An “open science” approach to modeling particle accelerators could extend the capabilities of today’s machines and usher in the next generation, according to ATAP’s Axel Huebl and Jean-Luc Vay.

However, as these machines become more complex, so do “the tools used to simulate and model accelerator and beam dynamics,” says Axel Huebl, a research scientist and the lead software architect for the Accelerator Modeling Program (AMP) in the Accelerator Technology & Applied Physics (ATAP) Division at Berkeley Lab.

“Advanced software tools are necessary to enhance the capabilities of existing accelerator facilities and enable the creation of future accelerators that use laser-plasma acceleration. But to accurately model the dynamics of beam evolution and their interactions with all the beamline’s components, these tools must have high-fidelity, be efficient and scalable, and function at speeds fast enough to run in tandem with accelerator operations.”

While accelerators and colliders have benefited from exceptional levels of coordination and long-range planning—leading to large-scale accelerator complexes such as those at Berkeley Lab, Fermilab, and CERN. The development of the tools used for modeling beam and accelerator physics “has been largely uncoordinated,” says Jean-Luc Vay, a senior scientist and the head of AMP.

Developing the next generation of advanced modeling software for accelerators, says Vay, will mean moving the modeling community toward an “open science” ecosystem. “This ecosystem would follow open-source principles by providing current and future collaborators with access to open-source software, open documentation, open data, and open data standards.”

Towards this end, AMP researchers are working with several other national labs, including Fermilab, SLAC National Accelerator Laboratory, Argonne National Laboratory, Oak Ridge National Laboratory, as well as colleagues from the University of California, Los Angeles, to develop an open-science ecosystem for accelerator modeling. The work is being carried out under the Department of Energy Scientific Discovery Through Advanced Computing (SciDAC) Collaboration for Advanced Modeling of Particle Accelerators.

This ecosystem “must also include standardized data-handling protocols and user interfaces to allow everyone in the community—scientists, engineers, mathematicians—to contribute to new software frameworks and their associated tools,” says Huebl.

Building such an ecosystem, he adds, will require “the cooperation and coordination of multi-institutional and multi-disciplinary teams of experts working with the latest high-performance computing hardware, which includes the use of desktops, medium-sized computers, and the world’s largest supercomputers.”

According to Huebl and Vay, their open science approach is inspired by the AMP team’s experience working openly with interdisciplinary experts to develop the award-winning WarpX code. A Particle-in-Cell (PIC) simulation code, WarpX was developed under the Department of Energy’s Exascale Computing Project by an international team of researchers led by Vay and in collaboration with Berkeley Lab’s Applied Mathematics & Computational Research Division, as well as many other organizations. Last November, WarpX won the prestigious Associated Computing for Machinery Gordon Bell Prize, which recognizes outstanding achievements in high-performance computing applied to science, engineering, and large-scale data analytics challenges.

“The new ecosystem,” explains Huebl, “will provide the accelerator modeling community with a common or shared ‘language’ that describes machine and beam parameters, design optimization workflows, and which can evolve—like human languages—to not only meet current modeling needs but can also address the needs of future beamlines.”

Central to the success of this ecosystem, he says, will be common standards for controlling simulations and for sharing and exchanging data among the collaborators, with accompanying documentation and simple installation steps that provide access for new members of the community. For instance, Huebl leads the international open particle-mesh data (openPMD) metadata community to ensure data compatibility with the scientists working on a reusable user interface called the particle-in-cell modeling interface (PICMI).

“The development of standardized protocols will allow the ecosystem to be extended to other codes or frameworks that also adopt the standards,” says Vay, and could open the door to a community of interconnected accelerator modelers and offer “limitless new opportunities for creating more powerful modeling tools for both conventional and laser-plasma accelerators.”

Furthermore, he says the openness of this ecosystem would allow for the seamless integration of novel data-driven approaches based on machine-learning techniques that offer the potential for trained models to be used as fast “surrogates” in long-time particle tracking and storage ring models.

“Holistic, start-to-end accelerator modeling requires tools capable of covering orders of magnitude in temporal and spatial scales,” says Huebl. “Therefore, compatible open source and data models will need to connect short-scale element modeling with long-scale beam dynamics modeling, like those used in our new ImpactX code,” which he notes has recently entered the public testing phase.

ATAP Division Director Cameron Geddes said the research “builds on the great success of WarpX and is another excellent demonstration of how ATAP researchers, working with colleagues from across the national labs and other institutions, are creating new tools for enabling advanced particle accelerator technologies that promise to deliver scientific discoveries and open new areas of research.”

 

Learn More

  1. Huebl, A. and Vay, J.-L. “Accessible Exascale Simulation Tools for Particle Accelerators and Beams,” APS-DPB Newsletter (2023), https://engage.aps.org/dpb/resources/newsletters
  2. Huebl, A. et al. “From Compact Plasma Particle Sources to Advanced Accelerators with Modeling at Exascale,” 20th Advanced Accelerator Concepts Workshop (AAC22), https://arxiv.org/abs/2303.12873
  3. Huebl, A., Lehe, R., Mitchell, C. E., Qiang, J., Ryne, R. D., Sandberg, J., Vay, J.-L. “Next Generation Computational Tools for the Modeling and Design of Particle Accelerators at Exascale,” Proceedings of the North American Particle Accelerator Conference (NAPAC22), TUYE2, https://doi.org/10.18429/JACoW-NAPAC2022-TUYE2
  4. SciDAC Collaboration for Advanced Modeling of Particle Accelerators
  5. Huebl, A., Lehe, R., Vay, J.-L., Grote, D. P., Sbalzarini, I., Kuschel, S., Sagan, D., Pérez, F., Koller, F., Bussmann, M. “OpenPMD1.0: Base paths for mesh- and particle-only files and updated attributes,” 2018, https://doi.org/10.5281/zenodo.591699
  6. Particle-in-Cell Modeling Interface (PICMI)
  7. WarpX Documentation
  8. ImpactX Documentation
Tags: