Transforming the Bootstrap: Using Transformers to Compute Scattering Amplitudes in Planar N = 4 Super Yang-Mills Theory

Abstract

We pursue the use of deep learning methods to improve state-of-the-art computations in theoretical high-energy physics. Planar N = 4 Super Yang-Mills theory is a close cousin to the theory that describes Higgs boson production at the Large Hadron Collider; its scattering amplitudes are large mathematical expressions containing integer coefficients. In this paper, we apply Transformers to predict these coefficients. The problem can be formulated in a language-like representation amenable to standard cross-entropy training objectives. We design two related experiments and show that the model achieves high accuracy (> 98%) on both tasks. Our work shows that Transformers can be applied successfully to problems in theoretical physics that require exact solutions.

26+10 pages, 9 figures, 7 tables, application of machine learning aimed at physics and machine learning audience; v2: clarifications added, matches published version

References in corpus (34)

• Adam: A Method for Stochastic Optimization
• Neural Machine Translation by Jointly Learning to Align and Translate
• LLaMA: Open and Efficient Foundation Language Models
• Higgs boson gluon-fusion production in N3LO QCD
• Classical Polylogarithms for Amplitudes and Wilson Loops
• Multiple polylogarithms and mixed Tate motives
• Space-time S-matrix and Flux-tube S-matrix at Finite Coupling
• Bootstrapping the three-loop hexagon
• Hopf algebras, coproducts and symbols: an application to Higgs boson amplitudes
• Bootstrapping a Five-Loop Amplitude Using Steinmann Relations
• Higgs Boson Production at Hadron Colliders at N3LO in QCD
• Vector-boson fusion Higgs production at N${}^3$LO in QCD
• Two-Loop QCD Corrections to the Helicity Amplitudes for $H\to$ 3 partons
• The Drell-Yan cross section to third order in the strong coupling constant
• Collider Physics at the Precision Frontier
• Six-Gluon Amplitudes in Planar $\mathcal{N}=4$ Super-Yang-Mills Theory at Six and Seven Loops
• Analytic two-loop form factors in N=4 SYM
• Multiple Elliptic Polylogarithms
• Charged Current Drell-Yan Production at N3LO
• Higgs production in bottom-quark fusion to third order in the strong coupling
• Branes with Brains: Exploring String Vacua with Deep Reinforcement Learning
• Lepton-pair production at hadron colliders at N${}^3$LO in QCD
• A Three-Point Form Factor Through Five Loops
• Les Houches 2021: Physics at TeV Colliders: Report on the Standard Model Precision Wishlist
• An Operator Product Expansion for Form Factors
• Functions Beyond Multiple Polylogarithms for Precision Collider Physics
• The Impact of Positional Encoding on Length Generalization in Transformers
• Investigating the Limitations of Transformers with Simple Arithmetic Tasks
• Deep Symbolic Regression for Recurrent Sequences
• SYMBA: Symbolic Computation of Squared Amplitudes in High Energy Physics with Machine Learning
• Linear algebra with transformers
• SALSA: Attacking Lattice Cryptography with Transformers
• Teaching Arithmetic to Small Transformers
• xVal: A Continuous Numerical Tokenization for Scientific Language Models

Cited by in corpus (2)

• Learning the Simplicity of Scattering Amplitudes
• Meta-Designing Quantum Experiments with Language Models