jiangpy 0.1.0

A swarm intelligence optimization library for simple, rapid construction of improved strategies

jiangpy

A swarm intelligence optimization library with 120+ algorithms and a pluggable strategy system.

Design: Algorithm = Base Optimizer + Pluggable Strategies

Installation

pip install jiangpy

For visualization support:

pip install jiangpy[visualize]

Quick Start

import numpy as np
from jiangpy import FloatVar, PSO, HillClimbing, LatinHypercube

def objective(x):
    return np.sum(x ** 2)

problem = {
    "bounds": FloatVar(lb=-10.0, ub=10.0, n_dims=30),
    "minmax": "min",
    "obj_func": objective,
}

model = PSO(pop_size=50, epoch=500)
model.with_init(LatinHypercube())
model.with_local_search(HillClimbing(step_size=0.01))
result = model.solve(problem)
print(f"Best fitness: {result.target.fitness}")

Features

• 120+ optimization algorithms across swarm, evolutionary, physics, math, bio, and human-inspired categories
• Pluggable strategies for initialization, selection, crossover, mutation, local search, and replacement
• Parallel evaluation support (single / thread / process)
• Multi-objective optimization via weighted sum
• Benchmark functions including classical and CEC benchmarks
• Hyperparameter tuning via grid search

Documentation

Run the examples:

python -m jiangpy.examples_usage

Requirements

• Python >= 3.8
• numpy >= 1.20.0
• matplotlib >= 3.3.0 (optional, for visualization)

License

MIT