Axel Huebl is a Research Software Engineer with ATAP’s Accelerator Modeling Program (AMP). He had previously been a postdoctoral scholar with AMP after earning his PhD and then working as a scientific researcher at Helmholtz-Zentrum Dresden-Rossendorf (HZDR).
With a background in both the user and the computer-science aspects of modeling, Axel researches, oversees, and participates in the development and integration of codes (computer programs) that run on high performance computers to simulate accelerators, particle and laser beams, and plasmas.
1. Advanced software expertise is in great demand. What led you to accelerator modeling and to the Lab?
During my undergraduate studies, I started out early as an intern at the National Laboratory in Dresden, Helmholtz-Zentrum Dresden-Rossendorf (HZDR). A newly established group on laser-particle acceleration caught my interest and I started writing parallel applications, performed simulations on laser-wakefield acceleration and later on we wrote our own particle-in-cell code that ran on graphics processing units, PIConGPU, which I maintained and improved with my colleagues over many years. In 2012/13, the first large-scale GPU-powered supercomputer came online with DOE’s Titan cluster at Oak Ridge National Laboratory and we were the first from our community that could use its new technology. Since then, I have been a regular user of these facilities – performing science runs on laser-ion acceleration, tackling new challenges to run simulations at very large scale, and collaborating with the vibrant U.S. computational and computer science community.
My way to Berkeley Lab started in 2014 when I met Jean-Luc Vay at the International Particle Accelerator Conference (IPAC) in Dresden. He soon after invited me to visit his group as an affiliate over the course of my PhD studies. Together with Rémi Lehe, whom I had met a few years earlier during an Erasmus internship in Lisbon, we collaborated to kick-start an open standard to exchange particle-mesh data (openPMD) and from there on stayed in regular contact with each other. At the end of my PhD studies, the Exascale Project funded WarpX and I was excited to join, contribute with my experience, and deepen my knowledge on accelerator modeling.
2. You have a talent for communication as well as computer modeling. How do you use that in your role here?
During my time working in the German national laboratory environment, I learned that the truly great physics challenges are only achievable as a team. Efficient communication over open channels, be it in outreach, open data or open source publishing, is a way for me to collaborate and integrate scientific work with many knowledgeable people. This helped me a lot in the past to learn, collaborate, help others, and later on guide new students.
I see computer modeling of particle accelerator science questions as a community challenge that needs to be addressed with interdisciplinary teams. While contributing to individual daily challenges, we use our peers’ work and publish our own research as openly as possible, so that others can reuse or contribute to it. I pre-think and establish mechanisms that make this possible — and surprisingly many technical challenges are rooted in finding ways of efficient and scalable communication, e.g., exchange of data, software, results and reproducibility of computational studies.
3. What is the big-picture goal of what you would like to achieve in your career?
My current career path moves along the boundary of computational physics and computer science. I work in a field with still exponentially growing capabilities, where a computing system that comes online today is pretty much obsolete ten years later. I thrive when I am able to actively shape how we model advanced accelerators with cutting edge technology and develop new workflows on how we can derive insight into complex plasma problems. My background in both physics and computer science inspires me in both domains.
My career goal is to advance our capabilities and understanding in both aforementioned fields to address exciting challenges in laser-plasma physics. Using the most powerful available supercomputers is always an appealing prospect, and building communities that are open and inclusive for members to join, enable each other, build and rely upon each other, and compete openly, is an enjoyable and rewarding way to work.