Fast Algorithm for Computing Spectral Transforms of Boolean and Multiple-Valued Functions on Circuit Representation

  • Authors:
  • René Krenz;Elena Dubrova;Andreas Kuehlmann

  • Affiliations:
  • -;-;-

  • Venue:
  • ISMVL '03 Proceedings of the 33rd International Symposium on Multiple-Valued Logic
  • Year:
  • 2003

Quantified Score

Hi-index 0.00

Visualization

Abstract

In this paper we present a fast algorithm for computingthe value of a spectral transform of Boolean or multiple-valuedfunctions for a given assignment of input variables.Our current implementation is for arithmetic transform, becauseour work is primarily aimed at optimizing the performanceof probabilistic verification methods. However, thepresented technique is equally applicable for other discretetransforms, e.g. Walsh or Reed-Muller transforms. Previousmethods for computing spectral transforms used truthtables, sum-of-product expressions, or various derivativesof decision diagrams. They were fundamentally limited bythe excessive memory requirements of these data structures.We present a new algorithm that partitions the computationof the spectral transform based on the dominator relationsof the circuit graph representing the function to be transformed.As a result, the presented algorithm can handlelarger functions than previously possible.