Think Locally, Act Globally: Perfectly Balanced Graph Partitioning
This work addresses graph partitioning for applications requiring perfect balance, offering incremental improvements in speed and solution quality.
The paper tackles the perfectly balanced graph partitioning problem by introducing a novel local improvement scheme that encodes unrestricted local searches into a model for balanced combinations, integrated into a parallel multi-level evolutionary algorithm, resulting in a fast system that improves or reproduces most best-known results.
We present a novel local improvement scheme for the perfectly balanced graph partitioning problem. This scheme encodes local searches that are not restricted to a balance constraint into a model allowing us to find combinations of these searches maintaining balance by applying a negative cycle detection algorithm. We combine this technique with an algorithm to balance unbalanced solutions and integrate it into a parallel multi-level evolutionary algorithm, KaFFPaE, to tackle the problem. Overall, we obtain a system that is fast on the one hand and on the other hand is able to improve or reproduce most of the best known perfectly balanced partitioning results ever reported in the literature.