I am working on a simulated annealing algorithm for graph coloring. I am following this model, but I am having troubles understanding the cooling schedule and more specifically, the section with the variable M. In my understanding M represents the number of iterations after which the temperature should be changed.
Also, as I understand it, the algorithm focuses on finding a proper color distribution, given the chromatic number and not on finding its value. Since there is no focus on finding the chromatic number, but rather on a coloring with cost low enough, I decided to try "repurpose" the algorithm by setting an upper boundary of colors via Brook's Theorem and using binary search to adjust the target cost, eventually finding a solution with low chromatic number that works. This model, however, may prove slower, so I wanted to ask if you could suggest any better solutions?
Thank you :)