A Virtual Tribute to Quark Confinement and the Hadron Spectrum 2021

Aug 2 – 6, 2021
online
Europe/Brussels timezone

Machine learning regression and error quantification for lattice QCD Monte Carlo observables

Aug 4, 2021, 5:00 PM
20m
online

online

Parallel contribution H. Statistical Methods for Physics Analysis in the XXI Century

Speaker

Boram Yoon (Los Alamos National Laboratory)

Description

In lattice QCD simulations, a large number of observables are calculated on each Monte Carlo sample of gauge fields, and their statistical fluctuations are correlated with each other as they share the same background gauge field. By exploiting the correlation, a machine learning regression model can be trained to predict the values of the computationally expensive observables from the values of the computationally cheap observables for each Monte Carlo sample of the gauge field. I will present the machine learning algorithm and its applications to the prediction of lattice QCD observables and discuss the bias correction and error quantification procedure of the machine learning predictions on statistical data.

Primary author

Boram Yoon (Los Alamos National Laboratory)

Co-authors

Dr Tanmoy Bhattacharya (Los Alamos National Laboratory) Dr Rajan Gupta (Los Alamos National Laboratory)

Presentation materials

 ConfXIV-Yoon.mp4 ConfXIV-Yoon.pdf