Discriminating Between Superior UE(s2)-optimal Supersaturated Designs

For binary factors, a design is supersaturated for the main effects model if the number of runs is smaller than the number of factors. Supersaturated designs (SSDs) cannot have all orthogonal columns, and so, the traditional notions of D-, A-, E-optimality are not applicable here. SSDs are studied under criteria such as E(s(2)) or UE(s(2)) which are near-orthogonality measures. In this work, following some of the latest works, we provide algorithms to construct better UE(s(2))-optimal designs. We also provide a few design examples to demonstrate the proposed algorithms.