Efficient incomplete block designs through linear integer programming

Abstract

The purpose of this paper is to present a linear integer programming approach to construct efficient binary incomplete block designs for any given number of treatments v, number of blocks b, with common block-size k, and with a nearly balanced concurrence matrix. The proposed approach is illustrated by constructing an efficient incomplete block design. A-efficient and D-efficient incomplete block designs have been constructed and catalogued using the proposed algorithm for a restricted range of parameters 3 ⩽ v ⩽ 20, b ⩾ v, and 2 ⩽ k ⩽ min(10, v − 1), with vb⩽1, 000. An R package is developed to implement the proposed approach.