A neighborhood decoupling algorithm for truncated sum minimization

A heuristic, called the neighborhood decoupling (ND) algorithm, is described. It first selects a minterm and then selects an implicant, a two-step process employed in previous heuristics. The approach taken closely resembles the G.W. Dueck and D.M. Miller (1987) heuristic; however, it makes more efficient use of minterms truncated to the highest logic value. The authors present the algorithm, discuss its implementation, show that it performs consistently better than others, and explain the reason for its improved performance