[Paper Review] A Bit-Parallel Russian Dolls Search for a Maximum Cardinality Clique in a Graph

Ricardo C. Corrêa, Philippe Michelon+3 more|arXiv (Cornell University)|Jul 4, 2014

Advanced Graph Theory Research27 references3 citations

TL;DR

This paper proposes an enhanced Russian Dolls Search (RDMC) algorithm for finding the maximum clique in undirected graphs, leveraging bit-parallel operations, approximate vertex coloring for pruning, and a recursive subproblem-solving strategy. The method outperforms existing combinatorial algorithms, especially on dense graphs, achieving more than twofold speedup in several cases compared to the best prior exact solvers.

ABSTRACT

Finding the clique of maximum cardinality in an arbitrary graph is an NP-Hard problem that has many applications, which has motivated studies to solve it exactly despite its difficulty. The great majority of algorithms proposed in the literature are based on the Branch and Bound method. In this paper, we propose an exact algorithm for the maximum clique problem based on the Russian Dolls Search method. When compared to Branch and Bound, the main difference of the Russian Dolls method is that the nodes of its search tree correspond to decision subproblems, instead of the optimization subproblems of the Branch and Bound method. In comparison to a first implementation of this Russian Dolls method from the literature, several improvements are presented. Some of them are adaptations of techniques already employed successfully in Branch and Bound algorithms, like the use of approximate coloring for pruning purposes and bit-parallel operations. Two different coloring heuristics are tested: the standard greedy and the greedy with recoloring. Other improvements are directly related to the Russian Dolls scheme: the adoption of recursive calls where each subproblem (doll) is solved itself via the same principles than the Russian Dolls Search and the application of an elimination rule allowing not to generate a significant number of dolls. Results of computational experiments show that the algorithm outperforms the best exact combinatorial algorithms in the literature for the great majority of the dense graphs tested, being more than twice faster in several cases.

Motivation & Objective

• To develop a more efficient exact algorithm for the maximum clique problem, which is NP-hard and central to many applications.
• To improve upon existing Russian Dolls Search (RDS) implementations by integrating advanced pruning and parallelization techniques.
• To evaluate the performance of the proposed algorithm on dense graphs, where prior methods struggle due to high computational complexity.
• To demonstrate that the Russian Dolls approach can outperform classical Branch and Bound methods in practice for hard combinatorial instances.

Proposed method

• The algorithm uses a Russian Dolls Search framework where each node in the search tree represents a decision subproblem rather than an optimization subproblem, enabling recursive decomposition of the problem.
• It applies bit-parallel operations to efficiently represent and manipulate vertex sets, significantly accelerating set operations like intersection and union.
• Approximate vertex coloring—using both standard greedy and greedy-with-recoloring heuristics—is employed to compute upper bounds and prune suboptimal branches early.
• An enhanced elimination rule is introduced to avoid generating redundant subproblems, reducing the search space without losing optimality.
• The algorithm recursively applies the same principles to each subproblem, ensuring consistency and depth in pruning and search space reduction.
• The implementation integrates techniques from Branch and Bound, such as dynamic candidate ordering based on color values, to improve pruning efficiency.

Experimental results

Research questions

• RQ1 Can the Russian Dolls Search framework be effectively adapted to the maximum clique problem with performance gains over existing exact algorithms?
• RQ2 To what extent do bit-parallel operations and approximate coloring improve the efficiency of RDS-based clique solvers?
• RQ3 How does the proposed elimination rule reduce the number of subproblems generated without compromising solution quality?
• RQ4 Does the RDMC algorithm achieve significant speedups on dense graphs, where the problem is known to be particularly challenging?

Key findings

• The RDMC algorithm outperforms the best existing exact combinatorial algorithms on the majority of dense graphs tested, with speedups exceeding 2x in several instances.
• For graphs with density above 0.8, the algorithm achieves more than double the performance of prior methods, demonstrating strong scalability on hard instances.
• The use of bit-parallel operations reduced computation time by accelerating set operations, particularly in subgraph induction and neighborhood queries.
• The integration of approximate coloring via greedy and recoloring heuristics significantly improved pruning efficiency, reducing the number of explored nodes.
• The enhanced elimination rule successfully pruned a substantial number of redundant subproblems, contributing to faster convergence.
• On the 1/3-all and 2/3-all graph families with 1000 vertices and 50% edge density, RDMC solved instances in under 100 seconds, outperforming PBBMC and other state-of-the-art solvers.