500x Speedup on NP-Hard Vertex Cover, a Multi-Strategy Meta-Heuristic Framework Achieving

Speedup on NP-Hard Vertex CoverA Multi-Strategy Meta-Heuristic Framework Achieving

This paper also presents high-speed enhancements for the Dominating Set (DS) problem, achieving up to 425x speedup over exact solvers.

Abstract

We present Porta Aviones v3.8, a hybrid meta-heuristic framework for solving

NP-hard combinatorial optimization problems with near-optimal quality and excep-

tional speed. On the vertex cover problem, Porta Aviones achieves 97.8% solution

quality with 503x speedup compared to Google OR-Tools CP-SAT solver across 10

independent trials (n=4,000). For large-scale instances (n=15,000), Porta Aviones

produces valid solutions in 13.5 seconds, while exact solvers time out without finding

solutions. The framework combines three key innovations: (1) Rosa Hybrid multi-

strategy parallel execution, (2) fast reduction to minimal cover in O(k · davg) time,

and (3) adaptive mode selection based on problem size. We validate the framework

on vertex cover, dominating set, and demonstrate a practical application to graph

neural network sampling where vertex cover-based sampling preserves 39% more

graph structure than random sampling. Our results demonstrate that intelligent

meta-heuristics can achieve near-optimal solutions orders of magnitude faster than

exact methods, enabling real-time optimization on real-world scale problems.

Contact: For feedback, collaborations, apirolo@abc.gob.ar

Code: The full GitHub repository for this project is forthcoming.

Current Research: I am currently focused on a new project