Chaotic Optimization with Adaptive Mutation and Feasibility Control

Summary

This article continues a description of a chaotic optimization algorithm for searching solution spaces. It combines logistic, sinusoidal, and tent maps with an initial broad search, weighted-gradient refinement, and a final local search whose scope narrows adaptively. Inertia-based velocity vectors, stagnation checks, and mutations are intended to maintain exploration while refining promising candidates.

The installment details mutation of selected coordinates through random, global-best-relative, or chaotic-map updates, resetting velocity after a mutation. It also describes adapting a constraint penalty according to the share of feasible candidates, recording recent best values, and checking convergence from their history. Testing is conducted on optimization test functions, and the document reports comparative test results and presents strengths and drawbacks, including many tuning parameters. These benchmark results do not establish trading performance; the method is an optimization technique that would need separate evaluation on financial objectives and realistic data.

Key ideas

• The algorithm uses three chaotic maps to generate candidate search sequences.
• Its search proceeds from broad exploration to gradient-based refinement and adaptive local search.
• Mutation changes a subset of coordinates using random, best-relative, or chaotic updates.
• A penalty parameter adapts to the proportion of feasible candidates in the population.
• Benchmark comparisons on test functions are not evidence of effectiveness in live trading.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.