#optimization #python #algebra #ndarray #cma-es #cmaes #hansen

haru_cmaes

A simple CMA-ES optimization algorithm implementation based on Hansen's purecma Python implementation

6 releases (breaking)

0.5.1 Sep 13, 2024
0.5.0 Sep 9, 2024
0.4.0 Aug 3, 2024
0.3.0 Jul 30, 2024
0.1.0 Jul 26, 2024

#240 in Machine learning

Download history 323/week @ 2024-07-24 158/week @ 2024-07-31 13/week @ 2024-08-07 1/week @ 2024-08-14 183/week @ 2024-09-04 199/week @ 2024-09-11 17/week @ 2024-09-18 10/week @ 2024-09-25 4/week @ 2024-10-02

273 downloads per month

MIT/Apache

37KB
413 lines

CMAES in Rust

Motivation

This is my own implementation of the CMA-ES optimization algorithm based in Hansen's purecma python implementation.

This is version 0.3.0 so expect more enhancements and changes along the way.

Roadmap

Although functional at this point, the roadmap is to convert this crate to use ngalgebra as evidenced in the benchmark: eigen decomposition is faster, nice!. So, expect changes in the short term.

Other improvements will follow as well.

Stay tuned.

Simple usage example

use haru_cmaes::{
    params::CmaesParams, 
    state::CmaesState, 
    strategy::Cmaes
    fitness::square_and_sum, 
    };
use anyhow::Result;

fn example() -> Result<()> {
    let params = CmaesParams {
        popsize: 10,
        xstart: vec![0.0; 10],
        sigma: 0.75,
    };

    let cmaes = Cmaes::new(&params)?;
    let mut state = CmaesState::init_state(&params)?;
    for _i in 0..150 {
        let mut pop = cmaes.ask(&mut state)?;
        let mut fitness = square_and_sum(&pop)?;
        state = cmaes.tell(state, &mut pop, &mut fitness)?;
    }

    println!("Best y: {:+.4?}", &state.best_y);
    println!("Best y (fitness): {:+.4?}", &state.best_y_fit);

    Ok(())
}

fn main() {
    example();
}

Requirements for (ndarray and friends): BLAS algebra

I assume you have a clean brand new linux environment, so follow the following instructions. You can also refer to the working Github actions, if that helps you better.

1) Install Build Tools (GCC)

The build-essential package includes the GCC compiler and other necessary tools for building C programs whic are needed for low-level C algebra utilities wrapped by rust crates. This is most likely a requirement for BLAS C bindings used by ndarray and friends.

sudo apt install build-essential

2) Install pkg-config and OpenSSL Development Libraries

If you encounter OpenSSL and pkg-config related issues during compilation:

sudo apt install pkg-config libssl-dev

3) Setting Up Rust Dependencies

Ensure the following dependencies are specified in your Cargo.toml:

anyhow = { version = "1.0.86" }
rand = { version = "0.8.5" }
rayon = { version = "1.10.0" }
ndarray = { version = "0.15", features = ["blas", "rayon"] }
blas-src = { version = "0.10", features = ["openblas"] }
openblas-src = { version = "0.10", features = ["cblas", "system"] }
ndarray-linalg = { version = "0.16", features = ["openblas-system"] }
ndarray-rand = { version = "0.14" }

4) Installing OpenBLAS

To use OpenBLAS system-wide for ndarray and others, install the libopenblas-dev package:

sudo apt install libopenblas-dev

For Lapack do:

sudo apt-get install liblapack-dev libblas-dev

If you want to check where did it got installed dpkg-query -L libopenblas-dev

5) Additional Tools

Install cargo-depgraph and graphviz for dependency visualization:

sudo apt install graphviz
cargo install cargo-depgraph

6) Git (if needed)

Since it's a fresh ubuntu build, for git:

git config --global user.name "Your Name" git config --global user.email "your.email@example.com"

Then, check github key, if ssh -T git@github.com says git@github.com: Permission denied (publickey), then, probably the key pair was lost, due to new ubuntu fresh install, so do ls -al ~/.ssh and see if you indeed have keys stored. If not, then ssh-keygen -t ed25519 -C "youremail@example.com", ssh-add. Then add it to github.com cat ~/.ssh/ided25519.pub. Then paste that under Settings, SSH and GPG Keys and that's it.

7) Run simple example

cargo test --lib

Requirements for Benchmarks

1) Cmake

Install cmake by downloading the tar file https://cmake.org/download/, extracting it, cd into it, do ./bootstrap, then do gmake, then sudo gmake install, lastly verify with cmake --version

Dependencies

~66MB
~864K SLoC