Thermal effective action found for the 3D critical O(N) vector model in the large-N limit
This paper computes the leading terms of the thermal effective action for the critical O(N) vector model in three dimensions. The authors work at large N and in the high-temperature regime, where the thermal circle is much smaller than the spatial scales. They extract concrete numbers for three coefficients that control the free energy density and the theory’s response to spatial curvature and to a Kaluza–Klein gauge field.
The basic idea is to start from the conformal field theory (CFT) on a product of a two-dimensional spatial manifold M and a thermal circle S1 of length β (the inverse temperature). When β is small one can perform a Kaluza–Klein reduction down to an effective two-dimensional local action on M. The first terms of that action include a Casimir energy density f, a curvature coupling c1, and a gauge-field strength coupling c2. Those coefficients encode universal CFT data such as the Casimir energy and the response to a Kaluza–Klein gauge field.
The authors determine the coefficients in two independent ways. First they compute a “twisted” thermal partition function on the sphere S2 with an angular potential Ω (this inserts a rotation chemical potential). Second they do a direct path-integral computation on a weakly curved, perturbed flat background. To make the path integral tractable they use a Hubbard–Stratonovich transformation that introduces an auxiliary scalar σ and then evaluate the resulting large-N effective action by saddle point. A key technical point is that, at the order studied, the saddle value of σ can be treated as spatially constant; the authors justify this using general results about perturbed optimization problems collected in their appendix.
Their main numerical results for the leading coefficients (in the large-N limit) are f = (2 ζ(3) / (5 π)) N, c1 = 0, and c2 = −(√5 log φ) / (96 π) N, where ζ(3) is the Riemann zeta function at 3 and φ = (1+√5)/2 is the golden ratio. The two independent computations give the same answers, which provides a non-trivial check of the thermal effective action framework and of their direct calculations.