Core Concepts
This paper initiates the study of binary decision diagrams (BDDs) and zero-suppressed binary decision diagrams (ZDDs) as relatively compact data structures for representing matroids in a computer. The focus is on analyzing the sizes of BDDs and ZDDs for representing matroids.
Abstract
The paper makes the following key contributions:
Comparison of different variations of BDDs and ZDDs for representing matroids:
The size of the ZDD Z(B(M)) and the BDD B(I(M)) are never greater than the BDD B(B(M)).
The BDDs B(B(M)) and B(B(M*)) have the same size.
The ZDDs Z(I(M)), Z(B(M)), and the BDD B(I(M*)) all have the same size.
Upper bounds on the width (size) of BDDs and ZDDs for representing matroids:
For free matroids, the width is at most 1.
For uniform matroids, the width is at most λ(E⪯,i) + 1.
For partition and nested matroids, the width is at most 2λ(E⪯,i).
For transversal and laminar matroids, the width is unbounded.
Improved upper bounds for partition and nested matroids using the pathwidth of the matroid:
The width is at most pathwidth(M) + 1 for a certain total order on the ground set.
Implementation of a rank oracle using the ZDD structure.
The results provide insights into the compact representation of matroids using BDDs and ZDDs, and reveal new strongly pigeonhole classes of matroids.