One-loop matrix element emulation with factorisation awareness

Abstract

In this article we present an emulation strategy for one-loop matrix elements. This strategy is based on the factorisation properties of matrix elements and is an extension of the work presented in arXiv:2107.06625. We show that a percent-level accuracy can be achieved even for large multiplicity processes. The point accuracy obtained is such that it dwarfs the statistical accuracy of the training sample which allows us to use our model to augment the size of the training set by orders of magnitude without additional evaluations of expensive one-loop matrix elements.

20 pages, 10 figures, minor corrections for v2

References in corpus (6)

• The automated computation of tree-level and next-to-leading order differential cross sections, and their matching to parton shower simulations
• LHAPDF6: parton density access in the LHC precision era
• Optimising simulations for diphoton production at hadron colliders using amplitude neural networks
• Uncertainty Aware Learning for High Energy Physics
• Phase Space Sampling and Inference from Weighted Events with Autoregressive Flows
• Loop Amplitudes from Precision Networks