Deterministic Parallel Hypergraph Partitioning

TL;DR

This paper introduces a scalable deterministic parallel algorithm for hypergraph partitioning, matching or surpassing existing methods in quality and speed, with broad applications in computational and data management fields.

Contribution

It presents a novel deterministic parallel hypergraph partitioning algorithm integrated into Mt-KaHyPar, outperforming existing deterministic approaches in quality and efficiency.

Findings

Achieves similar partition quality as non-deterministic methods

Outperforms existing parallel deterministic algorithms in speed and quality

Demonstrates scalability on extensive benchmark instances

Abstract

Balanced hypergraph partitioning is a classical NP-hard optimization problem with applications in various domains such as VLSI design, simulating quantum circuits, optimizing data placement in distributed databases or minimizing communication volume in high-performance computing. Engineering parallel heuristics for this problem is a topic of recent research. Most of them are non-deterministic though. In this work, we design and implement a highly scalable deterministic algorithm in the state-of-the-art parallel partitioning framework Mt-KaHyPar. On our extensive set of benchmark instances, it achieves similar partition quality and performance as a comparable but non-deterministic configuration of Mt-KaHyPar and outperforms the only other existing parallel deterministic algorithm BiPart regarding partition quality, running time and parallel speedups.