Split multi-terminal binary decision diagrams

A new data structure called Split Multi-terminal Binary Decision Diagrams (Split MTBDD) is introduced for representing Multi–Output logic Functions (MOF). Split MTBDDs are efficient for some functions where conventional BDDs are not. A Split MTBDD comprises interconnected MTBDD components, each associated with a “dichotomic fragment”. The “dichotomy” reflects cognitive patterns introduced by the designer of a MOF specification. The paper describes a method of transforming an arbitrary MOF into a corresponding Split MTBDD. Experimental results indicate that Split MTBDDs are more compact than conventional MTBDDs for many benchmarks. Criteria for prediction of the Split MTBDD compactness are formulated and justified.