Development of New Strategy and Tools for Simulations and Real Data Analysis in High Energy Physics
Lead Research Organisation:
University of Southampton
Department Name: Electronics and Computer Science
Abstract
The task of Decoding the Fundamental Theory of Nature is the central task for modern particle physics but it can only be solved by using advanced computing strategy and techniques. Many promising models have been suggested, however no generic approach for mapping from particle physics data to different theories has been proposed. This project will develop new ideas related to implementation of fast Monte-Carlo simulation and Symbolic Matrix element evaluation using parallel processing techniques. An important element of the project is an effective comparison between simulated and real data, by reinterpretation of existing analyses or prototyping of new analysis to discriminate between models. Furthermore, this interdisciplinary project will explore new computing approaches and large data techniques and their application to particle physics. This project will develop further the High Energy Physics Model Database - the novel project created at the University of Southampton which enables users to store models and run Monte Carlo simulations on the IRIDIS4 higher performance computing cluster. A new extension, PhenoData, was devised to enable researchers to share digitized plots and tables, in addition to cataloguing publicly available analyses code - allowing users to search by collider signatures.
Organisations
People |
ORCID iD |
| Markus Brede (Primary Supervisor) | |
| Daniel Locke (Student) |
Studentship Projects
| Project Reference | Relationship | Related To | Start | End | Student Name |
|---|---|---|---|---|---|
| EP/N509747/1 | 01/10/2016 | 30/09/2021 | |||
| 1808100 | Studentship | EP/N509747/1 | 01/10/2016 | 30/09/2020 | Daniel Locke |
| Description | The study of particle dark matter (DM) at colliders (the LHC and future colliders) has been a key focus of this research. A project is nearing completion to study the potential impact of the proposed International Linear Collider (ILC) on the characterisation of DM. If a signal consistent with DM (stable, no electric charge, with interactions consistent with astronomical data) is discovered at the LHC, then precision measurements would be required to deduce the properties of such a candidate. A lepton collider such as ILC could enable us to determine the mass of such a candidate by observation of kinematic cusps, and the spin may also be determined through angular observables. An additional project is also underway to catalogue and extend so-called minimal consistent models of DM to include additional mediators. These models are renormalisable theories of the dark sector, with a single electroweak multiplet (of any spin) in addition to an extra mediator. The development of tools to improve and assist in the comparison of beyond the standard model theories and their experimental constraints is an additional focus of this research. The High Energy Physics Model Database (HEPMDB) is a Southampton based project to enable the central storage of HEP model files in several common formats. It also allows users to run several of the most popular HEP tools asynchronously on the IRIDIS high-performance computing cluster via a web-based interface. The current development of this web-platform is focused on the so called "bottom-up" approach of phenomenology; whereby data informs higher scale theory directly. To facilitate this, the development of a database of experimental signatures is under way, alongside an intuitive user interface to enable the user to be presented with possible theories which display observed phenomena. This research also led to the development of PhenoData - an online database that allows the user to store digitized data from plots or tables used in a given paper, for which there is no public data available. There was a demand for such a service in the phenomenology community, stemming from the lack of availability of data associated with plots or tables on some papers. This resulted in multiple people digitizing the same plots. Although the initial intended usage is withing the HEP phenomenology community, PhenoData may have further application to other fields. PhenoData has a RESTful API, allowing the database to be easily interfaced with other projects. Also under development is PhenoAnalysis; a new data analysis framework which also enables streamlined interface of ROOT files and common python based Machine Learning (ML) tools. This framework allows users to quickly compute common collider observables or easily add their own to the catalogue. Users may then apply cuts and generate publication ready plots. There are also inbuilt statistical tools to calculate 95% confidence level exclusions for signal regions via toy Monte Carlo. Alternatively, users may generate plots for data-exploration, to elucidate the correlation of their chosen observables, enabling effective extraction of higher-level features. Towards more sophisticated analysis prototyping, users may generate pandas dataframes (a commonly used data format in the ML community) of such high-level features to be passed to commonly used tools such as TensorFlow or scikit-learn. |
| Exploitation Route | Public tools HEPMDB and PhenoData enable HEP researchers to store and document both models and data, key for data preservation and reproducibility in the phenomenology community. |
| Sectors | Digital/Communication/Information Technologies (including Software) |
| Title | PhenoData |
| Description | PhenoData - an online database that allows the user to store digitized data from plots or tables used in a given paper, for which there is no public data available. There was a demand for such a service in the phenomenology community, stemming from the lack of availability of data associated with plots or tables on some papers. This resulted in multiple people digitizing the same plots. Although the initial intended usage is withing the HEP phenomenology community, PhenoData may have further application to other fields. PhenoData has a RESTful API, allowing the database to be easily interfaced with other projects. |
| Type Of Material | Database/Collection of data |
| Year Produced | 2017 |
| Provided To Others? | Yes |
| Impact | Has been cited in several hep-ph papers. |
| URL | https://hepmdb.soton.ac.uk/phenodata |