The idea of Markov point processes (Ripley and Kelley, 1977) suggests adding the clique of next higher order to get a process that permits positive attraction of pairs of points. Define
the number of triples of points that are mutual neighbours, where `neighbour' has the same definition as for the Strauss process (separation by distance no greater than r). Define t(x) to be the 3-dimensional vector (n(x), s(x), w(x)), where s(x) given by (1.16) is the same as in the Strauss process. The exponential family having canonical statistic t(x) and unnormalized densities (1.3) is called here the triplets process.
