Crate metaheuristics_nature[−][src]

Expand description

A collection of nature-inspired metaheuristic algorithms.

use metaheuristics_nature::{Report, RGA, RGASetting, Solver, Task, ObjFunc};

let a = RGA::solve(
    MyFunc::new(),
    RGASetting::default().task(Task::MinFit(1e-20)),
    |_| true // Run without callback
);
let ans: f64 = a.result(); // Get the result from objective function
let (x, y): (Array1<f64>, f64) = a.parameters(); // Get the optimized XY value of your function
let history: Vec<Report> = a.history(); // Get the history reports

There are two traits Algorithm and Solver. The previous is used to design the optimization method, and the latter is a simple interface for obtaining the solution, or analyzing the result.

Solver will automatically implement for the type which implements Algorithm.

Objective Function

You can define your question as a objective function through implementing ObjFunc.

First of all, the array types are ndarray::ArrayBase. And then you should define the upper bound, lower bound, and objective function ObjFunc::fitness by yourself.

The final answer is ObjFunc::result, which is generated from the design parameters.

Random Function

This crate use 32bit PRNG algorithm to generate random value, before that, a random seed is required. The seed is generated by getrandom, please see its support platform.

Features

std: Default feature. Enable standard library function, such as timing and threading.
parallel: Enable parallel function, let objective function running without ordered, uses std::thread::spawn. Disable it for the platform that doesn’t supported threading, or if your objective function is not complicate enough.
wasm: Support for webassembly, especial for random seed generating.