We propose a dynamic extension of the Petford-Welsh coloring algorithm that estimates the chromatic number of a graph without requiring k as an input. The basic algorithm is based on the model that is closely related to the Boltzmann machines that minimize the Ising model Hamiltonian. The method begins with a minimal coloring and adaptively adjusts the number of colors based on solution quality. We evaluate our approach on a variety of graphs from the DIMACS benchmark suite using different initialization strategies. The results show that the algorithm designed is not only capable of providing near optimal solutions but also is very robust. We demonstrate that our approach can be surprisingly effective on real-world instances, although more adaptive or problem-specific strategies may be needed for harder cases. The main advantage of the proposed randomized algorithm is its inherent parallelism that may be explored in future studies.
Bihani et al. (Thu,) studied this question.