Confinement in Holographic Theories at Finite Theta

Abstract

A strongly coupled confining gauge theory with a non-zero vacuum angle undergoing a deconfinement to confinement phase transition is studied in the holographic gravitational description. A simplified five-dimensional setup is constructed where a bulk scalar models the effect of the vacuum angle, and the suitable boundary conditions on the ultra-violet (UV) and the infra-red (IR) boundaries are identified. The IR boundary condition is motivated by higher dimensional examples where the bulk scalar comes from a Wilson loop on a shrinking cycle. In this five-dimensional dual geometry, and in the limit of small backreaction in the infra-red, the critical temperature for the phase transition is shown to reduce quadratically with the vacuum angle, matching lattice results. The topological susceptibility has a sharp reduction across the critical temperature, also matching lattice results. The rate for the phase transition is estimated as a function of the vacuum angle, and is seen to be enhanced (reduced) when the field theory has a relevant (irrelevant) deformation at high energies. Crucially, for the irrelevant case, the confined phase can get destabilized for a range of parameters. In the context of early universe dynamics, if the vacuum angle is time-dependent, the transition history changes strongly: the deconfined phase can last till much lower temperatures than naively expected, and one can trigger a transition to the confined phase by a change in the vacuum angle, thus providing a controlled way to generate supercooling. As a phenomenological application, the peak frequency and the power of resulting gravitational wave signal from bubble collisions change, affecting their visibility in detectors. Possible generalizations of the scenario are discussed.

Comments: v1: 36 pages, 5 figures; v2: typos fixed, citations added, discussion on topological susceptibility expanded; v3: additional discussion added on IR boundary condition, branch choice, theta = pi, domain wall, glueball mass and string tension, comparison with lattice, and GW signals. Version accepted in JHEP