If the root of the remaining branch-and-bound tree changes, but the new root has been processed earlier, then it can be advantageous to optimize the corresponding subproblem again, in order to get improved conditions for fixing variables by reduced costs. Therefore, we provide the reoptimization of a subproblem. The difference to the ordinary optimization is that no branching is finally performed even if the subproblem is not fathomed. If it turns out during the reoptimization that the subproblem is fathomed, then we can fathom all subproblems contained in the subtree rooted at this subproblem.