The structure of the 4-separations in 4-connected matroids (Q651044): Difference between revisions

\textit{W. H. Cunningham} and \textit{J. Edmonds} [``A combinatorial decomposition theory,'' Can. J. Math. 32, 734--765 (1980; Zbl 0442.05054)] considered the structure of 2-separations in a 2-connected matroid \(M\) and showed that there is a labeled tree that displays all 2-separations of \(M\). Further \textit{J. Oxley}, \textit{C. Semple} and \textit{G. Whittle} [``The structure of the 3-separations of 3-connected matroids,'' J. Comb. Theory, Ser. B 92, No. 2, 257--293 (2004; Zbl 1059.05032)], described the structure of 3-separations in a 3-connected matroid the way in which such 3-separations can interlock led them to define the following equivalence relation: In a 3-connected matroid \(M\), two exactly 3-separating partition \((X_1,Y_1)\) and \((X_2,Y_2)\) of \(E(M)\) are equivalent that is if \(fcl(X_1) = fcl(X_2)\) and \(fcl(Y_1) = fcl(Y_2)\) (where \(fcl(X)\) means \textit{full closure} of \(X\)). A 3-separation \((X, Y)\) is sequential if \(Y\) or \(X\) is sequential that is \(fcl(X) = E(M)\) or \(fcl(Y)=E(M)\). For all 3-connected matroids having at least nine elements tree decomposition is given in [Oxley, Semple and Whittle, loc. cit.] that guarantees to display up to equivalence all non-sequential 3-separations of the matroid. Some of the vertices of this tree decomposition are labeled by flower vertices, a flower being a structure that was introduced to deal with crossing separations. This notion has been generalized by \textit{J. Aikin} and \textit{J. Oxley} [``The structure of crossing separations in matroids,'' Adv. Appl. Math. 41, No. 1, 10--26 (2008; Zbl 1139.05010)]. For integers known exceeding one, a partition \((P_1, P_2, P_3\dots P_n)\) of ground set \(E\) of a matroid \(M\) is a \(k\)-flower with petals \(P_1, P_2, P_3,\dots,P_n\), if each \(P_i\) is exactly \(k\)-separating. The aim of this paper is to generalize this main result of [Oxley, Semple and Whittle, loc. cit.] by giving a tree decomposition got the 4-separations in a 4-connected matroid and hence to establish Theorem: Let \(M\) be 4-connected matroid having at least 17 elements and \(T\) be a 4-tree for \(M\). Then every non-sequential 4-separation of \(M\) is 2-eqivalent to a 4-seperation of \(M\) is 2-equivalent to a 4-separation displayed by \(T\). If \(M\) has no non-sequential 4-separations, then the tree \(T\) consisting of a single bag vertex labeled by \(E(M)\) satisfies the theorem. If \(M\) has a non-sequential 4-separation \((R, G)\) then we know from Lemma 5.3 that: \((X, E-X)\) is a non-sequential 4-separation of a 4-connected matroid \(M\), then there is a tight irredundant maximal flower that displays a 4-separation 2-equivalent to \((X, E-X)\), consequently we will have a tight irredundant maximal flower that displays a 4-separation 2-equivalent to \((R, G)\). And therefore from Corollary 5.2: Tight irredundant maximal flowers for 4-connected matroids on at least 17 elements are partial 4-trees. It follows that there is a partial 4-tree that displays a 4-separation 2-equivalence to \((R, G)\). Let \(T\) be a maximal partial 4-tree for \(M\). Then clearly \(T\) has at least on edge and now by applying lemma 5.4 to \(T\) for \(M\). Then clearly \(T\) has at 17 elements. That is let \(M\) be a 4-connected matroid with \(|E(M) | \geq 17\) and let \(T\) be a partial 4-tree for \(M\) having at least one edge. If \(M\) has a non-sequential 4-separation \((w, E-w)\) that is not 2-equivalent to any 4-separation displayed by \(T\). Then there is a partial 4-tree \(T'\) such that \(T' \geq T\) and \(T'\) displays some non-sequential 4-separation that is not 2-equivalent to any 4-separation displayed by \(T\). That establishes the proof of the main theorem 5.1. stated above.