Title: A Canonical Partition of the Primes of Logic Functions

Abstract: This paper presents a set of theorems and algorithms related to the problem of finding a minimum-cost sum-of-primes representation of a logic function f. This problem has received much attention because of its many applications. (For example, a stage in a multi-level logic circuit is modeled by a sum-of-products logic function.) The algorithms generate a canonical partition of the primes of f into independent disjoint primesets,
f = sum(FreePrimes) + sum(PS1) + . . . + sum(PSn) + sum(< Surplus >) where
FreePrimes = the essential primes + primes covered by the essential primes,
PS1,...,PSn are disjoint canonical sets of non-free primes,
< Surplus >= non-free primes that cannot be part of any minimum-cost solution. The partition of the primes holds for any cost function that assigns a positive cost to each prime. The algorithms calculate the members of FreePrimes, PSi and < Surplus >.
The key to the results is the study of objects called ancestor sets. The Ancestor Theorem proves that a minimum-cost sum-of-primes for a logic function must equal the essential primes plus the sum of the minimum-cost covers QSi (contained in PSi) of canonical independent ancestor sets Ai. The ancestor sets are easy to generate when the calculation of the primes is within computational scope. The worst-case estimate of the total number of primes associated with a logic function is exponential. Thus the immediate practical use of the result is limited to logic functions for which the primes are within computational range. However, it is likely that these theoretical results may be used to improve the methods currently in use to solve large problems.