Projects per year
Abstract
Combinatorial optimization problems are ubiquitous in real-world applications, yet the increasing complexity and dynamic nature of real-world problems continue to outpace traditional algorithm design based on intuitions. While heuristic and meta-heuristic frameworks offer an effective path forward, it is non-trivial to design effective and efficient ones in practice, relying more on the researcher’s expertise than on a unified methodology. This lack of structured design principles often leads to brittle solvers that fail to address those complex and hard problems. In this paper, we synthesize a set of core design lessons, which are learned from our experience in solving combinatorial optimization problems, particularly in capacitated arc routing problems as a representative case. These lessons are organized into a conceptual algorithm-design framework, comprising three core dimensions: problem characterization, search-space and representation design, and search strategy design, together with learning-enabled adaptation as a cross-cutting mechanism. Ultimately, this perspective provides an algorithm design framework for researchers to systematically investigate the next generation of robust, self-evolving optimization solvers.
| Original language | English |
|---|---|
| Article number | 2743482 |
| Journal | Cogent Engineering |
| Volume | 13 |
| Issue number | 1 |
| Early online date | 8 Oct 2026 |
| DOIs | |
| Publication status | Published - 2026 |
Bibliographical note
Publisher Copyright:© 2026 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group.
Funding
This work was supported by the Internal Grants (DR26F7, SDS25A3, ISRG252602) of Lingnan University.
Keywords
- Evolutionary computation
- capacitated arc routing problems
- evolutionary algorithms
- heuristics
- meta-heuristic optimization
Fingerprint
Dive into the research topics of 'A perspective on algorithm design for combinatorial optimization: lessons learned from solving capacitated arc routing problems'. Together they form a unique fingerprint.Projects
- 1 Active
Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver