Using 3 ways to solve the problem: Tabu Search, Hybrid Evolutionary Algorithms and HEA in Duet.

Tabu08.cpp uses algorithm from Using tabu search techniques for graph coloring.

HEA02.cpp uses algorithm from Hybrid Evolutionary Algorithms for Graph Coloring.

HEAD01.cpp uses algorithm from Variations on memetic algorithms for graph coloring problems.

DSJC500.5.col is the test instance.

The last one is published in Feb, 2018 and is the fastest one among these 3 ways. And its best solution is K=47.

Tabu search is the foundation of other 2 ways.

Thanks for chenfengkg 's article 禁忌搜索算法解决图着色问题. Tabu08.cpp borrowed his ideas and I made some changes in judgment condition to make it more concise.

The main difference from his code is the way to select move, shown as following.

if (tmp_delt <= delt && ( iter> h_tabu[j]|| (tmp_delt + f)<best_f)) { // 1&&(2||3)
    if (tmp_delt < delt) {//当前解小于本次迭代最优解,则重新开始保存解
    equ_count = 0;
    delt = tmp_delt;
    }
    equ_delt[equ_count][0] = i;
    equ_delt[equ_count][1] = j;
    equ_count++;
}

 int tmp = rand() % equ_count;//有多个解时，随机选择
    sel_vertex = equ_delt[tmp][0];
    sel_color = equ_delt[tmp][1];

Using DSJC500.5.col to test.

K=49

K=49, L=4000

K=48,L=10000

平均迭代频率（次/s) 67.20430108

K=48,L=10000