pswarm 0.1.1

A particle swarm optimization implementation

pswarm A clean, simplified implementation of the Particle Swarm Optimization algorithm for parameter optimization problems. Installation pip install pso-optimizer Features • Simple, well-documented implementation of PSO • Customizable inertia weight decay • Parameter-specific boundary handling • Built-in early stopping • Optimization history tracking • Utility functions for visualization and result saving Quick Start import numpy as np from pso_optimizer import PSOOptimizer, plot_convergence, save_results

Define objective function (example: minimize a simple quadratic function)

def objective_function(x): return -np.sum(x**2) # Negative because PSO maximizes by default

Define parameter bounds

param_min = np.array([-5.0, -5.0, -5.0]) param_max = np.array([5.0, 5.0, 5.0])

Create optimizer

optimizer = PSOOptimizer( objective_function=objective_function, param_min=param_min, param_max=param_max, swarm_size=30, max_iterations=100, maximize=False, # We're minimizing save_history=True )

Run optimization

best_params, best_fitness, history = optimizer.optimize()

Print results

print(f"Best parameters: {best_params}") print(f"Best fitness: {best_fitness}")

Plot convergence

plot_convergence(history, "convergence.png")

Save results

save_results(best_params, best_fitness, ['x', 'y', 'z'], "results.txt") Advanced Usage Special Boundary Handling You can specify special boundary handling for certain parameters:

Define parameter with wrapping boundary (like an angle)

special_boundaries = { 2: {'type': 'wrap'} # Parameter at index 2 will wrap around its bounds }

optimizer = PSOOptimizer( # ... other parameters ... special_boundary_handling=special_boundaries ) Loading Parameters from Configuration from pso_optimizer import load_parameter_bounds_from_config

Load parameter bounds from YAML config

param_min, param_max = load_parameter_bounds_from_config('config.yaml')

optimizer = PSOOptimizer( # ... other parameters ... param_min=param_min, param_max=param_max ) License This project is licensed under the MIT License - see the LICENSE file for details.