Reproducible analysis frameworks in Lattice Field Theory and STFC-enabled computational research in Wales
Lead Research Organisation:
Swansea University
Department Name: College of Science
Abstract
Lattice Field Theory is a technique used to study many models of physics, most notable Quantum ChromoDynamics, the theory of the strong interaction which holds the nuclei of atoms together. Calculations at the very smallest scales-many thousands of times smaller than the size of an atom-can make predictions for properties of matter that can be observed at experiments such as those at the Large Hadron Collider in Switzerland. Even though (and because) the length scales considered are so small, these calculations require vast amounts of computing power, using some of the largest supercomputers in the world. This produces large amounts of data, which must be carefully analysed by researchers to extract the properties being studied and make predictions. As the size of computers and the amount of data that can be produced grows, the likelihood of making errors in this process increases. To reduce the chance of this happening, this project will develop a set of tools to automate these analyses, so that they can always run consistently, and their results can be re-run and checked by anyone. The project will work with researchers to adapt their existing software to make use of these tools, and train them to do the same themselves for new software.
The Solar System Physics group at Aberystwyth University studies our Sun and its interactions with the rest of our Solar System. This includes studying phenomena like solar storms, which have the potential to significantly disrupt electronics and telecommunications on Earth; better understanding these phenomena allows us to predict when they may occur and take steps to protect critical infrastructure. These computations make use of programs written in a computer language called IDL, which requires paying money to use, with the amount you pay determining the number of processing units you can use at once, and so how fast your computations will be performed. While this was a good choice when the programs were written, the amount of data needing to be analysed has grown to the point that waiting for IDL to run an analysis is causing a bottleneck in research. This project will support researchers at Aberystwyth in adapting the programs to instead use the commonly-used Python programming language instead. This will let them use many more processing units simultaneously, giving answers more quickly. This will also mean that other researchers who haven't paid for IDL will be able to verify the results.
The Laser Interferometer Gravitational-Wave Observatory (LIGO) is an international collaboration to observe gravitational waves, which are ripples in spacetime itself, predicted by Einstein's theory of General Relativity. When very heavy objects in space, like black holes, collide with each other, their acceleration is strong enough to create gravitational waves that we can detect on Earth. These waves cause lengths and distances to change ever so slightly, which can be measured using devices called interferometers. In order to understand what kind of astronomical event created a gravitational wave, the readings from the interferometers must be compared with the results of simulations. At first, LIGO could only sense the largest of large events, the collisions of large black holes. Researchers are working to make LIGO more sensitive, to be able to detect collisions between smaller objects. This means that it observes a lot more gravitational waves than previously, so there is a lot more data to analyse. This project will work with the LIGO team at the Gravity Exploration Institute at Cardiff University to redevelop their software so that each analysis takes less time, which will allow LIGO to keep up with the increasing number of gravitational wave events that they observe.
The Solar System Physics group at Aberystwyth University studies our Sun and its interactions with the rest of our Solar System. This includes studying phenomena like solar storms, which have the potential to significantly disrupt electronics and telecommunications on Earth; better understanding these phenomena allows us to predict when they may occur and take steps to protect critical infrastructure. These computations make use of programs written in a computer language called IDL, which requires paying money to use, with the amount you pay determining the number of processing units you can use at once, and so how fast your computations will be performed. While this was a good choice when the programs were written, the amount of data needing to be analysed has grown to the point that waiting for IDL to run an analysis is causing a bottleneck in research. This project will support researchers at Aberystwyth in adapting the programs to instead use the commonly-used Python programming language instead. This will let them use many more processing units simultaneously, giving answers more quickly. This will also mean that other researchers who haven't paid for IDL will be able to verify the results.
The Laser Interferometer Gravitational-Wave Observatory (LIGO) is an international collaboration to observe gravitational waves, which are ripples in spacetime itself, predicted by Einstein's theory of General Relativity. When very heavy objects in space, like black holes, collide with each other, their acceleration is strong enough to create gravitational waves that we can detect on Earth. These waves cause lengths and distances to change ever so slightly, which can be measured using devices called interferometers. In order to understand what kind of astronomical event created a gravitational wave, the readings from the interferometers must be compared with the results of simulations. At first, LIGO could only sense the largest of large events, the collisions of large black holes. Researchers are working to make LIGO more sensitive, to be able to detect collisions between smaller objects. This means that it observes a lot more gravitational waves than previously, so there is a lot more data to analyse. This project will work with the LIGO team at the Gravity Exploration Institute at Cardiff University to redevelop their software so that each analysis takes less time, which will allow LIGO to keep up with the increasing number of gravitational wave events that they observe.
Publications
Athenodorou A
(2021)
Investigating the conformal behavior of SU(2) with one adjoint Dirac flavor
in Physical Review D
Athenodorou A
(2022)
Open Science in Lattice Gauge Theory community
Bennett E
(2023)
Update on SU(2) with one adjoint Dirac flavor
Bennett E
(2022)
Color dependence of the topological susceptibility in Yang-Mills theories
in Physics Letters B
Bennett E
(2023)
Status of reproducibility and open science in hep-lat in 2021
Bennett E
(2022)
Lattice studies of the S p ( 4 ) gauge theory with two fundamental and three antisymmetric Dirac fermions
in Physical Review D
Bennett E
(2024)
Singlets in gauge theories with fundamental matter
in Physical Review D
Bennett E
(2022)
Status of reproducibility and open science in hep-lat in 2021
Bennett E
(2022)
S p ( 2 N ) Yang-Mills theories on the lattice: Scale setting and topology
in Physical Review D
Julian J Lenz
(2023)
Magnetic catalysis in the (2+1)-dimensional Gross-Neveu model
in [submitted]
Lee J
(2022)
Composite dynamics in Sp(2 N ) gauge theories
in EPJ Web of Conferences
Lenz J
(2023)
Magnetized ( 2 + 1 )-dimensional Gross-Neveu model at finite density
in Physical Review D
Lenz J
(2023)
Magnetic catalysis in the ( 2 + 1 )-dimensional Gross-Neveu model
in Physical Review D
Title | Let it (Wilson) flow |
Description | Visualisation of the topological charge density distribution of a field configuration of Minimal Walking Technicolor. Successive time-slices are also iterated using the Wilson flow, which removes ultraviolet noise, which is clearly visible on the left, with the actual instantons (orange) and anti-instantons (blue) becoming visible at longer flow times to the right. The configuration was generated, and the charge density computed, using the DiRAC 3 Extreme Scaling service in Edinburgh. Credit: Ed Bennett, Biagio Lucini. |
Type Of Art | Image |
Year Produced | 2022 |
Impact | DiRAC image competition prize |
URL | https://telescoper.wordpress.com/2022/11/09/dirac-research-image-competition-the-winning-entries/ |
Description | LaVA (Lattice Virtual Academy) |
Geographic Reach | Multiple continents/international |
Policy Influence Type | Influenced training of practitioners or researchers |
URL | https://indico.ectstar.eu/event/161/ |
Description | Training material on publishing of data analysis code |
Geographic Reach | Local/Municipal/Regional |
Policy Influence Type | Influenced training of practitioners or researchers |
URL | https://edbennett.github.io/publishing-analysis |
Title | Investigating the conformal behaviour of SU(2) with one adjoint Dirac flavor --- data release |
Description | The full data generated in preparation for the publication "Investigating the conformal behaviour of SU(2) with one adjoint Dirac flavor" drafted in March 2021; this dataset was tidied and documented as part of this project. |
Type Of Material | Database/Collection of data |
Year Produced | 2021 |
Provided To Others? | Yes |
Impact | The dataset enabled publication of the paper mentioned above. |
URL | https://zenodo.org/record/5139618 |
Title | Lattice studies of the Sp(4) gauge theory with two fundamental and three antisymmetric Dirac fermions-code release |
Description | This release contains the analysis code used to prepare the plots and tables included in Lattice studies of the Sp(4) gauge theory with two fundamental and three antisymmetric Dirac fermions. |
Type Of Material | Computer model/algorithm |
Year Produced | 2022 |
Provided To Others? | Yes |
Impact | This work underpinned the publication listed above. |
URL | https://zenodo.org/record/6637743 |
Title | Lattice studies of the Sp(4) gauge theory with two fundamental and three antisymmetric Dirac fermions-data release |
Description | This dataset contains the raw data and metadata used to prepare the publication Lattice studies of the Sp(4) gauge theory with two fundamental and three antisymmetric Dirac fermions. Included are: The raw log output from the configuration generation, correlation function calculation, and Dirac eigenvalue computation, as well as metadata describing the ensembles used for the mass spectrum calculation, in `raw_data.zip`. These include all numbers used in the publication (aside from fit parameters) in plaintext form. All numbers included in the above logs, restructured into HDF5 format for convenience, in `data.h5`. The fit parameters used to compute the spectrum, including the thermalisation length, and the plateau start and end points, in `fit_params.zip`. The data underlying tables 2-6 of the publication above, in CSV format. |
Type Of Material | Database/Collection of data |
Year Produced | 2022 |
Provided To Others? | Yes |
Impact | These data underpinned the corresponding narrative publication mentioned above. |
URL | https://zenodo.org/record/6637515 |
Title | Sp(2N) Yang-Mills theories on the lattice: scale setting and topology-analysis workflow |
Description | This repository contains the analysis code used to prepare the plots and tables included in Topological susceptibility in Yang-Mills theories and Sp(2N) Yang-Mills theories on the lattice: scale setting and topology. |
Type Of Material | Computer model/algorithm |
Year Produced | 2022 |
Provided To Others? | Yes |
Impact | This work underpinned the two publications listed above. |
URL | https://zenodo.org/record/7260975 |
Title | Sp(2N) Yang-Mills theories on the lattice: scale setting and topology-data release |
Description | This release contains all data and metadata used to prepare the publications Topological susceptibility in Yang-Mills theories and Sp(2N) Yang-Mills theories on the lattice: scale setting and topology. Included are: The raw log output from the Wilson flow computation, as well as metadata describing the ensembles used, in `raw_data.zip`. These include all numbers used in the publication (aside from fit parameters) in plaintext form. The archive contains a separate `README.md` describing the layout of the data. All numbers included in the above logs, restructured into HDF5 format for convenience, in `datapackage.h5`. The data presented in all tables in both papers, in CSV format, as described in more detail below. |
Type Of Material | Database/Collection of data |
Year Produced | 2022 |
Provided To Others? | Yes |
Impact | These data underpinned the narrative publication mentioned above. |
URL | https://zenodo.org/record/6678411 |
Title | Survey of reproducibility in hep-lat publications in 2021 |
Description | Results of a survey of submissions to the hep-lat arXiv category in 2021, assessing how well each conforms to certain standards of reproducibility and openness. |
Type Of Material | Database/Collection of data |
Year Produced | 2022 |
Provided To Others? | Yes |
Impact | Two invited talks, a conference talk, and feed forward to other aspects of the project's work. |
URL | https://zenodo.org/record/7373413 |
Title | Survey on lattice data analysis, presentation, and curation practices |
Description | This repository contains the results of a survey on software workflows and open science in lattice field theory conducted in 2022 by Andreas Athenodorou, Ed Bennett, Julian Lenz, and Elli Papadopolou. These data were collected using LimeSurvey, and were first presented in a talk at Lattice 2022 by Andreas Athenodorou. The analysis is based on Julian Lenz's LimeSurvey CSV parser. The survey results are included in survey-results-redacted.csv. The survey structure is included in survey-structure.lss. Further details of the structure of the data, setup, see the included README.md file. |
Type Of Material | Database/Collection of data |
Year Produced | 2022 |
Provided To Others? | Yes |
Impact | These data underpinned a conference talk and a conference poster, as well as significantly informing the further development of this project. |
URL | https://zenodo.org/record/6980070 |
Title | Optimisation of Dynesty software |
Description | Dynesty is a library underpinning the Parallel Bilby application used by the LIGO Scientific Collaboration. It was identified as being a performance bottleneck in the parallelised code, and so work has been undertaken to reduce the time spent in serial portions of this code. |
Type Of Technology | New/Improved Technique/Technology |
Year Produced | 2022 |
Open Source License? | Yes |
Impact | Significant reduction in execution time of hot serialised functions. |
Title | Optimisation of Grid code |
Description | Grid is a toolkit for performing computations in lattice field theory that can be run on most high-performance computing architectures. One aspect of this computation was optimised resulting in an up to 20% reduction in execution time. |
Type Of Technology | New/Improved Technique/Technology |
Year Produced | 2021 |
Open Source License? | Yes |
Impact | N/A |
URL | https://github.com/edbennett/Grid_epcc/tree/accelerate-setcheckerboard |
Description | DiRAC Federation Project workshops |
Form Of Engagement Activity | Participation in an activity, workshop or similar |
Part Of Official Scheme? | No |
Geographic Reach | National |
Primary Audience | Other audiences |
Results and Impact | Met with 20-30 experienced researchers in science enabled by high-performance computing to identify opportunities for collaboration on future UKRI-led science-specific HPC procurements. |
Year(s) Of Engagement Activity | 2022 |
Description | Invited talk at Maynooth University |
Form Of Engagement Activity | A talk or presentation |
Part Of Official Scheme? | No |
Geographic Reach | National |
Primary Audience | Postgraduate students |
Results and Impact | Invited talk on open science and reproducibility given at Maynooth University research week. Audience was primarily academics and postgraduate research students. Significant questions and discussion followed. |
Year(s) Of Engagement Activity | 2023 |
URL | http://edbennett.github.io/maynooth-talk-20221024/ |