Closing the gap: Near-optimal Steiner trees in polynomial time

The authors enhance the Iterated 1-Steiner (I1S) MRST heuristic of A.B. Kahng, and G. Robins (1992). For typical nets, the methods obtain average performance of less than 0.25% from optimal, and produce optimal solutions in 90% of cases. The authors generalize I1S to higher dimensions, and prove that in 2-D and 3-D the MST degree of any point can be made /spl les/4 and /spl les/14, respectively. Aside from reducing the running times of the algorithms, these results are of independent theoretical interest.<
>