An Approximate Solution to the Minimum Vertex Cover Problem: The Hvala Algorithm
Significance
This paper presents Hvala, a linear-time ensemble approximation algorithm for the Minimum Vertex Cover problem. Hvala combines several complementary linear-time strategies — a maximal-matching (2)-approximation, a bucket-queue maximum-degree greedy method, the degree-(1) weighted-reduction Hallelujah heuristic, and redundant-vertex pruning — and returns the smallest cover found. The paper proves unconditionally that Hvala runs in (\mathcal{O}(n+m)) time and space and achieves worst-case approximation ratio at most (2) on every finite simple graph. It also reports a broad empirical evaluation on (252) instances, including NPBench graphs, real-world networks, and adversarial graph families. The significance is not a claim to beat known hardness barriers for all graphs, but a rigorously bounded, reproducible, linear-time algorithm with strong empirical performance and a public implementation through the `hvala` Python package.
Abstract
Key Findings
* Introduces Hvala, a linear-time ensemble algorithm for Minimum Vertex Cover.
* Combines maximal matching, maximum-degree greedy, Hallelujah degree-(1) weighted reduction, and redundant-vertex pruning.
* Proves unconditional (\mathcal{O}(n+m)) time and space complexity.
* Proves unconditional worst-case approximation ratio (\rho\le 2) for every finite simple graph.
* States a conditional strict pointwise bound (|S|<2\cdot\mathrm{OPT}(G)), dependent on the imported Hallelujah hypothesis.
* Evaluates the algorithm on (252) graph instances across NPBench, real-world networks, and adversarial graph families.
* Reports mean approximation ratio (1.015) and maximum (1.192) on the (173) instances with published reference values.
* Separates certified approximation ratios from ratios-to-best-known reference values.
* Provides public package availability via PyPI as `hvala`.
