Have you found anything interesting yet?

[Updated 15 July 1999]

Yes, I have! I have obtained a complete characterization of the representability of a given signed graph. I’d like to take a moment to provide some links, then let you know where some of the ideas came from, before proceeding.

Links to my papers:

·        “Binary Signed Graphs”, in Word 2000 and html formats.

·        “GF(4) Representability of Bias Matroids of 3-connected Signed Graphs”, in Word 2000 and html formats.

·        “Vector Representability of Bias Matroids of Signed Graphs”, in Word 2000 and html formats.

Note: The above html versions of my papers use Microsoft’s conversion tools from Word 2000. Your browser may not display them.

***

My work in the area of representability has been based on the roads of study suggested by the following lemma, derived from works of Zaslavsky, Whittle and Tutte:

  • The bias matroid of a signed graph is regular iff it is GF(2)-representable, and near-regular iff it is GF(4)-representable.

Thus there are but three possibilities to consider:

  • The bias matroid is GF(2)-representable. (regular)
  • The bias matroid is GF(4)-representable but not GF(2)-representable. (near-regular but not regular)
  • The bias matroid is neither GF(2)- nor GF(4)-representable. (dyadic but not near-regular)

I now give a few more definitions integral to my theorems. A block of a graph G is a maximal subgraph B of G for which any two edges of B are contained in a polygon. Given a signed graph G, by G^ we mean the signed graph obtained by contracting all balanced blocks of G.

Theorem 1: The bias matroid of a connected signed graph G is binary iff G^ does not have a pair of vertex-disjoint negative polygons.

Corollary 1: If the signed graph G is two-connected and loopless, then its bias matroid is binary iff G does not have a pair of vertex-disjoint negative polygons.

Corollary 2: If G^ has two or more blocks, then the bias matroid of G is binary iff there is a vertex v in G^ such that

    1. The intersection of any two or more blocks of G^ is v, and
    2. The deletion of v and its incident edges from G leaves a balanced graph.

When I first wrote this page in 1996, I'd only obtained Corollary 1. Since then, I’ve come quite some way, earning my doctorate and eventually obtaining the complete characterization, including all items shown below. We need a few more definitions.

Let A, B and C be subgraphs of G, where neither A nor C is contained wholly within B. We say that B is a barrier between A and C if all paths from A to C intersect B. For a subgraph B of G, the relation ~ on the collection of edges G\B given by (x~y iff B is not a barrier between x and y) is an equivalence relation, and the subgraphs of G determined by the equivalence classes of ~ are called the bridges of B. A nontrivial bridge of B is one that contains vertices not in B – so if B is a set of vertices, then B is a cutset of G iff B has two or more nontrivial bridges.

G is said to satisfy the non-barrier condition if there exist three vertex-disjoint negative circles in G, none of which is a barrier between the other two. If G has at least two vertex-disjoint negative circles but does not satisfy the non-barrier condition, then we say it satisfies the barrier condition.

Let {a, b, c} be a vertex cutset of G with a balanced, nontrivial bridge X. We may assume that X has only positive edges. We delete X from G and replace it with a triangle of all positive edges on {a, b, c}, (but we don’t duplicate a positive edge that’s already in place). This operation is called 3-plating. I have shown that if a 3-plating of G is GF(4)-representable, then so is G. A 2-plating is defined similarly, and has the same preservation property just cited. Platings lower the number of vertices in the signed graphs in consideration, so any signed graph can be 2- and 3-plated until no more 2- and 3-platings remain, and the resultant signed graph, which is called completely plated, is GF(4)-representable iff G is.

The signed graph Pr3 has six vertices and twelve edges: vertices {v1, v2, v3, w1, w2, w3}, a positive triangle on {v1, v2, v3}, a positive triangle on {w1, w2, w3}, and negative digons on each {vi, wi}.

A cylindrical signed graph is one that can be embedded (without crossings) on a cylinder such that a circle C in G can be continuously deformed into a single point on the cylinder iff C is positive.

With these definitions, we are able to state the main result of my doctoral thesis.

Theorem 2. Let G be a completely plated 3-connected signed graph. Then G is GF(4)-representable iff one of the following holds:

1)     G satisfies the non-barrier condition and is Pr3.

2)     G satisfies the non-barrier condition and is cylindrical.

I completed these two theorems, along with a collection of other results, as part of my doctoral thesis. I graduated from Binghamton University in the Spring of 1998, and am now working as an Instructor at the University of Kentucky, and I will be looking for a full-time permanent position for the 2000-2001 school year. I have been at UK for a year now, and during my stay here I perfected the characterization begun in Theorems 1 and 2. Theorem 3 gives a complete characterization. There is one missing definition – that of CNO – but I leave that definition to the papers, to which I have placed links above.

Theorem 3. Let G be a connected signed graph. Let G^ be the signed graph obtains by contracting all balanced blocks of G.

1)     G is binary (and hence regular) iff G^ contains no pair of vertex-disjoint negative circles.

2)     G is quaternary (and hence representable over all fields except possibly GF(2)) iff it is binary or it is nonbinary but each nonloop block of G^ can be obtained from a negative digon by repeated application of a (possibly empty) sequence of the following transformations:

a)     Duplication of edges.

b)     Reversing a 2-plating or 3-plating.

c)      For a negative digon D on vertices v and w, pasting one of the following subgraphs onto the graph at v and w, possibly deleting one or both edges of D:

i)                    Pr3, where {v, w} support a negative digon D' in Pr3; we may delete any edges of D';

ii)                  A plate-cylindrical signed graph, where v and w are on the same end-circle;

iii)                A binary signed graph G0 with a balancing vertex, for which G0 has a compatible negative orientation (CNO) that is perfect from v to w.

3)      If G is neither binary nor quaternary, then G is representable over all fields of characteristic other than two, but no other fields.

Previous | Matroid Index | Next