Pairwise balanced designs with prescribed minimum dimension (Q2250055): Difference between revisions

In this paper, the authors construct linear spaces with a prescribed set of sizes for the lines, admitting a minimum dimension, and having any large enough numerically admissible number of points. The dimension is defined as the maximum positive integer \(d\) such that any set of \(d\) points is contained in a proper subspace (a subspace being a subset of points closed under taking all points of each joining line). The construction goes roughly as follows. One starts with an affine space of large enough dimension, and then applies standard tricks like substituting a line with a subspace of higher dimension and smaller line sizes, deleting points, adding a point if a parallelism is given. The construction gives some freedom in the number of points, leading to intervals which overlap for large enough size. Comment of the reviewer: the dimension of a linear space is very general, but in many cases one is only interested in the dimension of certain types of linear spaces, and then one only considers subspaces of the same type to define the dimension. As an example, we correct a slight oversight of the authors, when they remark that the dimension of an affine space coincides with the dimension of the underlying vector space. This is, with the general definition, only true over a field with at least 3 elements since otherwise one needs to require that the dimension is the maximum positive integer \(d\) such that any set of \(d\) points is contained in a proper \textit{affine} subspace (meaning, the line through a point of the subspace parallel to a line of the subspace also belongs to the subspace). Indeed, every subset of points is a subspace of the linear space where each line has size 2 (a complete graph).