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:20260731T133318Z LOCATION:Bldg. 6 - 001 - Plenary Room DTSTART;TZID=Europe/Stockholm:20260630T121500 DTEND;TZID=Europe/Stockholm:20260630T124500 UID:submissions.pasc-conference.org_PASC26_sess129@linklings.com SUMMARY:Flash Poster Session - Part II DESCRIPTION:Access the recording\n\nP28 - Performance-Portable and Highly Scalable Spectral Transforms with ecTrans\n\nThe continued increase in the skill of weather forecasts observed over the past decades depends crucial ly on the efficient exploitation of the next generation of high-performanc e computers by Earth-system models. The European Centre for Medium-Range W eather Forecast (ECMWF)'s model, the IFS, is one ...\n\n\nSam Hatfield (EC MWF)\n---------------------\nP29 - Perspectives on Teamwork and AI in Scie ntific Computing\n\nThe development and use of high-quality software—a pri mary mechanism for sustained collaboration and progress in scientific comp uting—is undergoing profound change, driven by increasing complexity in sc ientific drivers and computing architectures and the rapid adoption of AI, placing dem...\n\n\nOlivia B. Newton (University of Montana), Anshu Dubey (Argonne National Laboratory), Denice Ward Hood (University of Illinois U rbana-Champaign), Lois Curfman McInnes (Argonne National Laboratory), and Santiago Ospina Tabares (University of Illinois Urbana-Champaign)\n------- --------------\nP31 - Predictive Alerts and Atmospheric Data for Airport W indshear by CPAS 200m Weather Model\n\nLow-level wind shear (LLWS) at Hong Kong International Airport (HKIA) poses significant aviation risks, prima rily driven by terrain-induced turbulence from Lantau Island and sea-breez e interactions. While current mitigation relies on real-time detection and short-term forecasting, for operational pu...\n\n\nSai Lun Tin, Chi Chiu Cheung, Ka Ki Ng, and Wai Pang Sze (ClusterTech Limited)\n---------------- -----\nP32 - Profile-Guided-Optimisation of Lattice QCD Contractions on CP U and GPU\n\nWe present a performance optimisation study of the 2+2 discon nected component bottleneck of Lattice QCD computation of the hadronic lig ht-by-light contribution to the muon's anomalous magnetic moment. Optimiza tion on CPU and GPU architectures were guided by popular profiling tools p erf, valgrind, ns...\n\n\nJingJing Li, Urs Wenger, and Roman Gruber (Unive rsity of Bern)\n---------------------\nP33 - RDQ: A Zero-Copy Remote Data Queue for In-Situ Machine Learning in HPC\n\nThe integration of machine le arning into the computational sciences is increasingly pursued to reduce t ime-to-solution, alleviate I/O bottlenecks, and enable adaptive analysis d uring simulation. We present RDQ (Remote Data Queue), a library for coupli ng HPC simulations and machine learning training ...\n\n\nMaximilian Sande r (TU Dresden) and Jens Domke (RIKEN)\n---------------------\nP35 - Sheddi ng Light on the Solar Dynamo Using Bayesian Data Science\n\nSolar magnetic activity exhibits an approximately 11-year cycle that is far from stable, showing strong long-term variability and recurrent episodes of strongly r educed activity known as Grand Minima. Understanding the origin of these f eatures remains a fundamental challenge in solar physics and is ...\n\n\nS imone Ulzega (Zurich University of Applied Sciences, Institute of Computat ional Life Sciences) and Carlo Albert (Swiss Federal Institute of Aquatic Science and Technology (EAWAG))\n---------------------\nP36 - Stencil Comp utation on Tenstorrent Wormhole\n\nThe rapid ascent of large language mode ls (LLMs) has prioritized domain-specific accelerators (DSAs) optimized fo r dense matrix-based deep learning. However, the suitability of these arch itectures for traditional high-performance computing (HPC) kernels, like s tencil-based partial differential equat...\n\n\nLorenzo Piarulli and Danie le De Sensi (Sapienza University)\n---------------------\nP37 - Structured Reinforcement Learning for Loop Transformation in MLIR\n\nOptimizing poly hedral kernels for modern multicore architectures is a high-dimensional, n on-convex problem where small structural changes often yield orders-of-mag nitude runtime variation. While traditional compilers rely on rigid static heuristics and autotuners require prohibitive search times, th...\n\n\nAb rar Hossain (University of Toledo)\n---------------------\nP38 - Tracking Mechanistic Evolution Across Brain Tissues and Cell Types Using Multiplex Networks\n\nTracking the evolution of biological function remains a major challenge in computational biology, as existing approaches are often limit ed to sequence conservation, gene presence, or predefined pathways. These methods can fail to identify conserved functional mechanisms even though c onstituent genes...\n\n\nKenneth Smith (Oak Ridge National Laboratory); Ma tthew Lane (University of Tennessee); Alice Townsend and Jean Merlet (Oak Ridge National Laboratory, University of Tennessee); Anna Vlot and Alana W ells (Oak Ridge National Laboratory); and Daniel Jacobson (Oak Ridge Natio nal Laboratory, University of Tennessee)\n---------------------\nP39 - Tun ing the Performance of Three-Body Interactions in Molecular Dynamics\n\nMo lecular Dynamics (MD) simulations predict thermophysical properties, yet s tandard pair potentials can lack the desired accuracy for certain applicat ions. Introducing three-body potentials, such as the Axilrod-Teller-Muto m odel, improves results but poses significant computational challenges. Thi s ...\n\n\nMarkus Mühlhäußer, Samuel James Newcome, Fabio Gratl-Gaßner, Ma nish Kumar Mishra, and Hans-Joachim Bungartz (Technical University of Muni ch)\n---------------------\nP41 - Using Generative Machine Learning to Pro duce High-Resolution Weather Data\n\nGenerative machine learning techniqu es show promise for performing atmospheric downscaling (super-resolution f or meteorological data) to produce high-resolution weather and climate sim ulations. Previous work has evaluated the quality of these models using st andard error scores such as RMSE, absolut...\n\n\nPetar Stamenkovic and Ma ry McGlohon (MeteoSwiss, ETH Zurich); David Leutwyler and Xavier Lapillonn e (MeteoSwiss); Fabian Bösch, Lukas Drescher, and Henrique Mendonça (ETH Z urich / CSCS); Sebastian Schemm (University of Cambridge); Siddhartha Mish ra (ETH Zurich); and Oliver Fuhrer (MeteoSwiss)\n---------------------\nAC MP01 - Adaptive Multidimensional Quadrature on Multi-GPU Systems\n\nThe pr esent work introduces a distributed adaptive deterministic quadrature fram ework for high-dimensional integration on multi-GPU systems, enabling accu rate evaluation of integrals arising in applications such as radiative tra nsfer and probabilistic design. The method is formulated as a hierarchic.. .\n\n\nMelanie Tonarelli (Università della Svizzera Italiana)\n----------- ----------\nACMP02 - Data Augmentation to Improve the Performance of Deep Learning-Based Seismic Inversion\n\nDeep Learning-based Seismic Inversion (DLI) is a promising alternative to Full Waveform Inversion (FWI), but str uggles with data scarcity. This work evaluates data augmentation technique s adapted from computer vision to address this bottleneck, and demonstrate s that generating synthetic earth models...\n\n\nCarlos Gomes de Carvalho Junior (Universidade Federal de São Carlos)\n---------------------\nACMP03 - Deep Reinforcement Learning for Algorithm Selection in Molecular Dynami cs Simulations\n\nShort-range particle simulations are crucial in physics and chemistry, requiring efficient neighbor search and interaction algorit hms. The C++ library AutoPas supports more than 100 algorithms, but no sin gle algorithm is universally optimal. Traditional manual tuning is impract ical due to dynamic si...\n\n\nPatrick Metscher (Technische Universität Mü nchen)\n---------------------\nACMP04 - Fine-Tuning Large Language Models for HPO-Term Recognition\n\nLarge language models (LLMs) offer a promising approach for extracting structured medical information from free-text cli nical notes. This work investigates fine-tuning LLMs for Human Phenotype O ntology (HPO) term recognition, a core task in clinical genetics that requ ires accurate identification and...\n\n\nSrinithi Krishnamoorthy (Cornell University)\n---------------------\nACMP05 - Hypergraph Partitioning for S parse Matrix Reordering\n\nFill-in during sparse matrix factorization rema ins a critical bottleneck in scientific computing. We present an efficient hypergraph partitioning approach for sparse matrix reordering based on th e Clique-Node Hypergraph (CNH) representation, building on prior work by Ç atalyurek et al. and Selvitopi ...\n\n\nRitvik Ranjan (ETH Zurich)\n------ ---------------\nACMP06 - Improving Deep Learning Based Seismic Inversion with Online Augmentation\n\nThis study investigates the application of dat a augmentation techniques to optimize Deep Learning-based Seismic Inversio n (DLI), aiming to overcome the scarcity of labeled datasets in the indust ry. Using the OpenFWI benchmark, the study evaluates four incremental stra tegies: horizontal flipping, syn...\n\n\nLucas Souza (Federal University o f São Carlos)\n---------------------\nACMP07 - Mapping the Productivity-To -Energy Trade-Off in Memory-Bound HPC via DVFS, Core Scaling, and C-State Control on Repurposed Hardware\n\nHigh-performance computing (HPC) systems in resource-constrained environments, such as Africa, often rely on repur posed hardware, shifting the primary financial burden from capital expendi ture to operational energy costs. This work aims to identify realistic and reproducible performance–energy...\n\n\nSuné Toerien (University of the W itwatersrand), Vele Nefale (UniverUniversity of the Witwatersrand), and Nt andoyenkosi Memela and Mubeen Dewan (University of the Witwatersrand)\n--- ------------------\nACMP08 - Matrix-Free vs. Matrix-Based Finite Element S olvers for 3D Advection–Diffusion–Reaction Equations\n\nHigh-order finite element methods are attractive for three-dimensional advection–diffusion–r eaction (ADR) problems, but their efficiency on distributed-memory systems is often limited by memory traffic and communication in large linear solv es. This work compares matrix-based (MAT) and ma...\n\n\nZhaohui Song (Pol itecnico di Milano)\n---------------------\nACMP09 - Medulla: Cluster- and Application-level I/O Performance Diagnosis with LLMs\n\nModern High-Perf ormance Computing (HPC) I/O systems offer immense bandwidth but are notori ously difficult to utilize effectively.\nIdentifying critical bottlenecks is a challenge that persists for both\ndomain scientists and cluster admin istrators. Existing analysis tools\ncan identify common bottlenec...\n\n\n Anish Sathyanarayanan (BITS Pilani K K Birla Goa Campus)\n---------------- -----\nACMP10 - Mixed Precision Acceleration of Light Matter Dynamics: Ena bling HPC Co-Design for Quantum Material Discovery\n\nLight-matter dynamic s in topological quantum materials holds the promise for a sustainable soc iety with ubiquitous and power-hungry artificial intelligence (AI) by enab ling ultralow-power (attojoule) and ultrafast (petahertz) computing and se nsing devices. A challenge is simulating multiple field an...\n\n\nTaufeq Mohammed Razakh (University of Southern California)\n--------------------- \nACMP11 - Mosaic: Automatic Categorization of I/O Patterns from Scientifi c Applications\n\nWhile newly deployed High-Performance Computing (HPC) sy stems embeds more compute capabilities than the generations before, parall el file systems (PFS) are struggling to keep up with this same trend, incr easing the gap between computing power and I/O performance. If new paradig ms and technologies ar...\n\n\nThéo Jolivel (INRIA, Universite de Rennes)\ n---------------------\nACMP12 - Semantic-Aware Implicit Neural Compressio n for Physics Simulations\n\nMachine learning surrogates and data-driven s cientific discovery require efficient access to simulation data, yet physi cs simulations generate terabyte-scale datasets. Traditional compression e ither achieves insufficient ratios or corrupts physics-critical features l ike conservation laws. Implicit n...\n\n\nJessica Ezemba (Carnegie Mellon University)\n---------------------\nACMP13 - Solver-Integrated Lossy and L ossless Compression for Scalable Flow Simulations\n\nLarge-scale computati onal fluid dynamics (CFD) simulations on modern GPU-accelerated supercompu ters generate terabytes of data per run, making checkpointing, storage, an d post-hoc analysis increasingly I/O-bound. This bottleneck limits data re tention and hinders downstream workflows such as visualiz...\n\n\nViral Su dip Shah (University of Illinois Urbana-Champaign)\n---------------------\ nACMP14 - Stability and Accuracy of the r²SCAN Functional for Group-IV Ele mental Solids in a Cost-Aware Workflow Perspective\n\nDensity-functional t heory (DFT) is a central tool in computational materials science, and its predictive power depends critically on the choice of exchange-correlation functional. Here we benchmark the meta-GGA r²SCAN in a study of the group- IV elemental solids C, Si, Ge, and Sn, focusing on stabili...\n\n\nAdonis Haxhijaj (EPFL, ETH Zurich)\n\nSession Chair: Tobias Hodel (University of Bern, Switzerland) END:VEVENT END:VCALENDAR