In graph theory, a threshold graph is a graph that can be constructed from a one-vertex graph by repeated applications of the following two operations:
For example, the graph of the figure is a threshold graph. It can be constructed by beginning with a single-vertex graph (vertex 1), and then adding black vertices as isolated vertices and red vertices as dominating vertices, in the order in which they are numbered.
Threshold graphs were first introduced by . A chapter on threshold graphs appears in, and the book is devoted to them.
An equivalent definition is the following: a graph is a threshold graph if there are a real number
S
v
w(v)
v,u
uv
w(u)+w(v)>S
Another equivalent definition is this: a graph is a threshold graph if there are a real number
T
v
a(v)
X\subseteqV
X
\sumva(v)\leT.
The name "threshold graph" comes from these definitions: S is the "threshold" for the property of being an edge, or equivalently T is the threshold for being independent.
Threshold graphs also have a forbidden graph characterization: A graph is a threshold graph if and only if it no four of its vertices form an induced subgraph that is a three-edge path graph, a four-edge cycle graph, or a two-edge matching.
From the definition which uses repeated addition of vertices, one can derive an alternative way of uniquely describing a threshold graph, by means of a string of symbols.
\epsilon
u
j
\epsilonuuj
\epsilonuj
\epsilonuuujuuj
Threshold graphs are a special case of cographs, split graphs, and trivially perfect graphs. A graph is a threshold graph if and only if it is both a cograph and a split graph. Every graph that is both a trivially perfect graph and the complementary graph of a trivially perfect graph is a threshold graph. Threshold graphs are also a special case of interval graphs. All these relations can be explained in terms of their characterisation by forbidden induced subgraphs. A cograph is a graph with no induced path on four vertices, P4, and a threshold graph is a graph with no induced P4, C4 nor 2K2. C4 is a cycle of four vertices and 2K2 is its complement, that is, two disjoint edges. This also explains why threshold graphs are closed under taking complements; the P4 is self-complementary, hence if a graph is P4-, C4- and 2K2-free, its complement is as well.
showed that threshold graphs can be recognized in linear time; if a graph is not threshold, an obstruction (one of P4, C4, or 2K2) will be output.