PyApso

PyApso is an implementation of the Adaptive Particle Swarm Optimization (APSO) algorithm and designed for solving continuous, multi-objective, and multi-dimensional optimization problems.

Installation

Install the package using pip:

pip install pyapso

Or from source:

git clone https://github.com/smohamadabedy/pyapso.git
cd pyapso
pip install .

Features

ADAPTIVE - key parameters: (inertia weight, avg weight) Continuous optimization support Single and multi-objective fitness evaluation Constraint-aware evolution Batch evaluation and parallel execution Excel and JSON logging Custom callbacks and visualizations

Usage

from pyapso import APSO
import numpy as np

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

if __name__ == "__main__":
    best, score, history, hd = APSO(
        objective_function=fitness,
        dim=1,
        bounds=([-10], [10]),
        num_particles=100,
        max_iter=10,
        verbose = 1,live_plot=False,folder="results_folder",save_prefix="results_file"
    ).run()    
            
    print("Best solution:", best)
    print("Best score:", score)

    # Save optimization history (JSON,CSV)
    hd.save()

    # Plot optimization history
    hd.plot()

Modes

apso.run("avg", c1=1.8, c2=1.8, w_min=0.5, w_max=0.9)

"avg" mode adapts the inertia weight dynamically between w_min and w_max using swarm behavior.

apso.run("inertia",c1=1.8, c2=1.8, inertia=0.05)

"inertia" mode adapts a fixed inertia weight throughout the optimization (default).

💡 Tip:

• A higher c1 encourages particles to explore their own search path.
• A higher c2 encourages convergence to the global best.
• Lower inertia means the swarm is more reactive and less explorative.

Examples

You can find runnable demos in the examples/ directory:

examples/
├── demo_1.py    # 5D Rastrigin function
├── demo_2.py    # 3D constrained benchmark
├── demo_3.py    # 2D McCormick function

Run an example:

python examples/demo_2.py