Smarter algorithms and detectors pave the way for the future of particle research at CERN

How far can we push the Standard Model of particle physics? According to young researchers Kunal Gautam and Eduardo Ploerer — doctoral students at the Vrije Universiteit Brussel (VUB) and the University of Zurich — the end is nowhere in sight. Their work provides crucial building blocks for the Future Circular Collider (FCC-ee), the planned successor to CERN’s current Large Hadron Collider (LHC).

The FCC-ee will be an unprecedented precision instrument — but that precision requires intelligent support. Gautam and Ploerer have therefore developed new machine learning algorithms and ultrathin detectors that together form the “eyes and brain” of the future of particle physics.

At the heart of their research lies the DeepJetTransformer, a neural network that identifies patterns in enormous amounts of data from collisions between elementary particles. The algorithm is based on the same transformer technology that powers language models such as ChatGPT. Thanks to this approach, scientists can now much more effectively distinguish which types of quarks — the fundamental building blocks of matter — are involved in a collision, including the elusive strange quarks.

The results are astonishing: processes that once disappeared completely into the noise, such as the rare decay of the Z boson into strange quarks, can now be recognized within seconds. Measurement precision has improved by a factor of a thousand, making it possible to observe subtle effects such as the forward–backward asymmetry for the first time.

Beyond software, the researchers also worked on new hardware. They designed monolithic active pixel sensors (MAPS) — ultrathin silicon detectors with a resolution of less than three micrometers and an efficiency greater than 99%. Even under intense radiation, these sensors remain reliable, making them suitable not only for the future FCC-ee, but also for ongoing experiments such as ALICE at the LHC.

With their innovative combination of artificial intelligence and advanced detection technology, Gautam and Ploerer are giving a significant boost to the search for new physical insights. “We don’t just want to confirm what we already know,” they say, “but stay open to surprises. The FCC-ee might very well be the place where the Standard Model begins to crack.”