We develop a diagrammatic approach to effective field theories (EFTs) corresponding to deep neural networks at initialization, which dramatically simplifies computations of finite-width corrections to neuron statistics. The structures of EFT calculations make it transparent that a single condition governs criticality of all connected correlators of neuron preactivations. Understanding of such EFTs may facilitate progress in both deep learning and field theory simulations.

United States

Ian Banta

Tianji Cai

Nathaniel Craig

Zhengkang Zhang

Structures of neural network effective theories

We revisit the effective field theory of the two Higgs doublet model at tree level. The introduction of a novel basis in the UV theory allows us to derive matching coefficients in the effective description that resum important contributions from the Higgs vacuum expectation value. The new basis typically provides a significantly better approximation of the full theory prediction than the traditional approach that utilizes the Higgs basis, particularly for alignment away from the decoupling limit.

CERN

Switzerland

Timothy Cohen

Xiaochuan Lu

United Kingdom

Dave Sutherland

Effective field theory of the two Higgs doublet model

We study the effect of BSM particles receiving most of their mass from their coupling to the Higgs boson (“Loryons”) on the electroweak phase transition. The existence of BSM Loryons would imply that electroweak symmetry must be non-linearly realized in the effective theory of the Standard Model. Since, by definition, Loryons have a significant coupling to the Higgs, they are expected to have a significant effect on the Higgs effective potential and thereby the electroweak phase transition. We show that the BSM Loryon parameter space viable under current experimental and theoretical constraints overlaps heavily with the parameter space in which a strongly first-order phase transition is predicted. The portion of the experimentally allowed parameter space which gives a strongly first-order phase transition is significantly larger for Loryons as compared to non-Loryons.

A strongly first-order electroweak phase transition from Loryons

We initiate the study of a new class of beyond the Standard Model states that we call “Loryons.” They have the defining characteristic of being non-decoupling, in the sense that their physical mass is dominated by a contribution from the vacuum expectation value of the Higgs boson. The stakes are high: the discovery of a Loryon would tell us that electroweak symmetry must be non-linearly realized in the effective field theory of the Standard Model. Loryons have their masses bounded from above by perturbative unitarity considerations and thus define a finite parameter space for exploration. After providing a complete catalog of Loryon representations under mild assumptions, we turn to examining the constraints on the parameter space from Higgs couplings measurements, precision electroweak tests, and direct collider searches. We show that most fermionic candidates are already ruled out (with some notable exceptions), while much of the scalar Loryon parameter space is still wide open for discovery.

Italy

Non-decoupling new particles

Models with singlet fields coupling to the Higgs can enable a strongly first-order electroweak phase transition, of interest for baryogenesis and gravity waves. We improve on previous attempts to self-consistently solve for the bubble wall properties—wall speed <math><msub><mi>v</mi><mi>w</mi></msub></math> and shape—in a highly predictive class of models with <math><msub><mi>Z</mi><mn>2</mn></msub></math>-symmetric singlet potentials. A new algorithm is implemented to determine <math><msub><mi>v</mi><mi>w</mi></msub></math> and the wall profiles throughout the singlet parameter space in the case of subsonic walls, focusing on models with strong enough phase transitions to satisfy the sphaleron washout constraint for electroweak baryogenesis. We find speeds as low as <math><msub><mi>v</mi><mi>w</mi></msub><mo>≅</mo><mn>0.1</mn></math> in our scan over parameter space, and the singlet must be relatively light to have a subsonic wall, <math><msub><mi>m</mi><mi>s</mi></msub><mo>≲</mo><mn>135</mn><mtext> </mtext><mtext> </mtext><mi>GeV</mi></math>.