CASUS Institute Seminar, Axel Huebl, Lawrence Berkeley National Laboratory (LBNL), University of Berkeley, USA
bio // Axel is a computational physicist at Berkeley Lab. He is researching advanced particle accelerators with computational modeling tools, developing as a staff member the Beam pLasma Accelerator Simulation Toolkit (BLAST) that includes applications such as the DOE Exascale Computing Project (ECP) application WarpX. Before joining Berkeley Lab, he was part of a team of undergraduates that made it in the SC Gordon Bell finals in 2013, with the first particle-in-cell code running on the newly released Titan GPU cluster, called PIConGPU. Axel is a long-time user of scalable file formats and contributes to ECP libraries such as ADIOS2. For his applications, he also develops data reduction algorithms, including performance modeling, and interactive parallel data analysis. He co-authored and supervised the inception of the C++ performance portability framework Alpaka, the in situ visualization framework ISAAC and leads an open data standard, openPMD, which is now popular in plasma, beam and accelerator physics.
abstract // Particle accelerators are among the largest and most complex devices in science. The next generation of accelerators demands increasingly sophisticated computational tools for design and optimization so that the challenges of increasing energy, intensity, accuracy, compactness, complexity and efficiency will be met. It is key that contemporary software take advantage of the latest advances in computer hardware and scientific software engineering practices, delivering speed, reproducibility and feature composability for the aforementioned challenges.
In his talk, Axel will describe the software stack that is being developed at the heart of the BLAST by LBNL and collaborators. He will first describe how the US DOE (ECP application WarpX [1-3] will exploit the power of GPUs and its performance on Exascale supercomputers for the modeling of laser-plasma acceleration. Then the talk will continue about how the team is leveraging the ECP experience to develop a new generation ecosystem of codes that, combined with machine learning, will deliver modeling (from ultrafast to ultraprecise) for future accelerator design and operations, ultimately resulting in enabling virtual twins of accelerators.
[1] J.-L. Vay, A. Huebl, et al, Phys. Plasmas 28, 023105 (2021)
[2] J.-L. Vay, A. Huebl, et al, J. Instr. 16, T10003 (2021)
[3] A. Myers, et al (incl. A. Huebl), Parallel Comput. 108, 102833 (2021)
The event is organized with a videoconferencing tool by Zoom Inc. If you want to join the talk remotely, please ask for the login details via contact@casus.science.