Towards a new generation of parton densities with deep learning models
read the original abstract
We present a new regression model for the determination of parton distribution functions (PDF) using techniques inspired from deep learning projects. In the context of the NNPDF methodology, we implement a new efficient computing framework based on graph generated models for PDF parametrization and gradient descent optimization. The best model configuration is derived from a robust cross-validation mechanism through a hyperparametrization tune procedure. We show that results provided by this new framework outperforms the current state-of-the-art PDF fitting methodology in terms of best model selection and computational resources usage.
This paper has not been read by Pith yet.
Forward citations
Cited by 5 Pith papers
-
Extraction of the color dipole amplitude with physics-informed neural networks
Physics-informed neural networks extract a model-independent color dipole amplitude from inclusive HERA data that predicts exclusive J/ψ photoproduction cross-sections without parameter retuning.
-
A linear PDF model for Bayesian inference
Presents a linear PDF parametrization from dimensionality-reduced neural network bases for efficient Bayesian inference, tested via multi-closure tests on synthetic deep inelastic scattering data.
-
Hyperoptimisation algorithm for the next generation of PDF determinations: ensemble regression with an unbiased selection model
Presents a new ensemble regression hyperoptimisation procedure for PDF fitting that combines multiple statistically equivalent methodologies into one PDF set accounting for hyperparameter variation.
-
Scheme-invariant stratified factorization algebras for inclusive deep inelastic scattering
Proposes a scheme-invariant stratified factorization algebra framework that derives the DIS convolution formula independently of collinear scheme or operator basis choices.
-
Neural Network Representation of Generalized Parton Distributions (NNGPD)
A neural network trained solely on integral observables from a known GPD model recovers the main features of the underlying distributions in a closure test.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.