Non-Abelian T-duality and the AdS/CFT correspondence:new N=1 backgrounds
Georgios Itsios, Carlos Nunez, Konstadinos Sfetsos, Daniel C. Thompson
TL;DR
The authors explore non-Abelian T-duality as a powerful solution-generating technique in AdS/CFT by dualizing well-understood N=1 backgrounds, notably Klebanov–Witten and Klebanov–Tseytlin/Strassler geometries. They show that KW dualizes to a type IIA/M-theory setup connected to Sicilian quivers, while KT dualizes to smooth massive IIA backgrounds with a quantized Romans mass, mapping D3/D5 charges to D6/D8 and revealing a Seiberg-like cascade in the dual theory. Importantly, key CFT data such as the central charge and entanglement entropy remain invariant under the duality (up to global factors), and IR phenomena like confinement and domain walls persist in the dual field theories. The work also extends to smooth KS and baryonic-branch geometries, uncovering how duality reshapes IR physics and suggesting deeper connections to wrapped M5-brane constructions. Overall, the paper demonstrates the utility of non-Abelian T-duality for generating novel N=1 holographic backgrounds and guiding the identification of their dual QFTs.
Abstract
We consider non-Abelian T-duality on N=1 supergravity backgrounds possessing well understood field theory duals. For the case of D3-branes at the tip of the conifold, we dualise along an SU(2) isometry. The result is a type-IIA geometry whose lift to M-theory is of the type recently proposed by Bah et. al. as the dual to certain N=1 SCFT quivers produced by M5-branes wrapping a Riemann surface. In the non-conformal cases we find smooth duals in massive IIA supergravity with a Romans mass naturally quantised. We initiate the interpretation of these geometries in the context of the AdS/CFT correspondence. We show that the central charge and the entanglement entropy are left invariant by this dualisation. The backgrounds suggest a form of Seiberg duality in the dual field theories which also exhibit domain walls and confinement in the infrared.
