WLCG/HSF Workshop 2024

Europe/Zurich
DESY

DESY

Deutsches Elektronen-Synchrotron (DESY) Notkestraße 85 22607 Hamburg Germany
Description

Welcome to the WLCG/HSF Workshop at DESY Hamburg, May 13-17 2024

Registration is now OPEN and is done via the DESY indico.

Registration will close at midnight Friday 26th of April

    • 2:00 PM 3:30 PM
      Plenary: Opening Hoersaal

      Hoersaal

      DESY

      Main workshop plenary session

      • 2:00 PM
        Introduction: WLCG and DESY 25m
      • 2:35 PM
        WLCG Governance 20m
      • 3:00 PM
        HSF retrospective and future 20m
    • 3:30 PM 4:00 PM
      Coffee 30m
    • 4:00 PM 5:30 PM
      Plenary: Opening Hoersaal

      Hoersaal

      DESY

      Main workshop plenary session

      • 4:00 PM
        DE consolidation 20m
        Speaker: Sebastian Wozniewski (Georg August Universitaet Goettingen (DE))
      • 4:30 PM
        Technology outlook 1h
        Speaker: Bernd Panzer-Steindel (CERN)
    • 6:00 PM 8:00 PM
      Reception 2h Canteen Extension

      Canteen Extension

      DESY

    • 9:00 AM 10:30 AM
      HSF: Community Software and Software Projects Seminar Room 4

      Seminar Room 4

      DESY

      • 9:00 AM
        The ROOT Project: Status and Future Directions 25m

        In this contribution, we’ll review the current status of the ROOT project, characterising its structure, available effort and strategic goals. We’ll explain how in the recent years the energy flowing from the open source community changed ROOT and boosted the development, materialising in the form of code, reports, ideas and proposals. We’ll review the recently integrated features that are key for the remainder of Run 3’s analysis and data processing as well as the HL-LHC era. We’ll not only focus on I/O, statistical tools and analysis, but also on less advertised ROOT components, such as packaging, distribution, Python support and graphics, and how those interoperate with other tools in the ecosystem. Moreover, we’ll discuss how the project will evolve in the next years, continuing to be at the heart of CERN’s flagship activity, the LHC, and prepare to support forthcoming and future experiments.

        Speaker: Danilo Piparo (CERN)
      • 9:25 AM
        The Scikit-HEP project - overview and future 25m

        Scikit-HEP is a community-driven and community-oriented project with the goal of providing an ecosystem for particle physics data analysis in Python fully integrated with the wider scientific Python ecosystem. The project started in Autumn 2016 and has evolved into a toolset of approximately thirty packages and a few “affiliated” packages.
        It expands the typical Python data analysis tools for particle physicists,with packages spanning the spectrum from general scientific libraries for data manipulation to domain-specific libraries.Each package focuses on a particular topic, and interacts with other packages in the toolset, where appropriate. Interoperability between Particle Physics tools and the Python scientific ecosystem is an important aspect of the project. Most of the packages are easy to install in many environments; much work has been done to provide binary wheels on PyPI and conda-forge packages. The project has gained interest and momentum over the years, carefully building a user and developer community engaging collaboration across experiments. Some of the packages are being used by other projects and communities. Utilities started within Scikit-HEP have in the meantime made its way as contributions to the wide Scientific Python project - the development guide and repository reviewer.

        An overview of the overall project and toolset will be presented, with comments on its history and evolution. Areas of particular relevance to community software, impact and engagement will be stressed. Future developments and matters of sustainability will be discussed.

        Speaker: Eduardo Rodrigues (University of Liverpool (GB))
      • 9:50 AM
        Key4hep - The common software stack for future experiments 35m

        Providing and maintaining the necessary tools for studying and developing detectors for future colliders is non trivial. On the one hand it requires a substantially sized software stack with all complications arising therefrom. On the other hand the available person power is usually strongly limited. In order to tackle both the Key4hep project aims at providing a complete software stack that can be used by all future collider communities, e.g. FCC, ILC, CEPC, EIC and MuonCollider among others.

        In this presentation we give an overview and status update of the Key4hep project itself but will also dive deeper into some aspects. These include building and maintaining the stack with the spack package manager, key insights and experiences we gained while developing for different communities simultaneously and also how to connect different existing software tools into a coherent framework. Additionally, we will also talk about some of the currently ongoing developments and future plans.

        Speaker: Thomas Madlener (Deutsches Elektronen-Synchrotron (DESY))
    • 9:00 AM 10:30 AM
      WLCG: Data Challenge Hoersaal

      Hoersaal

      DESY

      • 9:00 AM
        LHC Experiments 30m
        Speaker: Katy Ellis (Science and Technology Facilities Council STFC (GB))
      • 9:30 AM
        Belle II 15m
        Speaker: Silvio Pardi (Universita Federico II e INFN Sezione di Napoli (IT))
      • 9:45 AM
        DUNE 15m
        Speaker: Doug Benjamin (Brookhaven National Laboratory (US))
      • 10:00 AM
        Site Perspective 30m
        Speakers: Andreas Petzold (KIT - Karlsruhe Institute of Technology (DE)), Ofer Rind (Brookhaven National Laboratory)
    • 10:30 AM 11:00 AM
      Coffee 30m
    • 11:00 AM 12:30 PM
      HSF: Community Software and Software Projects Seminar Room 4

      Seminar Room 4

      DESY

      • 11:00 AM
        The HSF-India Project 20m

        The HSF-India initiative, which aims to implement new and impactful research software collaborations between India, Europe and the United States. The intent of this project is to increase the engagement of software experts in Asia with the HSF community. The starting point of this collaboration is a series of software workshops focused on building software skills. These workshops are the basis of a mutual training network that enables early-career researchers to pursue impactful research software initiatives in ways that advance their careers in experimental data-intensive science. Other project components include student projects and bidirectional researcher exchange programs. The experimental scope of this project is relatively broad, aiming to bring together researchers across facilities with common problems in research spanning experimental high-energy physics, nuclear physics and particle astrophysics. This talk will describe the scope of this initiative, its mechanisms for fostering new collaborations, and ways for interested research groups to get involved.

        Speaker: David Lange (Princeton University (US))
      • 11:20 AM
        User analysis software in a large collaboration (xAODAnaHelpers): development, maintenance and training. 20m

        A number of analyses and performance groups in ATLAS use an analysis framework, written in C++ with python steering files, called xAODAnaHelpers (xAH). xAH is used to loop on events a variety of ATLAS analysis data formats, by using central software to calibrate, select and correct physics objects. xAH has been chosen as one of the EVERSE (European Virtual Institute for Research Software Excellence) pilot cases representing user analysis software in particle physics, given (a) its widespread use in a large collaboration (b) the fact that its modular and intuitive interface fits the needs of diverse analysis use cases that require custom calibrations and objects beyond traditional physics analyses and (c) the challenges that end-user analysis software faces when relying on centrally developed tools that are updated often but still need to have full backward compatibility for ongoing analyses.
        After a brief description of the framework itself, this contribution will focus on the software development and maintenance practices for such a framework and on the development of tutorials for newcomers. It will also discuss plans for future work on software sustainability.

        Speaker: Tobias Fitschen (University of Manchester (GB))
      • 11:40 AM
        Pythia8 status and future developments 25m

        I will describe the current status of the Pyhtia8 project and some future developments that we are working with. I will also describe services offered by the Pythia8 collaboration such as on-line tutorials, and our GitLab help desk.

        Speaker: Leif Lönnblad (Lund University (SE))
      • 12:05 PM
        The Phoenix Event Display 20m

        Phoenix is a TypeScript-based event display framework, created in response to the 2017 HSF community white paper.

        It uses industry standard web tools (such as the popular three.js library for 3D rendering), and runs entirely in the client's web browser. It is experiment agnostic by design, providing shared common functionality (such as custom menus, controls, propagators) but also has support for experiment specific extensions for geometry and event data. It consists of two packages: a plain TypeScript core library (phoenix-event-display) and an Angular application for the UI (a React example is also provided in the documentation). Phoenix has been selected as a Google Summer of Code project for several years, and its contributors come from a wide variety of backgrounds. Recent developments have focused on improving event navigation and comprehension, with tools to better understand the relative position of objects, as well as native support for common formats such as EDM4HEP.

        It is currently used by several experiments, including ATLAS, FCC, LHCb and Belle-II.

        Speaker: Edward Moyse (University of Massachusetts (US))
    • 11:00 AM 12:30 PM
      WLCG: Data Challenge Hoersaal

      Hoersaal

      DESY

      • 11:00 AM
        FTS Perspective 20m
        Speakers: Mihai Patrascoiu (CERN), Steven Murray (CERN)
      • 11:20 AM
        Rucio Perspective 20m
        Speakers: Dimitrios Christidis (CERN), Martin Barisits (CERN)
      • 11:40 AM
        IAM Perspective 15m
        Speaker: Berk Balci (CERN)
      • 11:55 AM
        Network Technologies 25m
    • 12:45 PM 2:15 PM
      Lunch 1h 30m
    • 2:15 PM 3:30 PM
      Plenary: HSF Strategy Hoersaal

      Hoersaal

      DESY

      Main workshop plenary session

    • 3:30 PM 4:00 PM
      Coffee 30m
    • 4:00 PM 5:30 PM
      HSF: Software Tools, Generators and File Formats Seminar Room 4

      Seminar Room 4

      DESY

      • 4:00 PM
        NuHepMC: A standardized event record format for neutrino event generators 20m

        Simulations of neutrino interactions are playing an increasingly important role in the pursuit of high-priority measurements for the field of particle physics. A significant technical barrier for efficient development of these simulations is the lack of a standard data format for representing individual neutrino scattering events. We propose and define such a universal format, named NuHepMC, as a common standard for the output of neutrino event generators. The NuHepMC format uses data structures and concepts from the HepMC3 event record library adopted by other subfields of high-energy physics. These are supplemented with an original set of conventions for generically representing neutrino interaction physics within the HepMC3 infrastructure.

        Speaker: Dr Luke Pickering (Royal Holloway, University of London)
      • 4:20 PM
        Recent developments in the HepMC event record 20m

        HepMC3 is a library developed to handle the simulated collision events from Monte Carlo event Generators in High Energy Physics. The library is a successor in spirit of the earlier HepMC library and incorporated multiple ideas which appeared in the recent decade in the HEP community.

        This contribution discusses in detail the recent developments of the HepMC3 project, the relation of HepMC3 to other projects in the community, and the perspective of future developments.

        Speaker: Andrii Verbytskyi (Max Planck Society (DE))
      • 4:40 PM
        The HSF Conditions Database 20m

        Conditions data is the subset of non-event data that is necessary to process event data. It poses a unique set of challenges, namely a heterogeneous structure and high access rates by distributed computing. As these challenges are similar across various High Energy Physics (HEP) and Nuclear Physics (NP) experiments, the HEP Software Foundation (HSF) hosted a forum to discuss and share experiences from different collaborations. This yielded a white paper on 'best practice' for conditions data access, and a corresponding chapter of the HSF Community White Paper. Based on this experience, the potential for an experiment-agnostic conditions database was evident. An HSF activity was created to publish a white paper on conditions data use cases and requirements, to provide the basis for a conditions database designed as Community Software.

        This presentation will discuss the reference implementation, an 'HSF project' that satisfies those use cases and requirements. The reference implementation was developed in collaboration with sPHENIX, serving as the first real world application. This direct feedback provided a clearer understanding of the requirements, and additional implementation recommendations, while using the experts in the HSF activity to maintain its experiment-agnostic nature. In addition to sPHENIX, Belle II has also expressed interest in adopting the HSF reference implementation to benefit from its demonstrated scalability and performance.

        Speaker: Lino Oscar Gerlach (Brookhaven National Laboratory (US))
      • 5:00 PM
        The Gaussino core simulation software 20m

        Gaussino is an experiment-independent simulation package built upon the Gaudi software framework. It provides generic core components and interfaces for a complete HEP simulation application: event generation, detector simulation, geometry, monitoring and output of the simulated data. The generator interface allows for a wide variety of external event generator packages to be used, with an example implementation included for Pythia8. Detector simulation relies on the Geant4 toolkit for particle transport. It also provides a fast simulation interface to offload the simulation of specific sub-detectors to external processes, including GPU-accelerated and machine-learning-based options. Geometry descriptions can be provided through DD4Hep, GDML, experiment-specific software, or simple volumes specified at configuration time. Visualisation of the geometry and simulated data can be performed using the Geant4 visualisation driver or by saving the necessary objects for visualisation with Phoenix. Gaussino ensures a consistent multi-threaded execution between the various components and the underlying Gaudi infrastructure. This talk will focus on the features of Gaussino as a generic standalone application, giving examples for a diverse range of HEP experiments. Finally, the use of Gaussino as a toolkit to build experiment-specific applications will be covered, with LHCb's Gauss as an example.

        Speaker: Adam Morris (CERN)
    • 4:00 PM 5:30 PM
      WLCG: Operations Hoersaal

      Hoersaal

      DESY

      • 4:00 PM
        Overview of the ongoing changes and OPS campaigns on the WLCG infrastructure 20m
        Speaker: Julia Andreeva (CERN)
      • 4:20 PM
        Transition from VOMS Admin to IAM. Token profile definition. Implications on the middleware. 20m
        Speaker: Maarten Litmaath (CERN)
      • 4:40 PM
        APEL status and plans 20m
        Speaker: Adrian Coveney (STFC UKRI)
      • 5:00 PM
        AUDITOR 15m

        The increasing computational demand in High Energy Physics as well as increasing concerns about energy efficiency in high performance/throughput computing are driving forces in the search for more efficient ways to utilize available resources. Since avoiding idle resources is key in achieving high efficiency, an appropriate measure is sharing of idle resources of under-utilized sites with fully occupied sites. The software COBalD/TARDIS can automatically, transparently and dynamically (dis)integrate such resources in an opportunistic manner.

        However, resource sharing also requires accounting. This is done with AUDITOR (AccoUnting DatahandlIng Toolbox for Opportunistic Resources), a flexible and extensible accounting ecosystem that can cover a wide range of use cases and infrastructures. Accounting data is gathered via so-called collectors and stored in a database. So-called plugins can access the data and can act based on the accounting information.

        An HTCondor collector, a Slurm collector and a TARDIS collector are currently available, and a Kubernetes collector is already being worked on.

        The APEL plugin, for example, enables the creation of APEL accounting summaries and their transmission to the APEL accounting server. While the original goal of developing AUDITOR was to enable accounting for opportunistic resources managed by COBalD/TARDIS, it can also be used for normal accounting of a WLCG computing resource. Because AUDITOR uses a highly flexible data structure to store accounting data, extensions such as accounting GPU resources can be added with minimal effort.

        Speaker: Michael Boehler (Albert Ludwigs Universitaet Freiburg (DE))
      • 5:15 PM
        Follow up on the outcome of the WLCG review on the accounting and pledge management tools 15m
        Speaker: Panos Paparrigopoulos (CERN)
    • 9:00 AM 10:30 AM
      WLCG: Operations Hoersaal

      Hoersaal

      DESY

      • 9:00 AM
        New WLCG helpdesk. Status and plans. 20m
        Speaker: Pavel Weber
      • 9:20 AM
        Handling of high memory jobs. Whole-node scheduling. Standard 16-core instead of 8-core slots. VO and site perspective. 1h 10m
        Speakers: Antonio Perez-Calero Yzquierdo (Centro de Investigaciones Energéticas Medioambientales y Tecnológicas), Latchezar Betev (CERN), Rodney Walker (Ludwig Maximilians Universitat (DE))
    • 10:30 AM 11:00 AM
      Coffee 30m
    • 11:00 AM 12:45 PM
      HSF: File Formats and Performance Hoersaal

      Hoersaal

      DESY

      • 11:00 AM
        Development Status of RNtuple, the future HEP Columnar Storage Software Technology 20m

        An intense collaborative work is ongoing about the development and testing of RNTuple, the future HEP columnar storage software technology, involving LHC experiments, DUNE and the ROOT team.

        In this contribution we’ll review the status of the plan of work of RNTuple, towards the freezing of the specification at the end of the year. We’ll review the new features of RNTuple, as well as the work items already delivered this year and the remaining ones. In particular, we’ll show how the currently available I/O infrastructure that already allows writing relevant experiment EDMs in the RNTuple format.

        We’ll complement the aforementioned information with the latest performance plots.

        Speaker: Danilo Piparo (CERN)
      • 11:20 AM
        BALER: Bespoke data compression using autoencoders 20m

        The storage, transmission and processing of data is a major challenge across many fields of physics and industry. Traditional generic data compression techniques are lossless, but are limited in performance and require additional computation.

        BALER [1,2] is an open-source autoencoder-based framework for the development of tailored lossy data compression models suitable for data from multiple disciplines. BALER models can also be used in FPGAs to compress live data from detectors or other sources, potentially allowing for massive increases in network throughput.

        BALER is developed by a cross-disciplinary team of physicists, engineers, computer scientists and industry professionals, and has received substantial contributions from a large number of master’s and doctoral students. BALER has received support from industry both in providing datasets to develop BALER, and to transfer industry best practices.

        This presentation will introduce BALER, demonstrate its performance on a range of data types, discuss the involvement of students and industry in the project and lessons learned, and include a live demonstration.

        [1] https://arxiv.org/pdf/2305.02283.pdf
        [2] https://github.com/baler-collaboration/baler

        Speaker: James Smith (University of Manchester (GB))
      • 11:40 AM
        Analysis Grand Challenge benchmarking tests on selected sites 20m

        A fast turn-around time and ease of use are important factors for systems supporting the analysis of large HEP data samples. We study and compare multiple technical approaches.
        This presentation will be about setting up and benchmarking the Analysis Grand Challenge (AGC) [1] using CMS Open Data. The AGC is an effort to provide a realistic physics analysis with the intent of showcasing the functionality, scalability and feature-completeness of the Scikit-HEP Python ecosystem.
        I will present the results of setting up the necessary software environment for the AGC and benchmarking the analysis' runtime on various computing clusters: the institute SLURM cluster at my home institute, LMU Munich, a SLURM cluster at LRZ (WLCG Tier-2 site) and the analysis facility Vispa [2], operated by RWTH Aachen.
        Each site provides slightly different software environments and modes of operation which poses interesting challenges on the flexibility of a setup like that intended for the AGC.
        Comparing these benchmarks to each other also provides insights about different storage and caching systems. At LRZ and LMU we have regular Grid storage (HDD) as well as and SSD-based XCache server and on Vispa a sophisticated per-node caching system is used.

        [1] https://github.com/iris-hep/analysis-grand-challenge
        [2] https://vispa.physik.rwth-aachen.de/

        Speaker: David Martin Koch (Ludwig Maximilians Universitat (DE))
      • 12:00 PM
        Cloud Data Lake Technologies 20m

        Cloud data lake technologies have been used successfully in industry for analysis of exabyte scale datasets. The technologies that underly this architecture are

        • Object Store
        • Parquet file format
        • Kubernetes
        • Distributed SQL

        We will describe our work using a Trino distributed SQL engine to join selected event data with inference results. We will show how this architecture can eliminate the need to maintain analysis specific copies of datasets.

        Speaker: Benjamin Galewsky (Univ. Illinois at Urbana Champaign (US))
    • 12:45 PM 2:15 PM
      Lunch 1h 30m
    • 2:15 PM 3:30 PM
      Plenary: WLCG Strategy Hoersaal

      Hoersaal

      DESY

      Main workshop plenary session

      • 2:15 PM
        Introduction on [TECH] strategy 20m
        • Discussion about establishing a Technical Coordination Board. define its role wrt e.g. the GDB. Define how activities, projects, working groups, task forces are in relations with the TCB (e.g. DOMA, Auth WG, ...)
    • 3:30 PM 3:35 PM
      Photograph 5m
    • 3:35 PM 4:00 PM
      Coffee 25m
    • 4:00 PM 5:30 PM
      HSF: Training Hoersaal

      Hoersaal

      DESY

      • 4:00 PM
        HSF Training 10m
        Speaker: Alexander Moreno Briceño (Universidad Antonio Nariño)
      • 4:10 PM
        HELIOS 10m
        Speaker: Francesca Calegari
      • 4:20 PM
        SMARTHEP 10m
        Speaker: Jamie Gooding
      • 4:30 PM
        ErUM-Data-Hub 10m
        Speaker: Angela Warkentin
      • 4:40 PM
        HEP Experiments 10m
        Speaker: Valeriia Lukashenko
      • 5:00 PM
        HSF Training Discussion 30m
    • 4:00 PM 5:30 PM
      WLCG: Collaboration Board Seminar Room 1

      Seminar Room 1

      DESY

      • 4:00 PM
        Introduction to the Collaboration Board 20m

        Items to be discussed:

        • Endorsement of the reviewed MoU Annexes
        • Review of Annex 5 and proposals for change, including the evolution of the GDB, the creation of a TCB, the process to endorse observers/partners, the LCG project leader role, the terms of the various roles and boards
    • 7:00 PM 11:00 PM
      Dinner 4h Altes Mädchen

      Altes Mädchen

    • 9:00 AM 10:30 AM
      HSF: Fitting and Reinterpretation Seminar Room 4

      Seminar Room 4

      DESY

      • 9:00 AM
        Contur: A tool for reinterpreting particle-level measurements 20m

        Contur (Constraints On New Theories Using Rivet) is a public python package sitting on top of Rivet and Yoda, which allows information on new BSM models to be extracted from particle-level differential cross section measurements from the LHC. BSM events simulated by a general-purpose MC event generator are "signal injected" into the fiducial phase space of hundreds of measurements simultaneously, allowing a rapid scan of a wide range of model parameters and signatures. Contur takes as input the Yoda histograms from Rivet, and so can interoperate with any generator producing HepMC events. However, it also has convenience methods for parameter scanning using Herwig, and is interfaced to the scanning machinery of Madgraph and GAMBIT.

        Speaker: Jonathan Butterworth (UCL)
      • 9:20 AM
        Constructing model-agnostic likelihoods, a method for the reinterpretation of particle physics results 20m

        Experimental High Energy Physics has entered an era of precision measurements. However, measurements of many of the accessible processes assume that the final states' underlying kinematic distribution is the same as the Standard Model prediction. This assumption introduces an implicit model-dependency into the measurement, rendering the reinterpretation of the experimental analysis complicated without reanalysing the underlying data. We present a novel reweighting method in order to perform reinterpretation of particle physics measurements. It makes use of reweighting the Standard Model templates according to kinematic signal distributions of alternative theoretical models, prior to performing the statistical analysis. The generality of this method allows us to perform statistical inference in the space of theoretical parameters, assuming different kinematic distributions, according to a beyond Standard Model prediction. We implement our method as an extension to the pyhf software and interface it with the EOS software, which allows us to perform flavor physics phenomenology studies. Furthermore, we argue that, beyond the pyhf or HistFactory likelihood specification, only minimal information is necessary to make a likelihood model-agnostic and hence easily reinterpretable. We showcase that publishing such likelihoods is crucial for a full exploitation of experimental results.

        Speaker: Lorenz Gärtner (LMU)
      • 9:40 AM
        Model fitting in Python with zfit and Scikit-HEP 20m

        The Python HEP analysis ecosystem and its user base grew significantly in the last few years, and with it the need for advanced statistical inference tools involving likelihood fits; a core part of most analyses in HEP.
        zfit started over five years ago with the goal to provide this capability, a library for model fitting in HEP: scalable - in terms of model building complexity and performance - and pythonic - well-integrated into the Python ecosystem.
        After many iterations with users and a long development process, zfit reaches a maturity stage.
        In this talk, we will go over the extensive feature set of zfit: from binned and unbinned fits, extensive model building and the ability to create custom models up to advanced likelihood building, weighted fits and a variety of available minimizers. Thanks to its modern numpy-like backend, TensorFlow, with just-in-time compilation and the ability to run on CPUs and GPU, zfit is highly performant. zfit is also well-embeded into the Scikit-HEP ecosystem and beyond: it seamlessly integrates for data loading, plotting and more statistical tools, and allows libraries that build sophisticade models, such as ComPWA and more, to use zfit for statistical inference.

        Speaker: Jonas Eschle (Syracuse University (US))
      • 10:00 AM
        NUISANCE <3 HEPData: Automated Neutrino Scattering Comparisons 20m

        NUISANCE is a neutrino event generator prediction comparison and tuning framework. It facilitates cross-section predictions for the five main event generators in use by the few-GeV neutrino scattering community, enabling non-expert users to compare predictions to over 350 neutrino cross-section measurements, from the historical to the cutting edge.

        We are currently in the process of re-designing NUISANCE to meet the needs of next generation of neutrino experiments. A key goal for this effort is to tightly couple NUISANCE to HepData, which will allow us to offload the responsibility of managing experimental data releases back to the experimental collaborations via a repository expressly built for the job, HepData. A technological requirement for this is the ability to execute analysis code packaged in the HepData releases. This talk will introduce NUISANCE and discuss our approach to solving this problem, which is based on providing a standardised and extensible language-agnostic event-processing framework, with a working implementation in C++, leveraging HEP standard tools: HepMC3 and cling.

        Speaker: Patrick Stowell
    • 9:00 AM 10:30 AM
      WLCG: WLCG Strategy Hoersaal

      Hoersaal

      DESY

      • 9:00 AM
        Introduction to [INFRA] strategy 20m

        In this session we will discuss the aspects of the strategy related to the infrastructure [INFRA]. The open items from the community feedback on the strategy include:

        • Site Capabilities, see [INFRA-1]. Do we put some effort in exposing them? Pros, cons, concerns?
        • Approach towards special sites (hpc, commercial clouds), see [INFRA-2 to 5] and [INFRA-6, 7]. Discuss how to define usage policies. Discussion about producing Cloud and HPC blueprints.
        • GPU provisioning and pledging. Should the focus in the next couple of years be on measuring the use or trying to set up a pledging model? How and when considering pledging? The timescale obviously is driven by the experiments.
        • Analysis Facilities. What is the role (and stake) of WLCG?
        • What is the need for experiment dedicated effort at the sites (particularly large sites)?
    • 10:30 AM 11:00 AM
      Coffee 30m
    • 11:00 AM 12:30 PM
      HSF: Accelerated Computing Seminar Room 4

      Seminar Room 4

      DESY

      • 11:00 AM
        Accelerating Monte Carlo event generator analysis using efficient scalable data handling 20m

        In the age of GPU-accelerated event generation, pivotal community tools like HepMC and Rivet, vital for event generation infrastructure and Monte Carlo event analysis, risk becoming significant bottlenecks in the near future.

        We present an adaptable and highly efficient approach to simulating collider events featuring multi-jet final states, encompassing both leading and next-to-leading order QCD calculations. Rooted in an enhanced parton-level event file format with streamlined scalable data management, our technique offers a scalable solution for producing high-precision calculations on HPC clusters using modern hardware architectures. We verify the efficacy of our framework across various processes, notably Higgs boson plus multi-jet production with up to seven jets, and showcase its integration within the Sherpa and Pythia event generators. Augmented by an enhanced interface for data management in massively parallel applications in Rivet4, our approach represents a significant step towards facilitating efficient data-model comparisons and statistical interpretations in collider physics.

        Speaker: Christian Gutschow (UCL (UK))
      • 11:20 AM
        Portable event generation on GPU-accelerated hardware 20m

        High-precision calculations are crucial for the success of the LHC physics programme. However, the rising computational complexity for high-multiplicity final states is threatening to become a limiting bottleneck in the coming years. At the same time, the rapid deployment of non-traditional GPU-based computing hardware in data centres around the world demands an overhaul of the event generator design.

        We propose a flexible and efficient approach for simulating collider events with multi-jet final states, based on the first portable leading-order parton-level event generation framework, along with an GPU-accelerated version of LHAPDF for fast and efficient evaluation of parton distribution functions. Our approach lends itself neatly to most modern GPU-accelerated hardware, allowing to better exploit computing resources in large-scale production campaigns, and paving the way for economically and ecologically sustainable event generation in the high-luminosity era.

        Speaker: Max Knobbe (University of Göttingen)
      • 11:40 AM
        Experiences on the software performance of LHCb's first level software trigger 20m

        Since 2022, the LHCb detector is taking data with a full software trigger at the LHC proton-proton collision rate, implemented in GPUs in the first stage and CPUs in the second stage. This setup allows to perform the alignment & calibration online and to perform physics analyses directly on the output of the online reconstruction, following the real-time analysis paradigm.
        This talk will focus on the first level of the LHCb trigger implementation on GPUs, discuss challenges of using a heterogeneous architecture and report on the experience from the first running periods in 2022 and 2023.

        Speaker: Arthur Hennequin (CERN)
      • 12:00 PM
        Traccc: Track Reconstruction on GPU in ACTS 20m

        Reconstructing the tracks left by charged particles in modern HEP detectors is one of the most computationally challenging tasks in analyzing the data of modern experiments. During the High-Luminosity LHC era the LHC experiments, including ATLAS, will have to be able to process much more complex data at much higher rates than ever before.

        To achieve this, GPU accelerated code has been developed as an R&D effort as part of the ACTS project (https://acts.readthedocs.io). With ATLAS preparing to use ACTS for all of its CPU based track reconstruction during LHC's Run-4, we plan to integrate the GPU accelerated algorithms/tools from ACTS into ATLAS's offline, and possibly trigger reconstruction.

        In this talk we present the latest status of the ACTS Parallelization R&D effort, with updated (physics and computing) performance figures.

        Speaker: Attila Krasznahorkay (CERN)
    • 11:00 AM 12:30 PM
      WLCG: WLCG Strategy Hoersaal

      Hoersaal

      DESY

      • 11:00 AM
        Introduction to [FIN] strategy 20m

        In this section we will discuss the financial part [FIN] of the WLCG strategy. Items to be discussed based on the feedback on the strategy document:

        • How to develop a multi-year resource planning collecting the input from the experiments.
        • What are the conditions for considering HPC capacity as part of the pledge? Should be documented as a formal WLCG document.
        • Should we monitor pledges against authorship levels (some think is a good idea, some think it is not so good)
        • Do we need fine grain pledges (monthly?, quarterly?) Discussed in the past, partially implemented, never used
        • Can we setup a lightweight mechanism to highlight the commitments for SW/MW development? Is this useful for the SW and MW developers to secure funding?
    • 12:45 PM 2:15 PM
      Lunch 1h 30m
    • 2:15 PM 3:30 PM
      Plenary: Analysis Facilities Hoersaal

      Hoersaal

      DESY

      Main workshop plenary session

      • 2:15 PM
        Session introduction 5m
        Speaker: Alessandra Forti (University of Manchester (GB))
      • 2:20 PM
        LHCC questions draft document discussion 1h 10m
        Speaker: Alessandra Forti (University of Manchester (GB))
    • 3:30 PM 4:00 PM
      Coffee 30m
    • 4:00 PM 5:50 PM
      Plenary: Analysis Facilities Hoersaal

      Hoersaal

      DESY

      Main workshop plenary session

      • 4:00 PM
        Demonstrator Analysis 200 Gb/s 20m
        Speaker: Brian Paul Bockelman (University of Wisconsin Madison (US))
      • 4:20 PM
        User experience discussion 30m
        Speaker: Gordon Watts (University of Washington (US))
      • 4:50 PM
        Data Access discussion 30m
        Speaker: Nick Smith (Fermi National Accelerator Lab. (US))
      • 5:20 PM
        User Monitoring 30m
        Speaker: Robert William Gardner Jr (University of Chicago (US))
    • 9:00 AM 10:30 AM
      Plenary: Closing Hoersaal

      Hoersaal

      DESY

      Main workshop plenary session

      • 9:00 AM
        WLCG governance - recap 10m
      • 9:15 AM
        HSF summary and next steps 10m
      • 9:30 AM
        Analysis Facilities - summary and next steps 15m
      • 10:00 AM
        OpenScience EVERSE 15m
        Speaker: Guido Juckeland (Helmholtz-Zentrum Dresden-Rossendorf (HZDR))
    • 10:30 AM 11:00 AM
      Coffee 30m
    • 11:00 AM 12:30 PM
      Plenary: Closing Hoersaal

      Hoersaal

      DESY

      Main workshop plenary session

      • 11:00 AM
        OpenScience OSCAR/ESCAPE 15m
        Speaker: Xavier Espinal (CERN)
      • 11:25 AM
        Data Challenges - DC24 - summary and next steps 15m
      • 11:50 AM
        Closing 10m