Summary
Lattice Quantum Chromodynamics (LQCD) is the only known systematic framework to obtain ab-initio results in the non-perturbative regime of strong interactions. Its relevance to high-energy and nuclear physics has grown significantly in recent years due in part to a series of algorithmic advancements.
This project aims to compute time-like observables using numerical simulations of LQCD. Specifically, I will study spectral functions including the R-ratio, that is linked to the hadronic vacuum polarization of the electromagnetic current, and the hadronic tensor, that contains information on deep-inelastic scattering.
It is extremely challenging to compute observables intrinsically defined in Minkowski spacetime with lattice techniques, with the main issue being that the simulated quantum field theory is defined in Euclidean spacetime. While Euclidean correlators contain all the information needed to extract real-time physics, performing the analytic continuation with finite-precision data points from numerical simulations is an ill-posed problem. A second issue is that the the computational cost is driven by the loss of the signal of hadronic correlators with Euclidean-time separation, that happens at an exponential rate.
I will address these issues and significantly reduce the computational effort needed thanks to algorithms advancements. I plan to solve the signal-to-noise ratio problem using and further developing multi-level Monte Carlo sampling methods, that I recently contributed to extend to theories with fermions. The resulting exponential gain in the quality of the signal is essential to be able to perform the analytic continuation, that I plan to control using state-of-the-art techniques based on the Backus-Gilbert algorithm that have recently been developed by the supervisor.
This project aims to compute time-like observables using numerical simulations of LQCD. Specifically, I will study spectral functions including the R-ratio, that is linked to the hadronic vacuum polarization of the electromagnetic current, and the hadronic tensor, that contains information on deep-inelastic scattering.
It is extremely challenging to compute observables intrinsically defined in Minkowski spacetime with lattice techniques, with the main issue being that the simulated quantum field theory is defined in Euclidean spacetime. While Euclidean correlators contain all the information needed to extract real-time physics, performing the analytic continuation with finite-precision data points from numerical simulations is an ill-posed problem. A second issue is that the the computational cost is driven by the loss of the signal of hadronic correlators with Euclidean-time separation, that happens at an exponential rate.
I will address these issues and significantly reduce the computational effort needed thanks to algorithms advancements. I plan to solve the signal-to-noise ratio problem using and further developing multi-level Monte Carlo sampling methods, that I recently contributed to extend to theories with fermions. The resulting exponential gain in the quality of the signal is essential to be able to perform the analytic continuation, that I plan to control using state-of-the-art techniques based on the Backus-Gilbert algorithm that have recently been developed by the supervisor.
Unfold all
/
Fold all
More information & hyperlinks
Web resources: | https://cordis.europa.eu/project/id/843134 |
Start date: | 01-11-2019 |
End date: | 31-10-2021 |
Total budget - Public funding: | 191 149,44 Euro - 191 149,00 Euro |
Cordis data
Original description
Lattice Quantum Chromodynamics (LQCD) is the only known systematic framework to obtain ab-initio results in the non-perturbative regime of strong interactions. Its relevance to high-energy and nuclear physics has grown significantly in recent years due in part to a series of algorithmic advancements.This project aims to compute time-like observables using numerical simulations of LQCD. Specifically, I will study spectral functions including the R-ratio, that is linked to the hadronic vacuum polarization of the electromagnetic current, and the hadronic tensor, that contains information on deep-inelastic scattering.
It is extremely challenging to compute observables intrinsically defined in Minkowski spacetime with lattice techniques, with the main issue being that the simulated quantum field theory is defined in Euclidean spacetime. While Euclidean correlators contain all the information needed to extract real-time physics, performing the analytic continuation with finite-precision data points from numerical simulations is an ill-posed problem. A second issue is that the the computational cost is driven by the loss of the signal of hadronic correlators with Euclidean-time separation, that happens at an exponential rate.
I will address these issues and significantly reduce the computational effort needed thanks to algorithms advancements. I plan to solve the signal-to-noise ratio problem using and further developing multi-level Monte Carlo sampling methods, that I recently contributed to extend to theories with fermions. The resulting exponential gain in the quality of the signal is essential to be able to perform the analytic continuation, that I plan to control using state-of-the-art techniques based on the Backus-Gilbert algorithm that have recently been developed by the supervisor.
Status
CLOSEDCall topic
MSCA-IF-2018Update Date
28-04-2024
Images
No images available.
Geographical location(s)