Fernandez Casani, A., Garcia Montoro, C., Gonzalez de la Hoz, S., Salt, J., Sanchez, J., & Villaplana Perez, M. (2023). Big Data Analytics for the ATLAS EventIndex Project with Apache Spark. Comput. Math. Methods, 2023, 6900908–19pp.
Abstract: The ATLAS EventIndex was designed to provide a global event catalogue and limited event-level metadata for ATLAS experiment of the Large Hadron Collider (LHC) and their analysis groups and users during Run 2 (2015-2018) and has been running in production since. The LHC Run 3, started in 2022, has seen increased data-taking and simulation production rates, with which the current infrastructure would still cope but may be stretched to its limits by the end of Run 3. A new core storage service is being developed in HBase/Phoenix, and there is work in progress to provide at least the same functionality as the current one for increased data ingestion and search rates and with increasing volumes of stored data. In addition, new tools are being developed for solving the needed access cases within the new storage. This paper describes a new tool using Spark and implemented in Scala for accessing the big data quantities of the EventIndex project stored in HBase/Phoenix. With this tool, we can offer data discovery capabilities at different granularities, providing Spark Dataframes that can be used or refined within the same framework. Data analytic cases of the EventIndex project are implemented, like the search for duplicates of events from the same or different datasets. An algorithm and implementation for the calculation of overlap matrices of events across different datasets are presented. Our approach can be used by other higher-level tools and users, to ease access to the data in a performant and standard way using Spark abstractions. The provided tools decouple data access from the actual data schema, which makes it convenient to hide complexity and possible changes on the backed storage.
|
Ferrer-Sanchez, A., Martin-Guerrero, J., Ruiz de Austri, R., Torres-Forne, A., & Font, J. A. (2024). Gradient-annihilated PINNs for solving Riemann problems: Application to relativistic hydrodynamics. Comput. Meth. Appl. Mech. Eng., 424, 116906–18pp.
Abstract: We present a novel methodology based on Physics-Informed Neural Networks (PINNs) for solving systems of partial differential equations admitting discontinuous solutions. Our method, called Gradient-Annihilated PINNs (GA-PINNs), introduces a modified loss function that forces the model to partially ignore high-gradients in the physical variables, achieved by introducing a suitable weighting function. The method relies on a set of hyperparameters that control how gradients are treated in the physical loss. The performance of our methodology is demonstrated by solving Riemann problems in special relativistic hydrodynamics, extending earlier studies with PINNs in the context of the classical Euler equations. The solutions obtained with the GA-PINN model correctly describe the propagation speeds of discontinuities and sharply capture the associated jumps. We use the relative l(2) error to compare our results with the exact solution of special relativistic Riemann problems, used as the reference ''ground truth'', and with the corresponding error obtained with a second-order, central, shock-capturing scheme. In all problems investigated, the accuracy reached by the GA-PINN model is comparable to that obtained with a shock-capturing scheme, achieving a performance superior to that of the baseline PINN algorithm in general. An additional benefit worth stressing is that our PINN-based approach sidesteps the costly recovery of the primitive variables from the state vector of conserved variables, a well-known drawback of grid-based solutions of the relativistic hydrodynamics equations. Due to its inherent generality and its ability to handle steep gradients, the GA-PINN methodology discussed in this paper could be a valuable tool to model relativistic flows in astrophysics and particle physics, characterized by the prevalence of discontinuous solutions.
|
KM3NeT Collaboration(Aiello, S. et al), Alves Garre, S., Calvo, D., Carretero, V., Colomer, M., Corredoira, I., et al. (2020). gSeaGen: The KM3NeT GENIE-based code for neutrino telescopes. Comput. Phys. Commun., 256, 107477–15pp.
Abstract: The gSeaGen code is a GENIE-based application developed to efficiently generate high statistics samples of events, induced by neutrino interactions, detectable in a neutrino telescope. The gSeaGen code is able to generate events induced by all neutrino flavours, considering topological differences between tracktype and shower-like events. Neutrino interactions are simulated taking into account the density and the composition of the media surrounding the detector. The main features of gSeaGen are presented together with some examples of its application within the KM3NeT project. Program summary Program Title: gSeaGen CPC Library link to program files: http://dx.doi.org/10.17632/ymgxvy2br4.1 Licensing provisions: GPLv3 Programming language: C++ External routines/libraries: GENIE [1] and its external dependencies. Linkable to MUSIC [2] and PROPOSAL [3]. Nature of problem: Development of a code to generate detectable events in neutrino telescopes, using modern and maintained neutrino interaction simulation libraries which include the state-of-the-art physics models. The default application is the simulation of neutrino interactions within KM3NeT [4]. Solution method: Neutrino interactions are simulated using GENIE, a modern framework for Monte Carlo event generators. The GENIE framework, used by nearly all modern neutrino experiments, is considered as a reference code within the neutrino community. Additional comments including restrictions and unusual features: The code was tested with GENIE version 2.12.10 and it is linkable with release series 3. Presently valid up to 5 TeV. This limitation is not intrinsic to the code but due to the present GENIE valid energy range. References: [1] C. Andreopoulos at al., Nucl. Instrum. Meth. A614 (2010) 87. [2] P. Antonioli et al., Astropart. Phys. 7 (1997) 357. [3] J. H. Koehne et al., Comput. Phys. Commun. 184 (2013) 2070. [4] S. Adrian-Martinez et al., J. Phys. G: Nucl. Part. Phys. 43 (2016) 084001.
|
KM3NeT Collaboration(Aiello, S. et al), Calvo, D., Coleiro, A., Colomer, M., Gozzini, S. R., Hernandez-Rey, J. J., et al. (2020). The Control Unit of the KM3NeT Data Acquisition System. Comput. Phys. Commun., 256, 107433–16pp.
Abstract: The KM3NeT Collaboration runs a multi-site neutrino observatory in the Mediterranean Sea. Water Cherenkov particle detectors, deep in the sea and far off the coasts of France and Italy, are already taking data while incremental construction progresses. Data Acquisition Control software is operating off-shore detectors as well as testing and qualification stations for their components. The software, named Control Unit, is highly modular. It can undergo upgrades and reconfiguration with the acquisition running. Interplay with the central database of the Collaboration is obtained in a way that allows for data taking even if Internet links fail. In order to simplify the management of computing resources in the long term, and to cope with possible hardware failures of one or more computers, the KM3NeT Control Unit software features a custom dynamic resource provisioning and failover technology, which is especially important for ensuring continuity in case of rare transient events in multi-messenger astronomy. The software architecture relies on ubiquitous tools and broadly adopted technologies and has been successfully tested on several operating systems.
|
Aiola, S., Amhis, Y., Billoir, P., Jashal, B. K., Henry, L., Oyanguren, A., et al. (2021). Hybrid seeding: A standalone track reconstruction algorithm for scintillating fibre tracker at LHCb. Comput. Phys. Commun., 260, 107713–5pp.
Abstract: We describe the Hybrid seeding, a stand-alone pattern recognition algorithm aiming at finding charged particle trajectories for the LHCb upgrade. A significant improvement to the charged particle reconstruction efficiency is accomplished by exploiting the knowledge of the LHCb magnetic field and the position of energy deposits in the scintillating fibre tracker detector. Moreover, we achieve a low fake rate and a small contribution to the overall timing budget of the LHCb real-time data processing.
|