BEGIN:VCALENDAR VERSION:2.0 PRODID:Linklings LLC BEGIN:VTIMEZONE TZID:Europe/Stockholm X-LIC-LOCATION:Europe/Stockholm BEGIN:DAYLIGHT TZOFFSETFROM:+0100 TZOFFSETTO:+0200 TZNAME:CEST DTSTART:19700308T020000 RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU END:DAYLIGHT BEGIN:STANDARD TZOFFSETFROM:+0200 TZOFFSETTO:+0100 TZNAME:CET DTSTART:19701101T020000 RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU END:STANDARD END:VTIMEZONE BEGIN:VEVENT DTSTAMP:20241120T082409Z LOCATION:HG E 3 DTSTART;TZID=Europe/Stockholm:20240604T160000 DTEND;TZID=Europe/Stockholm:20240604T180000 UID:submissions.pasc-conference.org_PASC24_sess139@linklings.com SUMMARY:MS4E - In Situ Coupling of Simulations and AI/ML for HPC: Software , Methodologies, and Applications - Part II DESCRIPTION:Minisymposium\n\nMotivated by the remarkable success of artifi cial intelligence (AI) and machine learning (ML) in the fields of computer vision and natural language processing, over the last decade there has be en a host of successful applications of AI/ML to a variety of scientific d omains. In most cases, the models are trained using the traditional offlin e (or post hoc) approach, wherein the training data is produced, assembled , and curated separately before training is deployed. While more straightf orward, the offline training workflow can impose some important restrictio ns to the adoption of ML models for scientific applications. To solve thes e limitations, in situ (or online) ML approaches, wherein ML tasks are per formed concurrently to the ongoing simulation, have recently emerged as an attractive new paradigm. In this minisymposium, we explore novel approach es to enable the coupling of state-of-the-art simulation codes with differ ent AI/ML techniques. We discuss the open-source software libraries that a re being developed to solve the software engineering challenges of in situ ML workflows, as well as the methodologies adopted to scale on modern HPC systems and their applications to solve complex problems in different com putational science domains.\n\nTorchFort: A Library for Online Deep Learni ng in Fortran HPC Programs\n\nDeep learning has shown promise in reducing computational cost or as an alternative method for modeling physical pheno mena for a broad range of scientific applications. In these domains, the d ata sources are numerical simulation programs typically implemented in C, C++, or still often, Fortran. This...\n\n\nThorsten Kurth, Josh Romero, an d Massimiliano Fatica (NVIDIA Inc.)\n---------------------\nScaling Couple d Simulation and AI Workflows on Aurora with Dragon\n\nThe advent of exasc ale computing has enabled computational workflows coupling simulation and AI of unprecedented scale and complexity. However, the scale of these work flows presents challenges for the efficient distribution of data between t he various compute tasks spread across large node counts. I...\n\n\nChrist ine Simpson, Riccardo Balin, Archit Vasan, and Sam Foreman (Argonne Nation al Laboratory); Peter Mendygral, Colin Wahl, and Nick Hill (HPE); and Venk atram Vishwanath (Argonne National Laboratory)\n---------------------\nSca lable and Consistent Mesh-Based Modeling of Fluid Flows with Distributed G raph Neural Networks\n\nGraph neural networks (GNNs) have shown considerab le promise in accelerated mesh-based modeling for applications like fluid dynamics, where models must be compatible with unstructured grids for prac tical simulation capability in complex geometries. To realize the vision o f robust mesh-based modeling...\n\n\nShivam Barwey, Riccardo Balin, Bethan y Lusch, Saumil Patel, Ramesh Balakrishnan, and Pinaki Pal (Argonne Nation al Laboratory); Romit Maulik (University of Pennsylvania, Argonne National Laboratory); and Venkatram Vishwanath (Argonne National Laboratory)\n---- -----------------\nSimAI-Bench: A Performance Benchmarking Tool for Couple d Simulation and AI Workflows\n\nIn situ AI/ML workflows, in which ML task s are coupled to an ongoing simulation, are an attractive new paradigm for developing robust and predictive surrogate models for accelerating time t o science by steering simulation ensembles and replacing expensive computa tions. In the world of high performan...\n\n\nRiccardo Balin, Shivam Barwe y, Ramesh Balakrishnan, Bethany Lusch, Saumil Patel, Tom Uram, and Venkatr am Vishwanath (Argonne National Laboratory)\n\nDomain: Chemistry and Mater ials, Climate, Weather, and Earth Sciences, Engineering, Computational Met hods and Applied Mathematics\n\nSession Chair: Alessandro Rigazzi (HPE) END:VEVENT END:VCALENDAR