A variable partition approach for disjoint decomposition

This paper deals with the problem of determining the set of best free and bound variables (variable partitioning problem) for disjoint serial decomposition, such that the decomposed functions are smaller in size and have maximal don´t cares. A pruned breadth first search (PBFS) approach is proposed to determine the set of good variable partitions with minimal time and computational complexity. The set of best variable partitions are selected from the set of good variable partitions by (1) determining the number of block partitions in the output partition of the predecessor function, and (2) heuristic evaluation of the number of don´t cares that may be introduced during the encoding of the decomposed functions. The proposed approach has been successfully implemented and tested with MCNC and Espresso benchmarks