A geometric approach to forbidden minors for GF(3) (Q794659)

Let \({\mathfrak F}(q)\) denote the class of minor-minimal matroids that are not representable over GF(q). \textit{G.-C. Rota} [Actes Congr. Int. Math. 1970, 3, 229-233 (1976; Zbl 0362.05044)] has conjectured that for all prime powers q, \({\mathfrak F}(q)\) is finite. This conjecture, one of the most important unsolved problems in matroid theory, is the motivation for this paper. The classes \({\mathfrak F}(2)\) and \({\mathfrak F}(3)\) are known explicitly and contain one and four members respectively. However, apart from some recent work of the author and P. D. Seymour [to appear] on \({\mathfrak F}(4)\), there is little more known on this problem. The author argues in this paper that a complete determination of \({\mathfrak F}(q)\) in general is impossible and therefore an answer to Rota's conjecture will have to rely on the development of techniques that do not depend on explicitly constructing the members of \({\mathfrak F}(q)\). He then gives a new proof that \({\mathfrak F}(3)\) is finite by showing that every matroid in \({\mathfrak F}(3)\) has rank at most four. This proof is based on a general theorem of \textit{W. M. Kantor} [J. Comb. Theory, Ser. A 18, 12-26 (1975; Zbl 0312.05018)] concerning representations of matroids. It also makes essential use of \textit{T. H. Brylawski} and \textit{D. Lucas's} result [Colloq. Int. Comb. Teorie, Roma 1973, Tomo I, 83-104 (1976; Zbl 0392.51007)] that a matroid representable over GF(3) is uniquely representable over that field. The author pinpoints the failure of the last result for larger fields as the fundamental obstacle to further progress on Rota's conjecture.