Multi-level and mixed-level \(k\)-circulant supersaturated designs (Q1744032)

Reducing considerably the experimental costs in fractional factorial experimental design, supersaturated designs (SSDs) are very often used in practice. In order to improve the method, (SSD)-procedures are determined by optimizing the function \(E(f_{NOD})\) describing the chi-square distance between the two-factor projections and the uniform distribution, hence, a distance between the design and an orthogonal array of strength \(2\). Thus, using this criterion for assessing the mixed-level (SSDs) from the viewpoint of orthogonality and uniformity, optimized balanced \(k\)-circulant multi-level and mixed-level designs are presented. Moreover, further lower bounds of \(E(f_{NOD})\) are derived.