Multi-Terminal Binary Decision Diagrams: An Efficient DataStructure for Matrix Representation

  • Authors:
  • M. Fujita;P. C. McGeer;J. C.-Y. Yang

  • Affiliations:
  • Fujitsu Laboratories of America, 3350 Scott Blvd., Bldg #34, Santa Clara, CA 95054;Cadence Berkeley Laboratories, 1919 Addison St. #303, Berkeley, CA 94704;Center for Integrated Systems, Stanford University On Leave Currently at YAHOO! Inc.

  • Venue:
  • Formal Methods in System Design
  • Year:
  • 1997

Quantified Score

Hi-index 0.01

Visualization

Abstract

In this paper, we discuss the use of binary decisiondiagrams to represent general matrices. We demonstrate thatbinary decision diagrams are an efficient representation forevery special-case matrix in common use, notably sparsematrices. In particular, we demonstrate that for any matrix, theBDD representation can be no larger than the correspondingsparse-matrix representation. Further, the BDD representation isoften smaller than any other conventional special-caserepresentation: for the n×n Walshmatrix, for example,the BDD representation is of size O(log n). No otherspecial-case representation in common use represents this matrixin space less than O(n²). We describe termwise, row, column,block, and diagonal selection over these matrices, standard anStrassen matrix multiplication, and LU factorization. Wedemonstrate that the complexity of each of these operations overthe BDD representation is no greater than that over any standardrepresentation. Further, we demonstrate that complete pivotingis no more difficult over these matrices than partial pivoting.Finally, we consider an example, the Walsh Spectrum of a Booleanfunction.