Neural networks optimized solely on crossing symmetry reconstruct CFT correlators from minimal input data to few-percent accuracy across generalized free fields, minimal models, Ising, N=4 SYM, and AdS diagrams.
The analytic bootstrap at finite temperature
9 Pith papers cite this work. Polarity classification is still indexing.
9
Pith papers citing it
abstract
We propose new universal formulae for thermal two-point functions of scalar operators based on their analytic structure, constructed to manifestly satisfy all the bootstrap conditions. We derive a dispersion relation in the complexified time plane, which fixes the correlator up to an additive constant and theory-dependent dynamical information. At non-zero spatial separation we introduce a formula for the thermal two-point function obtained by summing over images of the dispersion relation result obtained in the OPE regime. This construction satisfies all thermal bootstrap conditions, with the exception of clustering at infinite distance, which must be verified on a case-by-case basis. We test our results both in weakly and strongly-coupled theories. In particular, we show that the asymptotic behavior for the heavy sector proposed in~\cite{Marchetto:2023xap} and its correction can be explicitly derived from the dispersion relation. We combine analytical and numerical results to compute the thermal two-point function of the energy operator in the $3d$ Ising model and find agreement with Monte Carlo simulations.
citation-role summary
background 2
baseline 1
method 1
citation-polarity summary
representative citing papers
A thermal normal-ordering scheme yields systematic epsilon-expansions for thermal observables in PT-symmetric cubic and quintic O(N) models, agreeing with exact 2D results from minimal models M(2,5) and M(3,8)_D and providing higher-d extrapolations.
Establishes correspondence between flat, thermal, and defect conformal partial waves via shadow formalism, obtaining thermal blocks from flat four-point and defect two-point functions and reducing the Casimir equation diagonally.
Neural networks trained on crossing symmetry accurately reconstruct conformal correlators from minimal inputs due to alignment between their spectral bias and CFT smoothness.
Thermal two-point functions in holographic AdS black holes receive exponentially small non-perturbative corrections controlled by null geodesics bouncing off the singularity.
A neural-network approach with dispersion relations handles infinite OPE towers in thermal conformal correlators without positivity.
Conformal ladder integrals are represented via thermal free energies of massive scalars, obey a second-order differential equation in even dimensions at any loop order, and admit an all-loop resummation for arbitrary D.
citing papers explorer
-
Neural Spectral Bias and Conformal Correlators I: Introduction and Applications
Neural networks optimized solely on crossing symmetry reconstruct CFT correlators from minimal input data to few-percent accuracy across generalized free fields, minimal models, Ising, N=4 SYM, and AdS diagrams.
-
$\mathcal{PT}$-symmetric Field Theories at Finite Temperature
A thermal normal-ordering scheme yields systematic epsilon-expansions for thermal observables in PT-symmetric cubic and quintic O(N) models, agreeing with exact 2D results from minimal models M(2,5) and M(3,8)_D and providing higher-d extrapolations.
-
Thermal conformal partial waves from flat-space and defect CFT
Establishes correspondence between flat, thermal, and defect conformal partial waves via shadow formalism, obtaining thermal blocks from flat four-point and defect two-point functions and reducing the Casimir equation diagonally.
-
Neural Networks Reveal a Universal Bias in Conformal Correlators
Neural networks trained on crossing symmetry accurately reconstruct conformal correlators from minimal inputs due to alignment between their spectral bias and CFT smoothness.
-
Imprint of the black hole singularity on thermal two-point functions
Thermal two-point functions in holographic AdS black holes receive exponentially small non-perturbative corrections controlled by null geodesics bouncing off the singularity.
-
Neural Networks, Dispersion Relations and the Thermal Bootstrap
A neural-network approach with dispersion relations handles infinite OPE towers in thermal conformal correlators without positivity.
-
A thermal representation for conformal ladder integrals
Conformal ladder integrals are represented via thermal free energies of massive scalars, obey a second-order differential equation in even dimensions at any loop order, and admit an all-loop resummation for arbitrary D.
- Bouncing singularities in Schwarzschild: a geometric origin of the QNM convergence region
- Bouncing singularities and thermal correlators on line defects