Finite temperature lattice QCD in the large N limit

Our aim is to give a self-contained review of recent advances in the analytic description of the deconfinement transition and determination of the deconfinement temperature in lattice QCD at large N. We also include some new results, as for instance in the comparison of the analytic results with Monte Carlo simulations. We first review the general set-up of finite temperature lattice gauge theories, using asymmetric lattices, and develop a consistent perturbative expansion in the coupling beta(s) of the spacelike plaquettes. We study in detail the effective models for the Polyakov loop obtained, in the zeroth order approximation in beta(s),, both from the Wilson action (symmetric lattice) and from the heat kernel action (completely asymmetric lattice). The distinctive feature of the heat kernel model is its relation with two-dimensional QCD on a cylinder; the Wilson model, on the other hand, can be exactly reduced to a twisted one-plaquette model via a procedure of the Eguchi-Kawai type. In the weak coupling regime both models can be related to exactly solvable Kazakov-Migdal matrix models. The instability of the weak coupling solution is due in both cases to a condensation of instantons; in the heat kernel case, this is directly related to the Douglas-Kazakov transition of QCD2. A detailed analysis of these results provides rather accurate predictions of the deconfinement temperature. In spite of the zeroth order approximation they are in good agreement with the Monte Carlo simulations in 2 + 1 dimensions, while in 3 + 1 dimensions they only agree with the Monte Carlo results away from the continuum limit.