AQHSO: Adaptive Quantum Hybrid Swarm Optimizer

AQHSO is a state-of-the-art three-phase metaheuristic optimization algorithm that unifies Grey Wolf Optimization (GWO), Firefly Algorithm (FA), and Ant Colony Optimization (ACO) under an adaptive quantum rotation framework.

This standalone library executes locally and relies natively on mealpy and numpy.

📦 Installation

Install AQHSO directly from the repository using pip:

pip install git+https://github.com/Puneethreddy2530/Research_Paper.git#subdirectory=package

Alternatively, if you cloned the source directory:

pip install .

🚀 Quick Start Example

It integrates flawlessly as a <mealpy.Optimizer> module into any project.

import numpy as np
from aqhso import AQHSO
from mealpy.utils.space import FloatVar

# 1. Define your objective function (e.g. Sphere function)
def sphere_function(solution):
    return np.sum(solution ** 2)

# 2. Define problem boundaries
problem_dict = {
    "fit_func": sphere_function,
    "obj_func": sphere_function,
    "bounds": FloatVar(lb=[-10.0]*30, ub=[10.0]*30),
    "minmax": "min",      # Goal: minimize
}

# 3. Initialize the Adaptive Quantum optimizer
optimizer = AQHSO(
    epoch=500,           # Number of iterations
    pop_size=30,         # Agents in the swarm
    theta_max_init=0.1,  # Initial quantum rotation max angle
    levy_beta=1.5        # Heavy-tail tunneling parameter
)

# 4. Execute optimization
best_agent = optimizer.solve(problem_dict)

print(f"Optimal Minimum Discovered: {best_agent.target.fitness}")
print(f"Coordinates: {best_agent.solution}")

🧠 Architectural Phases

1. Phase 1 (Epochs 0–20%) | GWO Exploitation: Fast initial basin targeting. Uses OBL (Opposition-based learning) for initialization.
2. Phase 2 (Epochs 20–70%) | Quantum FA Attraction: Superpositioned spatial brightness evaluated strictly inside angular bounds ($\theta$-space). Dynamic stagnation handling.
3. Phase 3 (Epochs 70–100%) | Lévy Tunneling & ACO Pheromone: Final granular traps optimization utilizing mathematical heavy-tailed distributions and an historical 1-D coordinate probability matrix.