The Setting
The observed spin set and prescribed source spin set enter the zero-field Fortuin-Kasteleyn (FK) measure as two sets of wired sites. Let be a finite ferromagnetic weighted multigraph with vertex set and separately indexed edge set . Each edge has distinct endpoints and a real coupling ; parallel edges are allowed. Fix subsets , where is the set of pinned sites.
An undeleted vertex of is a vertex outside , and an undeleted edge of is an edge whose two endpoints lie outside . The set of undeleted vertices of consists of the vertices outside , and the set of undeleted edges of consists of the edges whose two endpoints lie outside , as in (1): (1)
For an open edge set , wired open-edge adjacency is the relation on defined by (2): (2)✦ The wired cluster equivalence on is the equivalence closure of .✦ The contracted cluster count is the number of equivalence classes of on , as in (3): (3)✦
The zero-field FK weight is the random-cluster weight with cluster weight and edge probabilities , given in (4): (4)✦ Parallel edges have separate factors. Loops created by contraction contribute constants that cancel from terminal connection probabilities.
For undeleted vertices , the FK terminal connection probability is the probability that and belong to the same wired open cluster, defined by (5): (5)✦
The Claim
Theorem 1 (Terminal exchange symmetry). Let be a finite ferromagnetic weighted multigraph, let be arbitrary subsets of its vertex set, and let . Then the connection probabilities in (5) satisfy (6)✦
In the paper, the result reads: the connection probability and wired open-edge adjacency have explicit swap identities, the reversed likelihood comparison follows by reapplying the comparison theorem with and exchanged, and no separately named theorem states symmetry of the extreme log odds.
What Is Checked
The statement (6) is over finite ferromagnetic weighted multigraphs, arbitrary terminal and pinned sets, and arbitrary undeleted representatives, with the connection probability defined by the zero-field FK weights in (4). When and , the probability is independent of the representatives chosen within these respective sets.