On the Minimization of Longest Path Length for Decision Diagrams

In this paper, we consider the minimization of the longest path length (LPL) for binary decision diagrams (BDDs) and heterogeneous multi-valued decision diagrams (MDDs). Experimental results show that: (1) For many logic functions, node minimization of BDDs also minimizes the LPLs of BDDs. (2) When we use heterogeneous MDDs for representing logic functions and minimize the memory sizes of heterogeneous MDDs, we can reduce both memory sizes and LPLs to 86% and 72% of corresponding BDDs, respectively. On the other hand, when the memory limitations are set to the memory sizes of BDDs, the LPLs can be reduced to 54% of BDDs.