \
  --steps_per_run 5 \
  --output_dir output \
  --configs_path examples/lotka_volterra/configs.json

In simsim, you create your own (JSON) configs—however you see fit. For instance, you can do this manually, or programmatically in a language like Python.

We hope that forcing the user to explicitly provide all configs they desire to run simplifies the experience of using this tool.

Presently, we write the following two files to your specified --output_dir:

1. results.csv: A .csv containing all of your State's (key, val) pairs in each step of each run.
2. params.csv: Those (key, val) pairs in your config that you designate as system-level parameters. Nominally, these are constants that characterize a given configuration.

Simply, we recommend duplicating the full Lotka-Volterra example then replacing its code with your own.

Additionally, if you are working from a fresh repository, you can specify the simsim dependency in your Cargo.toml as such:

[dependencies]
simsim = { git="https://github.com/cavaunpeu/simsim.git" }

Please open an issue, or DM me on Twitter.