Optimization of fixed-polarity Reed-Muller circuits using dual-polarity property

In the optimization of canonical Reed-Muller (RM) circuits, RM polynomials with different polarities are usually derived directly from Boolean expressions. Time efficiency is thus not fully achieved because the information in finding RM expansion of one polarity is not utilized by others. We show in this paper that two fixed-polarity RM expansions that have the same number of variables and whose polarities are dual can be derived from each other without resorting to Boolean expressions. By repeated operations, RM expansions of all polarities can be derived. We consequently apply the result in conjunction with a hypercube traversal strategy to optimize RM expansions (i.e., to find the best polarity RM expansion). A recursive route is found among all possible polarities to derive RM expansion one by one. Simulation results are given to show that our optimization process, which is simpler, can perform exhaustive search as efficiently as other good exhaustive-search methods in the field.