Computation of Spectral Coefficients

The advances in DD representations for discrete valued functions in terms of computational efficiency can be exploited in the calculation of the spectra of Boolean functions. The fundamental matrix approach and the fast transforms derived from the matrix structures were discussed in Chapter 3. This chapter begins with a review of cube list based approaches for computing spectral coefficients after which the computation of spectral coefficients based on output probabilities is considered in some depth. Implementation of all these approaches using DDs is then addressed. A recent method based on Cayley graphs for the computation of the Walsh spectrum is discussed. The chapter concludes with a discussion of incompletely-specified functions and the calculation of their spectra