Asymptotics of classical spin networks. With an appendix by Don Zagier (Q1938774): Difference between revisions

A spin network is a pair \((\Gamma,\gamma)\) of a cubic graph \(\Gamma\) with a cyclic ordering of the edges around its vertices, together with a colouring \(\gamma\) of the edges of \(\Gamma\) by natural numbers, satisfying some conditions. The Penrose evaluation associates a rational number \(\langle \Gamma,\gamma\rangle\) to a spin network. Physics suggests that the asymptotic behaviour of \(\langle\Gamma, k\gamma\rangle\) where \(k\) goes to infinity corresponds to a classical limit of quantum mechanics and should therefore recover Eulidean geometric quantities. For quantum spin networks, one such conjecture is the Volume Conjecture. The main result in the paper under review is that the sequence \(k\mapsto \langle \Gamma, k\gamma\rangle\) is a \(G\)--function (Theorem 1.9), implying that \(\langle \Gamma, k\gamma\rangle\) is of Nilsson type, which in turn implies the existence of an asymptotic expansion of the sequence \(\langle \Gamma, k\gamma\rangle\) for all spin networks. A second result in this paper is Theorem 1.7, which generalizes Westbury's formula for the evaluation of the `generating series' for a spin network from planar graphs to general (possibly non-planar) graphs \textit{B. W. Westbury} [Banach Cent. Publ. 42, 447--456 (1998; Zbl 0902.57010)]. The reviewer finds the combinatorics of this proof to be exceptionally beautiful. A third result is a number of explicit computations of \(6j\)-symbols (the case of \(\Gamma\) being a tetrahedron graph) using a new technique, the ``Wilf-Zielberger Method'' (Section 4.1). Finally, the appendix by Don Zagier contains computations and conjectures for the regular cube \(12j\) spin network. \textit{F. Costantino} and \textit{J. Marché} have an alternative approach to the asymptotics of spin networks, which allows for holonomies along edges [``Generating series and asymptotics of classical spin networks'', \url{arXiv:1103.5644}].