Random-cluster representation of the Blume-Capel model (Q867037)

M. Blume (1966) has introduced a variant of the Ising model with the physical motivation of studying magnetization in Uranium Oxide at a temperature about \(30^0\)K, with the Hamiltonian \(\mathcal{H}\sigma=-J\sum_{\langle x,y \rangle}\sigma_x\sigma_y+D\sum_{x\in V}\sigma_x^2-h\sum_{x\in V}\sigma_x\), \(\sigma\in\{ -1,0,+1 \}^V\), \(J\), \(D\), \(h\) are constants. In the special case \(h=0\) H.W. Capel (1967) has calculated the existence of a first-order phase transition when \(1/3J\delta\log 4<D<1/2J\delta\), and a second-order phase transition when \(D<1/3J\delta\log 4\), with the prediction that for \(D>1/2J\delta\) zero states would be dominant. The aim of the reviewed article is to demonstrate a random-cluster representation for the Blume-Capel model with \(h=0\). The Blume-Capel model has three local states and allows the extension to Blume-Capel-Potts (BCP) model with a local state space \(\{ 0,1,2,\ldots,q \}\), \(q\geq 1\). It is shown how to construct a random-cluster representation of the BCP model, named here diluted-random-cluster model. It has three parameters, a vertex parameter \(a\), an edge parameter \(p\), and a cluster weighting factor \(q\) and is formulated on a finite graph with boundary conditions on a hypercubic lattice. Two types of stochastic orderings of measures are introduced for the study of phase transitions. These techniques permit a rigorous study of a part of the phase diagram of the Blume-Capel model.