Sanjit Shashi

Holography, in the form of the anti-de Sitter/conformal field theory correspondence, offers a natural setting for generative modelling. Data on a boundary manifold lifts into a higher-dimensional bulk through a propagator, and this extra dimension plays the role of a flow parameter. We exploit this structure to build Generative AdS (GenAdS), an approach to generative flow matching in which the dynamics are represented by the evolution of fields in AdS, together with a residual correction learned by a neural network. Boundary samples are encoded as scalar sources, transported into the bulk along the flow, and decoded after numerical integration. Our paradigm combines a Fourier-space encoding scheme for the data as AdS sources, a normalised radial phase space in which to stage the flow-matching dynamics, and a Klein–Gordon backbone to guide the flow. On a two-dimensional checkerboard benchmark, our experiments show that most of the benefit of GenAdS comes from the Fourier representation and convolutional architecture. However, when we remove momentum-channel regularisation, our most physics-informed GenAdS variant rivals the strongest physics-free control on boundary violation. On modified National Institute of Standards and Technology, GenAdS models remain close to a convolutional baseline on fidelity while achieving significantly higher recall at comparable precision, suggesting a fidelity-coverage trade-off. Our findings establish GenAdS as a physically interpretable and experimentally controllable framework for generative modelling, with many avenues for future extension.

We present a framework for generative machine learning that leverages the holographic principle of quantum gravity, or to be more precise its manifestation as the anti-de Sitter/conformal field theory (AdS/CFT) correspondence, with techniques for deep learning and transport theory. Our proposal is to represent the flow of data from a base distribution to some learned distribution using the bulk-to-boundary mapping of scalar fields in AdS. In the language of machine learning, we are representing and augmenting the flow-matching algorithm with AdS physics. Using a checkerboard toy dataset and MNIST, we find that our model achieves faster and higher quality convergence than comparable physics-free flow-matching models. Our method provides a physically interpretable version of flow matching. More broadly, it establishes the utility of AdS physics and geometry in the development of novel paradigms in generative modeling.

Recent work has studied the thermodynamics of gravity subjected to conformal boundary conditions, where the trace of the extrinsic curvature K and the conformal class of metrics are kept fixed. The case K < 0 seems tied to solutions with cosmic horizons. To study this situation further, we analyze Einstein–Maxwell theory, where we find solutions with cosmic horizons corresponding to patches inside the inner horizon of the Reissner–Nordström black hole. We find that these solutions are necessary for consistency with the thermal effective action in the putative boundary dual at O(1) chemical potential. Furthermore, we introduce a (positive or negative) cosmological constant and find more cases where solutions with cosmic horizons dominate the ensemble. Transitions between solutions with cosmic horizons and solutions with black hole horizons are mediated by ‘extremal’ solutions that correspond to patches of (A)dS 2 × Σ d − 1 , for which we discuss an effective JT gravity description with dimensionally reduced conformal boundary conditions. Finally, we introduce rotation in some simple cases and find that the bulk theory reproduces the universal velocity-dependent modification of the free energy predicted by the thermal effective action.

In this work, we investigate the assumptions regarding spacetime backgrounds underlying the classical double copy. We argue (contrary to the norm) that single-copy fields naturally constructed on the original curved background metric are only interpretable on a flat metric when such a well-defined limit exists, for which Kerr-Schild coordinates offer a natural choice. As an explicit example where such a distinction matters, we initiate an exploration of single-copies for the Gödel universe. This metric lacks a (geodesic) Kerr-Schild representation yet is Petrov type-D, meaning the technology of the “Weyl double copy” may be utilized. The Weyl derived single copy has many desirable features, including matching the defining properties of the spacetime, and being sourced by the mixed Ricci tensor just as Kerr-Schild single copies are. To compare, we propose a sourced flat-space single-copy interpretation for the Gödel metric by leveraging its symmetries, and find that this proposal lacks the defining properties of the spacetime, and is not consistent with the flat limit of our curved-space single copy. Notably, this inconsistency does not occur in Kerr-Schild metrics. Our curved-space single copy also lead to the same electromagnetic analogue of the Gödel universe found separately through tidal force analogies, opening a new avenue of exploration between the double copy and gravitoelectromagnetism programs.

We study the thermodynamics of Einstein gravity with vanishing cosmological constant subjected to conformal boundary conditions. Our focus is on comparing the series of subextensive terms to predictions from thermal effective field theory, with which we find agreement for the boundary theory on a spatial sphere, hyperbolic space, and flat space. We calculate the leading Wilson coefficients and observe that the first subextensive correction to the free energy is negative. This violates a conjectured bound on this coefficient in quantum field theory, which we interpret as a signal that gravity does not fully decouple in the putative boundary dual.

In this work, we investigate the assumptions regarding spacetime backgrounds underlying the classical double copy. We argue (contrary to the norm) that single-copy fields naturally constructed on the original curved background metric are only interpretable on a flat metric when such a well-defined limit exists, for which Kerr--Schild coordinates offer a natural choice. As an explicit example where such a distinction matters, we initiate an exploration of single-copies for the G\"odel universe. This metric lacks a (geodesic) Kerr--Schild representation yet is Petrov type-D, meaning the technology of the ``Weyl double copy" may be utilized. The Weyl derived single copy has many desirable features, including matching the defining properties of the spacetime, and being sourced by the mixed Ricci tensor just as Kerr--Schild single copies are. To compare, we propose a sourced flat-space single-copy interpretation for the G\"odel metric by leveraging its symmetries, and find that this proposal lacks the defining properties of the spacetime, and is not consistent with the flat limit of our curved-space single copy. Notably, this inconsistency does not occur in Kerr--Schild metrics. Our curved-space single copy also lead to the same electromagnetic analogue of the G\"odel universe found separately through tidal force analogies, opening a new avenue of exploration between the double copy and gravitoelectromagnetism programs.

We propose swampland criteria for braneworlds viewed as effective field theories of defects coupled to semiclassical gravity. We do this by exploiting their holographic interpretation. We focus on general features of entanglement entropies and their holographic calculations. Entropies have to be positive. Furthermore, causality imposes certain constraints on the surfaces that are used holographically to compute them, most notably a property known as causal wedge inclusion. As a test case, we explicitly constrain the Dvali--Gabadadze--Porrati term as a second-order-in-derivatives correction to the Randall--Sundrum action. We conclude by discussing the implications of these criteria for the question on whether entanglement islands in theories with massless gravitons are possible in Karch--Randall braneworlds.

We propose swampland criteria for braneworlds viewed as effective field theories of defects coupled to semiclassical gravity. We do this by exploiting their holographic interpretation. We focus on general features of entanglement entropies and their holographic calculations. Entropies have to be positive. Furthermore, causality imposes certain constraints on the surfaces that are used holographically to compute them, most notably a property known as causal wedge inclusion. As a test case, we explicitly constrain the Dvali-Gabadadze-Porrati term as a second-order-in-derivatives correction to the Randall-Sundrum action. We conclude by discussing the implications of these criteria for the question on whether entanglement islands in theories with massless gravitons are possible in Karch-Randall braneworlds.