Fast optimal algorithms for the minimum rectilinear Steiner arborescence problem

In this paper, we present two optimal algorithms for solving the Minimum Rectilinear Steiner Arborescence (MRSA) Problem. The first algorithm is a recursive branch-and-bound variant of the RSA heuristic by Rao et al. (1992). The second algorithm uses dynamic programming to avoid solving the same subproblem more than once. Furthermore, both algorithms can be generalized to solve the All-Quadrant MRSA Problem. Extensive experimental results show that our algorithms significantly outperform existing exact methods for solving the MRSA problem.