The lower bound is the focus of the primary computational effort. It is computed by a branch and bound process, using a linear programming relaxation, and then tightening the relaxation with cutting planes to obtain the bound. When no more progress is being made on the bound on a problem, we branch, dividing the problem into two subproblems. Thus, the lower bound is the minimum lower bound of all of the current subproblems. The current subproblems are also referred to as the active nodes.