Chaotic Optimization Algorithms in Global Search Problems

Chaotic optimization algorithms integrate deterministic chaotic dynamics into metaheuristic frameworks to enhance the efficiency and robustness of global search. By exploiting properties such as ergodicity and sensitive dependence on initial conditions, these methods inject complex, nonrepetitive trajectories into population initialisation, parameter adaptation or search operators. This helps to avoid premature convergence to local optima and to balance exploration and exploitation across complex landscapes. Common chaotic maps—such as the logistic, sine and Chebyshev maps—are used to generate pseudo-random sequences that replace or augment conventional random number generators. Over the past decade, chaotic variants of classical algorithms like particle swarm optimization, differential evolution and cuckoo search have shown marked improvements in convergence speed and solution quality. Applications span engineering design, image processing, scheduling and feature selection, demonstrating global significance in solving high-dimensional, multimodal problems across science and industry.

Research from Nature Portfolio

Recent studies have proposed a modified particle swarm optimization algorithm that embeds several chaotic-inspired strategies to address NP-hard scheduling problems. The algorithm uses an elite-reverse strategy to initialise the swarm with enhanced diversity, an adaptive inertia weight adjusted via a subtraction function combined with a ladder strategy, and a jump-out mechanism to escape local traps. Evaluated on standard benchmark functions and a complex vehicle scheduling problem with soft time windows, the method consistently outperformed basic PSO, an improved PSO variant and a chaotic PSO model. In a real-world mineral-transport simulation, the enhanced algorithm improved operational profit by up to 71.8% compared with baseline approaches.

Research from all publishers

A recent work in neural computing has applied chaotic initialisation to both particle swarm and a wave-based search algorithm for multi-level image thresholding. By distributing initial populations according to logistic, Chebyshev, circle, sine and piecewise maps, the chaotically initialized algorithms achieved faster convergence and higher segmentation accuracy on 23 benchmark functions and five test images. A parallel study introduced a chaotic transient search algorithm that integrates ten different chaotic maps into a physics-inspired transient search framework. Benchmarked on CEC’17 functions and five engineering design tasks, and tested for feature selection on UCI datasets, this approach demonstrated superior ergodicity, faster convergence and higher precision than the standard transient search method. Conversely, an empirical investigation compared particle swarm and simulated annealing modified with chaotic maps against their pseudorandom counterparts on common benchmark functions. Results indicated no consistent advantage of chaos in most cases, except in noisy scenarios, highlighting the nuanced and problem-dependent impact of chaotic sequences on metaheuristic performance.

Chaotic Optimization Algorithms in Global Search Problems publication trend

The graph below shows the total number of articles in chaotic optimization algorithms in global search problems across all publications each year (not limited to Nature Index journals).

Technical terms

Chaotic map: A deterministic, non-linear function that generates sequences with high sensitivity to initial conditions and rich ergodic behaviour.

Metaheuristic optimization: A class of stochastic or deterministic search strategies designed to find near-optimal solutions in complex, high-dimensional spaces without exhaustive enumeration.

Global search problem: An optimisation challenge characterised by multiple local optima, requiring broad exploration to identify the true global optimum.

Exploration and exploitation: Two complementary phases in search algorithms where exploration covers diverse regions of the search space and exploitation intensifies the search around promising candidates.

Population initialization: The process of generating the initial set of candidate solutions in population-based algorithms; using chaotic distributions can enhance diversity and sampling uniformity.

References

1. New comparative approach to multi-level thresholding: chaotically initialized adaptive meta-heuristic optimization methods. Neural Computing and Applications (2025).
2. An investigation of the effects of chaotic maps on the performance of metaheuristics. Engineering Reports (2021).
3. A modified particle swarm optimization algorithm for a vehicle scheduling problem with soft time windows. Scientific Reports (2023).
4. A novel chaotic transient search optimization algorithm for global optimization, real-world engineering problems and feature selection. PeerJ Computer Science (2023).