Transfer matrices and partition-function zeros for antiferromagnetic Potts models. VI. Square lattice with extra-vertex boundary conditions (Q648130): Difference between revisions

This is the sixth of a series of impressive papers on ``Transfer matrices and partition-function zeros for antiferromagnetic Potts models'' (for Part V, see [the authors, ibid. 135, No. 2, 279--373 (2009; Zbl 1179.82040)]). Recall that the Potts model on a regular lattice is characterized by two parameters, i.e., the (positive integer) number \(q\) of Potts spin states, and the (real) number \(v = \exp(J/kT) -1\), characterizing the nearest-neighbor coupling of spins. It was shown that the partition function of this system for any finite graph is a polynomial in \(q\) and \(v\). The particular case with \(v = -1\), corresponding to zero temperature limit, is called a chromatic polynomial and is the central topic of this series of papers for different boundary conditions (free, cylindrical, cyclic, and toroidal, for square lattices and triangular lattices). The special motivation for this new paper is the peculiar feature ``for which the authors provide detailed numerical evidence -- that the partition-function zeros are becoming dense everywhere in the complex \(q\)-plane outside a given limiting curve; but that, despite this, the infinite-volume free energy is perfectly analytical in this region''. This result is conceptually interesting and relevant, because a general statistical mechanics approach to studying phase transitions involves investigating the zeros of the finite-volume partition function when one or more parameters (as for instance the temperature) are allowed to take complex values. Lee and Yang showed that under a set of mild conditions the accumulation points of these zeros in the infinite-volume limit constitute the only possible loci of phase transitions, and that away from such accumulation points the infinite-volume free energy is analytic. However, this paper shows an explicit extreme example in which the zeros of the finite-volume partition function appear to be dense in a large region of the complex plane, while the infinite-volume free energy is perfectly analytic there. The example thoroughly worked out by the authors involves some particular boundary conditions (obtained from an \(m \times n\) grid with free boundary conditions by adjoining one new vertex adjacent to all the sites in the leftmost column and a second new vertex adjacent to all the sites of the rightmost column). Despite the highly specialized character of the paper, the conceptual relevance of the result makes it worth of attention by a wider community than that of specialists.