On the non-existence of quasi-3 designs (Q1868838): Difference between revisions

A symmetric (or square) design is said to be quasi-3 for points if there exist two nonnegative integers \(x\) and \(y\) (called the triple containment sizes) such that for any three distinct points, the number of blocks containing all three is either \(x\) or \(y\). A square design is quasi-3 for blocks if the number of points in the intersection of any three distinct blocks takes on only two values. A design that is quasi-3 for points and also quasi-3 for blocks, with \(x=0\) in both cases, is called a semi-symmetric 3-design. A design \(D\) is quasi-3 for blocks if and only if the dual design \(D^T\) is quasi-3 for points. In this paper the authors write quasi-3 as meaning quasi-3 for points. A 2-design is said to be quasi-symmetric if the number of points in the intersection of two blocks takes on only two values. It is clear that the derived design \(D_B\) with respect to a block \(B\), a square design \(D\) that is quasi-3 for blocks, will be a quasi-symmetric design. In [Des. Codes Cryptography 18, 55-61 (1999; Zbl 0965.05023)] the authors discussed a particular family of parameters of quasi-3 designs arising from Hadamard matrices. In this paper they discuss the general existence question: ``For which \((v,k,\lambda)\) does there exist a quasi-3 \((v,k,\lambda)\) design?'' They begin by giving all the known examples of quasi-3 designs and then discuss the question of whether the dual design of a quasi-3 design is a quasi-3 design. In particular they show that this is true for all known examples. Finally they show that for all \(v < 144\) (and \(\lambda > 2\)) there are no quasi-3 designs apart from the already known ones. This leaves open the possibility of a classification. In particular it is possible that any quasi-3 design with \(y < \lambda\) and \(\lambda > 2\) has parameters \((4u^2,2u^2-u,u^2-u)\) for some \(u\).