Related papers: Structures of Neural Network Effective Theories
Effective field theories (EFTs) provide a powerful framework to parametrise unknown aspects of possible ultraviolet (UV) physics. For scalar fields in de Sitter space, however, new emergent phenomena can arise when the cut-off scale of the…
The 1D flux power spectrum ($P_{\mathrm{1D}}$) of the Ly$\alpha$ forest provides an exceptionally high-resolution probe of structure formation down to small scales ($k\approx1-10~\text{$h~$Mpc$^{-1}$}$). These scales carry the imprints of…
Given a theory containing both heavy and light fields (the UV theory), a standard procedure is to integrate out the heavy field to obtain an effective field theory (EFT) for the light fields. Typically the EFT equations of motion consist of…
We study nonlinear effective field theories (EFTs) with factorially growing perturbative expansions, focusing on a class in which the relative entropy encodes an infinite tower of higher-dimensional operators. Using the resummed relative…
These notes are an overview of effective field theory (EFT) methods. I discuss toy model EFTs, chiral perturbation theory, Fermi liquid theory, and non-relativistic QED, and use these examples to introduce a variety of EFT concepts,…
In this dissertation, I introduce the principles and methods of effective field theory and describe my work in three EFTs: First, in the perturbative QCD region, I use soft collinear effective theory (SCET) to prove that strong interaction…
The statistical properties of deep neural networks (DNNs) at initialization play an important role to comprehend their trainability and the intrinsic architectural biases they possess before data exposure Well established mean field (MF)…
We introduce an effective field theory (EFT) for conformal impurity by considering a pair of transversely displaced impurities and integrating out modes with mass inversely proportional to the separation distance. This EFT captures the…
We develop interactions from chiral effective field theory (EFT) that are tailored to the harmonic oscillator basis. As a consequence, ultraviolet convergence with respect to the model space is implemented by construction and infrared…
Network science can offer fundamental insights into the structural and functional properties of complex systems. For example, it is widely known that neuronal circuits tend to organize into basic functional topological modules, called…
We apply the effective field theoretic (EFT) approach to resum the large perturbative logarithms arising when partonic hard scattering cross-sections are taken to the threshold limit. We consider deep inelastic scattering, Drell-Yan lepton…
Effective Field Theory (EFT) is an efficient method for parametrizing unknown high energy physics effects on low energy data. When applied to time-dependent backgrounds, EFT must be supplemented with initial conditions. In these…
Neural networks enjoy widespread success in both research and industry and, with the imminent advent of quantum technology, it is now a crucial challenge to design quantum neural networks for fully quantum learning tasks. Here we propose…
Many studies of possible new physics employ effective field theory (EFT), whereby corrections to the Standard Model take the form of higher-dimensional operators, suppressed by a large energy scale. Fits of such a theory to data typically…
Debate topic for Effective Field Theory (EFT) is the choice of a "basis" for $\mrdim = 6$ operators Clearly all bases are equivalent as long as they are a "basis", containing a minimal set of operators after the use of equations of motion…
We review the basic ideas of the dynamical mean field theory (DMFT) and some of the insights into the electronic structure of strongly correlated electrons obtained by this method in the context of model Hamiltonians. We then discuss the…
Attention networks have proven to be an effective approach for embedding categorical inference within a deep neural network. However, for many tasks we may want to model richer structural dependencies without abandoning end-to-end training.…
Neural network models generally involve two important components, i.e., network architecture and neuron model. Although there are abundant studies about network architectures, only a few neuron models have been developed, such as the MP…
Both the path integral measure in field theory and ensembles of neural networks describe distributions over functions. When the central limit theorem can be applied in the infinite-width (infinite-$N$) limit, the ensemble of networks…
Large Language Models (LLMs) are composed of neurons that exhibit various behaviors and roles, which become increasingly diversified as models scale. Recent studies have revealed that not all neurons are active across different datasets,…