NGT at ACAT 2025
Last week, members of the Next Generation Triggers project joined the international community at the 23rd International Workshop on Advanced Computing and Analysis Techniques in Physics Research (ACAT 2025), held at the University of Hamburg between 8–12 September.
Jointly organised by DESY and the University of Hamburg, ACAT once again brought together experts from particle, nuclear, astro- and accelerator-physics, alongside specialists in computer science and high-performance computing. This year’s theme, “Transforming the Scientific Process: AI at the Heart of Theory, Experiment, and Computation in High-Energy and Nuclear Physics,” highlighted the central role of artificial intelligence in reshaping theory, experimentation, and computation.
NGT colleagues contributed to the programme through both presentations and poster sessions, sharing their latest research on machine learning, data analysis, and real-time computing for high-energy physics experiments and their work done in NGT. The event offered a unique opportunity for collaboration across disciplines, and many participants underlined the value of presenting their work in such an open and interactive forum.
“To me, ACAT was a great venue to connect with other CERN colleagues and researchers outside of CERN,” said our NGT colleague Jolly Chen, Doctoral Student from University of Twente. “I especially liked the poster sessions, as they provided a lot of opportunities to comfortably ask questions and learn more about the computing problems in HEP as a non-physicist.”
Poster sessions were indeed a highlight for several participants, fostering exchanges with researchers from a variety of experiments.
“It was well organized and I liked that there were only three parallel sessions, which made it easier to follow and created opportunities to talk with colleagues,” noted Simone Rossi Tisbeni, PhD student at the university of Bologna. “Attending the poster session was useful to present the research, and I got some interest also from other experiments such as LHCb.”
A special moment came when Christine Zeh, a NGT technical student, received the ACAT 2025 Poster Prize for her work titled “Efficient Data Movement for Machine Learning Inference in Heterogeneous CMS Software”.
“I wasn’t expecting to receive the prize at all, as there were so many great posters there,” she said. “However, I am very happy that people liked the topic as much as I do. The multiple poster sessions allowed me to discuss my approaches in depth and learn how other participants tackled similar issues. It was great to see that the project I did as a technical student is actually making a difference and solving a current issue.”
Her project tackles one of the main challenges in GPU-based workflows: avoiding unnecessary data transfers between CPU and GPU. By developing an interface that allows machine learning models to directly access data already stored in GPU memory, her solution streamlines the process, reduces overhead, and significantly boosts efficiency. Implemented within the CMS computing framework, this approach not only accelerates model execution but also simplifies integration with existing machine learning tools. You can access her winner poster by clicking here. Similarly, all presentations and posters by NGT members at ACAT2025 are available in the NGT repository, which you can access by clicking here.