Experiments for NCProcess
1. Prioritize categories by number of unknowns. Also (low priority)
categories that contain digested relations.
2. Focus on high priority categories.
a. Implement Dave's multiplicative
unknowns program.
b. Collect on knowns.
c. Scan each relation for expressions in parenthesis.
Set one of them to be a new letter z. Run a GB which places z and
its tp etc right above unknowns--includes relations defining inverses and
the indiviual relation R1 in which z appears. If this eliminates
all unknowns and return a relation involving only z tp[z] etc and knowns
then we have produced a 1-decomposition (or a symmetric 1-decomposion)
for R1. Thus, z is a legitimate motivated unknown.
d. Maybe we want to re-do part a. if part c fails.
NEW COMMANDS NEEDED
1. NCCategory[{A,Tp[A],B}] returns all equations in that category whose
unknowns are {A,Tp[A],B}.
motivation: this might make the output of NCProcess easier to use.