Resumen
We continue to study heuristics that can be applied to solve the problem of minimizing the states of nondeterministic finite automata by the branch and bound method, or rather, to the implementation of the most difficult stage of the solution associated with finding the minimum cover of an auxiliary logical matrix by special subsets of its elements with the value 1 (true). A new type of heuristic is described (the ?big step? heuristic) and various combinations of this heuristic and the two types of previously described heuristics are considered. An auxiliary heuristic based on the use of additional information about the analyzed matrix is also described. This auxiliary heuristic allows to more accurately assess the effectiveness of various combinations of basic heuristics.Along with new heuristics, one of the approaches to the analysis of the results of numerical experiments is described. It is based on constructing a reference set of the best quasi-optimal solutions and studying the distributions of the obtained solutions for different variants of the algorithm with respect to this set.