Automatic generation of numerical redundancies for nonlinear constraint solving (Q1371177)

The authors present a framework for cooperation of symbolic and propagation-based numerical solvers over the real numbers. The basic technique used in the paper consists of fixed points theorems for closure operators over a complete lattice of constraint systems. The second part contains a particular cooperation scheme, where propagation is associated to pruning operators implementing interval algorithms enclosing the possible solutions of constraint systems, whereas symbolic methods are mainly devoted to generating redundant constraints. When carefully chosen, it is well known that the addition of redundant constraints drastically improves the performances of systems based on local consistency. The authors propose here a method, which computes sets of redundant polynomials and show on some benchmarks the advantages of such computations.