A0251
Title: Generalized linear orthogonal arrays and applications to strong orthogonal arrays
Authors: Bochuan Jiang - Beijing Jiaotong University (China) [presenting]
Abstract: Orthogonal arrays are essential for designing experiments and have found extensive applications in the big data era. A new class of orthogonal arrays called generalized linear orthogonal arrays (GLOAs), is introduced, which encompasses linear orthogonal arrays, nonlinear orthogonal arrays from the Addelman-Kempthorne construction, and numerous other nonlinear orthogonal arrays as special cases. The most compelling feature of GLOAs is their highly deterministic projection properties, which facilitate the study of the complex relationships among columns of GLOAs. These properties are leveraged to explore the applications of GLOAs in computer experiments. Two methods for constructing strong orthogonal arrays (SOAs) and column-orthogonal SOAs of strength two plus are presented using GLOAs. These methods produce space-filling designs capable of accommodating a large number of factors, providing significant flexibility in terms of run sizes, and possessing appealing low-dimensional projection properties. Therefore, these designs are ideal for computer experiments.
