A Hybridizing-Enhanced Quantum-Inspired Differential Evolution Algorithm with Multi-Strategy for Complicated Optimization

Yu Chen, Haotian Xu, Jie Liu, Ming Hou, Yang Li, Shaopeng Qiu, Maohua Sun, Huimin Zhao, Wu DengJ. Artif. Intell. Soft Comput. Res 2026 · DOI

Abstract

High-dimensional optimization problems often suffer from the "curse of dimensionality," where traditional algorithms struggle with local optima and slow convergence. This paper presents QAHQDE, a hybrid algorithm combining quantum-inspired encoding with chaotic initialization and adaptive mutation strategies. By balancing global exploration and local exploitation through sub-population hybrid mutation, QAHQDE significantly outperforms standard quantum-inspired methods in accuracy and scalability across high-dimensional benchmarks.

Introduction

Optimization serves as the hidden engine for modern technological advancements, from logistics to deep learning. However, as problem complexity increases to include thousands of variables—a phenomenon known as the "curse of dimensionality"—traditional metaheuristics like Differential Evolution (DE) often stall or become trapped in local optima.

To address these challenges, researchers developed Quantum-inspired Differential Evolution (QDE). By using quantum bits (qubits) to represent potential solutions, QDE can explore a significantly larger state space simultaneously. Despite this advantage, standard QDE often suffers from "over-mutation," where aggressive jumps prevent the algorithm from converging on the absolute best solution. The proposed QAHQDE algorithm seeks to stabilize this quantum power while maintaining its exploratory benefits.

Methodology: The QAHQDE Framework

The Hybridizing-Enhanced Quantum-Inspired Differential Evolution Algorithm with Multi-Strategy (QAHQDE) introduces three core innovations to improve search efficiency and accuracy:

  1. Improved Chaotic Strategy: Instead of random initialization, the algorithm employs a modified Logistic map with noise and cosine smoothing. This ensures that the initial population is uniformly distributed, preventing early clustering at search space boundaries.
  2. Quantum-Adaptive Mutation: The mutation factor ($F$) is dynamically adjusted throughout the process. It begins with a high value to encourage global exploration and gradually decreases to allow for precise local fine-tuning in later stages.
  3. Sub-population Hybrid Mutation: The population is bifurcated into two distinct groups. One sub-population utilizes weighted dynamic mutation to balance search intensity, while the other uses standard random mutation to preserve diversity and prevent premature convergence.
Flowchart of the QAHQDE algorithm illustrating the multi-strategy hybrid approach.

Central to the algorithm is the use of quantum encoding, where solutions are represented as a superposition of states $(\alpha, \beta)$. This allows the algorithm to maintain a broader perspective on the problem space. The evolution of these states is guided by quantum rotation gates, which adjust the probability amplitudes of the qubits.

Illustration of the quantum rotation gate operation on a unit circle, showing the transition between states.

Experimental Evaluation

The researchers evaluated QAHQDE against 48 benchmark functions from the CEC2005, CEC2010, and CEC2013 competition sets. These functions range from simple unimodal peaks to complex multimodal traps and hybrid compositions. Testing was performed across massive dimensions ($D=100, 500, 1000, \text{ and } 3000$).

Key Results

  • Accuracy: QAHQDE outperformed standard QDE by over three orders of magnitude (1,000x) in convergence accuracy.
  • Comparative Ranking: In a head-to-head comparison with 38 other cutting-edge algorithms (including JADE, L-SHADE, and EO), QAHQDE achieved the #1 overall rank.
  • Scalability: Unlike many algorithms that degrade as dimensions increase, QAHQDE maintained high performance even at $D=3000$.
  • Efficiency: While quantum operations add complexity, QAHQDE offers a superior balance of time-to-accuracy compared to other high-performance variants like HMCFQDE.

Conclusion

QAHQDE demonstrates that the integration of structured chaos, adaptive mutation, and quantum-inspired encoding can effectively overcome the limitations of classical optimization in high-dimensional spaces. By balancing global "scouting" with local "drilling," the algorithm provides a robust solution for complex real-world optimization tasks. Future research will focus on further refining mutation mechanisms for specific function topographies and optimizing computational efficiency for ultra-high-dimensional scenarios.

Summary updated

BibTeX
@article{chen2026hybridizing,
  author = {Yu Chen and Haotian Xu and Jie Liu and Ming Hou and Yang Li and Shaopeng Qiu and Maohua Sun and Huimin Zhao and Wu Deng},
  title = {{A Hybridizing-Enhanced Quantum-Inspired Differential Evolution Algorithm with Multi-Strategy for Complicated Optimization}},
  journal = {J. Artif. Intell. Soft Comput. Res.},
  year = {2026},
  doi = {10.2478/jaiscr-2026-0001},
}
Figure