Flexipaths in matroids

Although the unavoidable minors of large 3-connected matroids were found nearly thirty years ago, there has been little progress on solving the corresponding problem for large 4-connected matroids. This paper aims to take a step towards solving that problem. The objects of study here are 4-paths, that is, sequences (L, P₁, P₂, , Pₙ, R) of sets that partition the ground set of a matroid so that the union of any proper initial segment of parts is 4-separating. Viewing the ends L and R as fixed, we call such a partition a 4-flexipath if (L, Q₁, Q₂, , Qₙ, R) is a 4-path for all permutations (Q₁, Q₂, , Qₙ) of (P₁, P₂, , Pₙ). A straightforward simplification enables us to focus on (4, c) -flexipaths for some c in \1, 2, 3\, that is, those 4-flexipaths for which (Qᵢ) = c and (Qᵢ Qⱼ) > c for all distinct i and j. Our main result is that the only non-trivial case that arises here is when c = 2. In that case, there are essentially only two possible dual pairs of (4, c) -flexipaths when n 5. A key technique in the proof of this result is of independent interest. We construct the clonal core of a partitioned matroid. From the relatively simple structure of this clonal core, we can deduce many properties of the original partitioned matroid from the local connectivities between unions of parts of the partition.