It serves as a component in sparse linear solvers. LIBRSB builds upon its "RSB" hierarchi cal and coarse-grained sparse matrices storage. The RSB data structure and algorithms ...\n\n\nMichele Martone (Leibniz Supercomputing Centre)\n---- -----------------\nP29 - GT4Py: A Python Framework for the Development of High-Performance Weather and Climate Applications\n\nGT4Py is a Python fra mework for weather and climate applications simplifying the development an d maintenance of high-performance codes in prototyping and production envi ronments. GT4Py separates model development from hardware architecture dep endent optimizations, instead of intermixing both togethe...\n\n\nMauro Bi anco and Till Ehrengruber (ETH Zurich / CSCS); Nicoletta Farabullini and A bishek Gopal (ETH Zurich); Linus Groner and Rico Häuselmann (ETH Zurich / CSCS); Peter Kardos, Samuel Kellerhals, and Magdalena Luz (ETH Zurich); Ch ristoph Müller (MeteoSwiss); Enrique G. Paredes (ETH Zurich / CSCS); Matth ias Roethlin (MeteoSwiss); Felix Thaler and Hannes Vogt (ETH Zurich / CSCS ); Benjamin Weber (MeteoSwiss); and Thomas C. Schulthess (ETH Zurich / CSC S)\n---------------------\nP08 - Analysis and Application of CNN to Improv e Deterministic Optical Flow Nowcasting at DWD\n\nOptical flow based nowca sting is essential for several operational productions at DWD, including t ime critical warnings. Precipitation and radar reflectivity nowcasts are p roduced every 5 minutes with a 5 minute stepping up to 2h lead time. The m ethod assumes stationarity of the input data. It is a ...\n\n\nUlrich Frie drich (DWD)\n---------------------\nP52 - Tunable and Portable Extreme-Sca le Drug Discovery Platform at Exascale: the LIGATE Approach\n\nToday digit al revolution is having a dramatic impact on the pharmaceutical industry a nd the entire healthcare system. The implementation of machine learning, e xtreme-scale computer simulations, and big data analytics in the drug desi gn and development process offers an excellent opportunity to lower...\n\n \nAndrea Beccari (Dompé farmaceutici), Silvano Coletti (Chelonia SA), Biag io Cosenza (Università di Salerno), Andrew Emerson (CINECA), Thomas Fahrin ger (University of Innsbruck), Daniele Gregori (E4 Engineering), Philipp G schwandtner (UIBK), Erik Lindahl (KTH Royal Institute of Technology), Jan Martinovic (IT4Innovations National Supercomputing Center), Gianluca Paler mo (Politecnico di Milano), and Torsten Schwede (University of Basel)\n--- ------------------\nP51 - Towards a Python-Based Performance-Portable Fini te-Volume Dynamical Core for Numerical Weather Prediction\n\nWe present re cent progress in the development of a high-performance Python implementati on of the FVM non-hydrostatic dynamical core at ECMWF and its member state partners. The FVM numerical formulation centred about 3D semi-implicit ti me integration of the fully compressible equations with finite-vo...\n\n\n Stefano Ubbiali (ETH Zurich), Till Ehrengruber (ETH Zurich / CSCS), Nicola i Krieger (ETH Zurich), Christian Kühnlein (ECMWF), and Lukas Papritz and Heini Wernli (ETH Zurich)\n---------------------\nP12 - Building a Physics -Constrained, Fast and Stable Machine Learning-Based Radiation Emulator\n\ nModeling the transfer of radiation through the atmosphere is a key compon ent of weather and climate models. The operational radiation scheme in the Icosahedral Nonhydrostatic Weather and Climate Model (ICON) is ecRad. The radiation scheme ecRad is accurate but computationally expensive. It is o perat...\n\n\nGuillaume Bertoli and Sebastian Schemm (ETH Zurich) and Fira t Ozdemir, Fernando Perez Cruz, and Eniko Szekely (Swiss data science cent er)\n---------------------\nP35 - Investigating the Mechanism of a Local W indstorm in the Swiss Alps Using Large-Eddy Simulations\n\nThe Laseyer win dstorm is a local and strong wind phenomenon in the narrow Schwende valley in northeastern Switzerland. The phenomenon has raised the interest of me teorologists as it has - in the past - led to derailments of the local tra in. It is characterised by easterly to southeasterly winds duri...\n\n\nNi colai Krieger (ETH Zurich), Christian Kühnlein (ECMWF), and Michael Spreng er and Heini Wernli (ETH Zurich)\n---------------------\nP44 - Numerical S imulation of Gradual Compaction of Granular Materials and the Uncertainty Quantification of the Proposed Mathematical Model\n\nThe poster deals with mathematical modelling of granular materials and focuses on the process o f their gradual compaction called ratchetting. The model of hypoplasticity introduced by E. Bauer et al. is investigated and the problem of stress-c ontrolled hypoplasticity is considered. The behaviour of ...\n\n\nJudita R uncziková and Jan Chleboun (Czech Technical University in Prague)\n------- --------------\nP06 - Accurate Electronic Properties and Intercalation Vol tages of Li-Ion Cathode Materials from Extended Hubbard Functionals\n\nThe design of novel cathode materials for Li-ion batteries requires accurate first-principles predictions of their properties. Density-functional theor y (DFT) with standard (semi-)local functionals fails due to the strong sel f-interaction errors of partially filled d shells of transition-metal (TM) ...\n\n\nIurii Timrov, Francesco Aquilante, and Michele Kotiuga (EPFL); M atteo Cococcioni (University of Pavia); and Nicola Marzari (EPFL)\n------- --------------\nP32 - High-Throughput Computational Screening of Fast Li-I on Conductors\n\nWe present a high-throughput computational screening to f ind fast Li-ion conductors to identify promising candidate materials for a pplication in solid-state electrolytes. Starting with ~30,000 experimental structures sourced from COD, ICSD and MPDS repositories, we performed hig hly automated calcula...\n\n\nTushar Thakur, Loris Ercole, and Nicola Marz ari (EPFL)\n---------------------\nP57 - Partial Charge Prediction and Pat tern Extraction from a AttentiveFP Graph Neural Network\n\nMolecular dynam ics (MD) simulations enable the time-resolved study of bio-molecular proce sses. The quality of MD simulations is, however, highly dependent on the s et of interaction parameters used, so-called force fields. The accurate pa rtial-charge assignment of all simulated atoms is hence a cruci...\n\n\nMa rc Thierry Lehner (ETH Zurich)\n---------------------\nP21 - Doppler-Boost ed Lasers: A New Path to Extreme QED Pair Plasmas in Light-Matter and Ligh t-Quantum Vacuum Interactions\n\nHow does light interact with matter or th e quantum vacuum at intensities where the physics is governed by Quantum E lectrodynamics (QED)? What are the properties of the QED electron-positron pair plasma produced in those interactions? Can the probing of this plasm a help address open problems in quant...\n\n\nHenri Vincenti, Luca Fedeli, Neil Zaim, Antonin Sainte-Marie, Pierre Bartoli, and Thomas Clark (CEA) a nd Jean-Luc Vay and Axel Huebl (Lawrence Berkeley National Laboratory)\n-- -------------------\nP27 - GPU-Accelerated Modelling of Greenhouse Gases a nd Air Pollutants in ICON with OpenACC\n\nReleasing excess greenhouse gase s into the atmosphere is the major cause of its natural composition altern ation and climate change. Computational modelling of the atmospheric chemi stry and transport processes has played a vital role in enhancing our unde rstanding of such complex phenomena and helped...\n\n\nArash Hamzehloo and Dominik Brunner (Empa)\n---------------------\nP45 - Optimization of Non- Conventional Airfoils for Martian Rotorcraft with Direct Numerical Simulat ions Using High-Performance Computing\n\nDesign of rotorcraft for Mars is challenging due to the very low density and low speed of sound compared to Earth. These conditions require Martian rotor blades to operate in a low- Reynolds-number (1,000 to 10,000 based on chord) compressible flow regime, atypical of conventional, terrestrial helico...\n\n\nLidia Caros, Oliver Buxton, and Peter Vincent (Imperial College London)\n--------------------- \nP04 - A Research Software Engineering Workflow for Computational Science and Engineering\n\nWe present a Research Software Engineering (RSE) workf low for developing research software in Computational Science and Engineer ing (CSE) in university research groups. Their members have backgrounds fr om different scientific disciplines and often lack education in RSE. Resea rch software development...\n\n\nMoritz Schwarzmeier, Tomislav Marić, Tobi as Tolle, Jan-Patrick Lehr, Ioannis Pappagianidis, Benjamin Lambie, Dieter Bothe, and Christian Bischof (TU Darmstadt)\n---------------------\nP42 - Multigrid in H(curl) on Hybrid Tetrahedral Grids\n\nIn many applications large scale solvers for Maxwell's equations are an indispensable tool. Thi s work presents theory and algorithms that are relevant to the solution of Maxwell's equations as well as their implementation in HyTeG. We focus on multigrid methods for the curl-curl-problem which arises...\n\n\nDaniel B auer (Friedrich-Alexander-Universität Erlangen-Nürnberg)\n---------------- -----\nP47 - Parallel Training of Deep Neural Networks\n\nDeep neural netw orks (DNNs) are used in a wide range of application areas and scientific f ields. The accuracy and the expressivity of the DNNs are tightly coupled t o the number of parameters of the network as well as the amount of data us ed for training. As a consequence, the networks and the amount...\n\n\nSam uel Cruz (Università della Svizzera italiana, UniDistance Suisse); Alena K opanicakova (Brown University, Università della Svizzera italiana); Hardik Kothari (Università della Svizzera italiana); and Rolf Krause (Università della Svizzera italiana, UniDistance Suisse)\n---------------------\nP24 - Geodynamo Simulations in a Full Sphere\n\nAlthough the geomagnetic field exists since about 4 Gyr, recent estimates for the formation of the Earth 's inner core go back no further than 500 Myr to 1 Gyr. Here we run rapidl y rotating dynamos in a full sphere geometry, representative of the Earth' s dynamo before the nucleation of the inner core...\n\n\nFabian Burmann, J iawen Luo, Philippe David Marti, and Andrew Jackson (ETH Zurich)\n-------- -------------\nP01 - A Language-Interoperable C++-Based Memory-Manager for the ICON Climate and Weather Prediction Model\n\nHPC machines now use acc elerators such as GPUs. In addition, CPUs themselves now feature many core s as well as special fast memory, like the Fujistu A64FX and Intel Sapphir e Rapids. These rapid changes create important challenges for simulation c odes to accommodate different parallel programming mod...\n\n\nClaudius Ho leksa (Karlsruhe Institute of Technology), Ralf Müller and Jörg Behrens (G erman Climate Computing Centre), Florian Prill (DWD), Christopher Bignamin i and Will Sawyer (ETH Zurich / CSCS), Xavier Lapillonne (MeteoSwiss), Ser gey Kosukhin and Daniel Klocke (Max Planck Institute for Meteorology), Ter ry Cojean and Yen-Chen Chen (Karlsruhe Institute of Technology), Hartwig A nzt (University of Tennessee), and Claudia Frauen (German Climate Computin g Centre)\n---------------------\nP14 - Closing the Gap: Aligning Develope rs’ Expectations and Users’ Practices in Cloud Computing Infrastructure\n\ nThere are often discrepancies between the uses that infrastructure develo pers envision for their technology and the way they are implemented in rea lity. We report on this gap between expectation and practice based on our ongoing study of the user-experience on a national cyberinfrastructure sys tem f...\n\n\nTamanna Motahar, Johanna Cohoon, Kazi Sinthia Kabir, and Jas on Wiese (University of Utah)\n---------------------\nP41 - MPI for Multi- Core, Multi Socket, and GPU Architectures: Optimised Shared Memory Allredu ce\n\nIn the literature the benefits of shared memory collectives especial ly allreduce have been shown. This intra-node communication is not only ne cessary for single node communications but it is also a key component of m ore complex inter-node communication algorithms [1]. In contrast to [2], o ur impleme...\n\n\nAndreas Jocksch and Jean-Guillaume Piccinali (ETH Zuric h / CSCS)\n---------------------\nP20 - Docker Container in DWD's Seamless INtegrated FOrecastiNg sYstem (SINFONY)\n\nAt Deutscher Wetterdienst (DWD ), the SINFONY project has been set up to develop a seamless ensemble pred iction system for convective-scale forecasting with forecast ranges of up to 12 hours. It combines Nowcasting (NWC) techniques with numerical weathe r prediction (NWP) in a seamless way. So far NWC...\n\n\nMatthias Zacharuk (DWD)\n---------------------\nP16 - Denoising Electronic Signals from Par ticle Detectors Using a Flexible Deep Convolutional Autoencoder\n\nIn this work, we present the use of a deep convolutional autoencoder to denoise s ignals from particle detectors. The study of rare particle interactions is crucial in advancing our understanding of the Universe. However, the pres ence of electronic noise makes signal events difficult to distinguish f... \n\n\nMark Anderson, Noah Rowe, and Tianai Ye (Queen's University)\n------ ---------------\nP02 - A Massively Parallel Approach to Forecasting Electr icity Prices\n\nWith the ongoing energy crisis in Europe, accurate forecas ting of electricity price levels and volatility is essential to planning g rid operations and protecting consumers from extreme prices. We present ho w massively parallel stochastic optimal power flow models can be deployed on modern many-core ...\n\n\nTimothy Holt (Università della Svizzera itali ana, Oak Ridge National Laboratory)\n---------------------\nP28 - GPU-Opti mized Tridiagonal and Pentadiagonal System Solvers for Spectral Transforms in QuiCC\n\nQuiCC is a code under development designed to solve the equat ions of magnetohydrodynamics in a full sphere and other geometries. It use s a fully spectral approach to the problem, with the Jones-Worland polynom ials as a radial basis and Spherical Harmonics as a spherical basis. We pr esent an alternat...\n\n\nDmitrii Tolmachev, Philippe Marti, and Giacomo C astiglioni (ETH Zurich); Daniel Ganellari (ETH Zurich / CSCS); and Andrew Jackson (ETH Zurich)\n---------------------\nP48 - ProtoX: A First Look\n\ nStencil operation is a key component in the numerical solution of partial differential equations. Developers tend to use different libraries that p rovide these operations for them. One such library is Proto. It is a C++ b ased domain specific library designed to provide an intuitive interface th at op...\n\n\nHet Mankad and Sanil Rao (Carnegie Mellon University), Phil Colella and Brian Van Straalen (Lawrence Berkeley National Laboratory), an d Franz Franchetti (Carnegie Mellon University)\n---------------------\nP2 5 - Ginkgo — A High-Performance Portable Numerical Linear Algebra Software \n\nNumerical linear algebra building blocks are used in many modern scien tific applications codes. Ginkgo is an open-source numerical linear algebr a software designed around the principles of portability, flexibility, usa bility, and performance. The Ginkgo library is integrated into the deal.II , MFEM, ...\n\n\nTerry Cojean and Isha Aggarwal (Karlsruhe Institute of Te chnology); Natalie Beams and Hartwig Anzt (University of Tennessee); and Y en-Chen Chen, Thomas Grützmacher, Fritz Göbel, Marcel Koch, Gregor Olenik, Pratik Nayak, Tobias Ribizel, and Yu-Hsiang Tsai (Karlsruhe Institute of Technology)\n---------------------\nP18 - Directive-Based, Fortran/C++ Int eroperable Approach to GPU Offloading of the High Performance Gyrokinetic Turbulence Code GENE-X\n\nThe achievement of high plasma confinement is th e key to realize commercially attractive energy production by magnetic con finement fusion (MCF) devices. Turbulence plays a significant role in main taining the plasma confinement within MCF devices. The GENE-X code is base d on an Eulerian (continuum) a...\n\n\nJordy Trilaksono, Philipp Ulbl, and Andreas Stegmeir (Max Planck Institute for Plasma Physics) and Frank Jenk o (Max Planck Institute for Plasma Physics, University of Texas at Austin) \n---------------------\nP36 - Iterative Refinement With Hierarchical Low- Rank Preconditioners Using Mixed Precision\n\nIt has been shown that the s olution to a dense linear system can be accelerated by using mixed precisi on iterative refinement relying on approximate LU-factorization. While mos t recent work has focused on obtaining such a factorization at a reduced p recision, we investigate an alternative via low-ra...\n\n\nThomas Spendlho fer and Rio Yokota (Tokyo Institute of Technology)\n---------------------\ nP34 - Interpretable Compression of Fluid Flows Using Graph Neural Network s\n\nNeural network (NN) based reduced-order models (ROMs) via autoencodin g have been shown to drastically accelerate traditional computational flui d dynamics (CFD) simulations for rapid design optimization and prediction of fluid flows. However, many real-world applications (e.g. hypersonic pro pulsion, ...\n\n\nShivam Barwey (Argonne National Laboratory) and Romit Ma ulik (Argonne National Laboratory, University of Pennsylvania)\n---------- -----------\nP30 - High Performance Computing Meets Approximate Bayesian I nference\n\nDespite the ongoing advancements in Bayesian computing, large- scale inference tasks continue to pose a computational challenge that ofte n requires a trade-off between accuracy and computation time. Combining so lution strategies from the field of high-performance computing with state- of-the-art stati...\n\n\nLisa Gaedke-Merzhäuser (Università della Svizzera italiana), Haavard Rue (King Abdullah University of Science and Technolog y), and Olaf Schenk (Università della Svizzera italiana)\n---------------- -----\nP31 - High-Performance Computing by and for Patient Specific Mechan ical Properties\n\nModeling the mechanical behavior of human trabecular bo nes improves technical applications and the treatment of fractures and bon e or joint related diseases. However, this type of bone consists of a larg e number of struts and plates, resulting in a highly anisotropic and patie nt specific behavior. F...\n\n\nJohannes Gebert, Ralf Schneider, and Micha el Resch (High-Performance Computing Center Stuttgart, University of Stutt gart)\n---------------------\nP26 - Global Sensitivity Analysis of High-Di mensional Models with Correlated Inputs\n\nGlobal sensitivity analysis is an important tool used in many domains of computational science to either gain insight into the mathematical model and interaction of its parameters or study the uncertainty propagation through the input-output interaction s. This works introduces a comprehensive framew...\n\n\nJuraj Kardos and O laf Schenk (Università della Svizzera italiana) and Derek Groen and Diana Suleimenova (Brunel University London)\n---------------------\nP17 - Detec ting Financial Fraud with Graph Neural Networks\n\nDetecting financial fra ud is a challenging classification problem that entails the discovery of s uspicious patterns in large-scale and time evolving data. Traditionally, f inancial institutions have been relying on rule-based methods to identify suspicious accounts, with such approaches becoming inef...\n\n\nJulien Sch midt, Dimosthenis Pasadakis, and Olaf Schenk (Università della Svizzera it aliana)\n---------------------\nP40 - Modeling a Novel Laser-Driven Electr on Acceleration Scheme: Particle-In-Cell Simulations at the Exascale\n\nIn tense femtosecond lasers focused on low-density gas jets can accelerate ul tra-short electron bunches up to very high energies (from hundreds of MeV to several GeV) over a few millimeters or a few centimeters. However, conv entional laser-driven electron acceleration schemes do not provide enough ch...\n\n\nLuca Fedeli (CEA), Axel Huebl (Lawrence Berkeley National Labor atory), France Boillod-Cerneux and Thomas Clark (CEA), Kevin Gott (Lawrenc e Berkeley National Laboratory), Conrad Hillairet (Arm), Stephan Jaure (At os), Adrien Leblanc (ENSTA), Rémi Lehe and Andrew Myers (Lawrence Berkeley National Laboratory), Christelle Piechurski (GENCI), Mitsuhisa Sato (RIKE N), Neil Zaim (CEA), Weiqun Zhang and Jean-Luc Vay (Lawrence Berkeley Nati onal Laboratory), and Henri Vincenti (CEA)\n---------------------\nP22 - E fficient Data Managment in Fully Spectral Dynamo Simulations on Heterogene ous Nodes\n\nOur CFD framework QuICC, based on a fully spectral method, ha s been successfully used for various dynamo simulations in spherical and C artesian geometries. It runs efficiently on a few thousands of cores using a 2D data distribution based on a distributed memory paradigm (MPI). In o rder to better ha...\n\n\nGiacomo Castiglioni, Philippe Marti, and Dmitrii Tolmachev (ETH Zurich); Daniel Ganellari (ETH Zurich / CSCS); and Andy Ja ckson (ETH Zurich)\n---------------------\nP09 - Analyzing Physics-Informe d Neural Networks for Solving Classical Flow Problems\n\nThe application o f Neural Networks (NNs) has been extensively investigated for fluid dynami c problems. A specific form of NNs are Physics-Informed Neural Networks (P INNs), which incorporate physics-based embeddings to account for physical laws. In this work, the performance of PINNs is compared to t...\n\n\nRish abh Puri (Forschungszentrum Jülich); Mario Rüttgers (Forschungszentrum Jül ich, RWTH Aachen University); and Rakesh Sarma and Andreas Lintermann (For schungszentrum Jülich)\n---------------------\nP19 - DNS of Strongly Turbu lent Thermal Convection in a Non-Rotating Full Sphere\n\nBody forces such as gravity can drive convective motion in fluids. Convection due to therma l gradients and the resulting buoyancy force is called thermal convection and occurs ubiquitously in nature. We present results on DNS of thermal co nvection in a non-rotating full sphere and with different bou...\n\n\nTobi as Sternberg, Philippe Marti, Giacomo Gastiglioni, and Andrew Jackson (ETH Zurich)\n---------------------\nP11 - Bridging the Language Gap: Classes for C++/Fortran Interoperability\n\nFortran and C++ remain popular languag es for high-performance scientific computing. Interoperation of these two languages is of great interest; be it to take advantage of a mature ecosys tem of libraries, or for coupling individual simulation codes into larger multi-scale or multi-physics application...\n\n\nIvan Pribec (Leibniz Supe rcomputing Centre)\n---------------------\nP15 - Compressing Multidimensio nal Weather and Climate Data into Neural Networks\n\nWeather and climate s imulations produce petabytes of high-resolution data that are later analyz ed by researchers in order to understand climate change or severe weather. We propose a new method of compressing this multidimensional weather and climate data: a coordinate-based neural network is traine...\n\n\nLangwen Huang and Torsten Hoefler (ETH Zurich)\n---------------------\nP03 - A Nov el Stochastic Parameterization for Lagrangian Modeling of Atmospheric Aero sol Transport\n\nIn recent years, it has become clear that the behavior of atmospheric aerosols has a non-negligible effect on radiative forcing wit hin Earth's climate and the computational models that simulate it [Carslaw , et al., Nature, 2013]. Thus, we must obtain descriptive aerosol models t hat are also predicti...\n\n\nMichael Schmidt (Sandia National Laboratorie s)\n---------------------\nP59 - A Scalable Interior-Point Method for PDE- Constrained Inverse Problems Subject to Inequality Constraints\n\nWe prese nt a scalable computational method for large-scale inverse problems with P DE and inequality constraints. Such problems are used to learn spatially d istributed variables that respect bound constraints and parametrize PDE-ba sed models from unknown or uncertain data. We first briefly overview P...\ n\n\nTucker Hatland and Cosmin Petra (Lawrence Livermore National Laborato ry), Noemi Petra (University of California Merced), and Jingyi Wang (Lawre nce Livermore National Laboratory)\n---------------------\nP07 - Addressin g Exascale Challenges for Numerical Algorithms of Strongly Correlated Latt ice Models\n\nStrongly Correlated Lattice Models play an important role fo r our understanding of Quantum Magnetism, High-Tc superconductors, and als o Quantum Simulators built from cold atoms, trapped ions, Rydberg atoms, o r superconducting qubits. Wave function based numerical algorithms, such a s Exact Diagonaliz...\n\n\nSamuel Gozel (Paul Scherrer Institute) and Andr eas M. Läuchli (Paul Scherrer Institute, EPFL) END:VEVENT END:VCALENDAR