On linear programming bounds for nets

It is well known that there are close connections between low-discrepancy point sets and sequences on the one hand, and certain combinatorial and algebraic structures on the other hand. E. g., Niederreiter 1] showed the equivalence between (t, t + 2, s)-nets and orthogonal arrays of strength 2. Some years later this was generalized and made precise in the work of Lawrence 2] as well as Mullen and Schmid 3] by introducing ordered orthogonal arrays. This large class of combinatorial structures yields both new constructions and bounds for the existence of nets and sequences. The linear programming bound for ordered orthogonal arrays was first derived by Martin and Stinson 4]. As in the case of error-correcting codes and orthogonal arrays, it yields a very strong bound for ordered orthogonal arrays, and consequently for nets and sequences. Solving linear programming problems in exact arithmetics is very time-consuming. Using different approaches to reduce the computing time, we have calculated the linear programming bound for a wide parameter range. (© 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim)