A variant of probe is suggested, by which, using a point infeasible to a linear programming solution space, a set of constraints can be identified whose intersection will contain at least one feasible point of the linear program. The authors implement this idea to solve lp-relaxations of the set covering problem.
