The graph in Figure 2 shows the threshold algorithm, and the dynamic algorithm with selection percentages of 30% and 50%. The x-axis shows the number of fitness evaluations, and the y-axis shows the percentage of experimental runs achieving a satisfactory solution (one line shown for each of the three algorithms) or a valid solution (again, one line per algorithm), as a function of the number of fitness evaluations. The x-axis is truncated at 250 evaluations to show more detail. Little improvement is seen after that, and in practice it would probably be more effective to restart the best algorithm after a smaller number of evaluations if a satisfactory solution has not been found.