An Updated Experimental Evaluation of Graph Bipartization Methods

We experimentally evaluate the practical state-of-the-art in graph bipartization (Odd Cycle Transversal (OCT)), motivated by need for good algorithms embedding problems into near-term quantum computing hardware. assemble a preprocessing suite of fast input reduction routines from OCT and Vertex Cover (VC) literature compare algorithm implementations using Quadratic Unconstrained Binary Optimization literature. also generate corpus frustrated cluster loop graphs, which have previously been used to benchmark annealing The diversity these graphs leads harder instances than existing benchmarks. In addition combinatorial branching solving directly, we study various reformulations other NP-hard such as VC Integer Linear Programming (ILP), enabling use solvers CPLEX. find that heuristic solutions with time constraints under second, iterative compression jump-started solution perform best, after point highly tuned solver like CPLEX is worthwhile. Results on exact are split between ILP formulations branch-and-reduce solver. extend our results large synthetic establishing robustness potential generalize domain data. total, over 8,000 evaluated, compared previous canonical 100 graphs. Finally, provide all code data an open source suite, including Python API accessing algorithms, along scripts fully replicating results.