In order to handle non-linear pseudo-Boolean constraints with a constraint solver for linear pseudo-Boolean inequalities, we have to linearize the nonlinear pseudo-Boolean constraints in advance. It is sufficient to consider only non-linear pseudo-Boolean inequalities in normal form like in Definition 3.1.8, since others can be brought into that form with (3.3). Thus, the problem is to compute a set of linear pseudo-Boolean inequalities equivalent to a nonlinear pseudo-Boolean inequality in normal form. After a review of well-known linearization techniques, we present a linearization method that fits well into the context of constraint logic programming.
Unable to display preview. Download preview PDF.