In a collider detector, each proton collision leaves a trail of dozens of particles, like the splash from a stone thrown into water. Making sense of that splash and understanding what exactly happened at the moment of impact is a task that humans can no longer handle alone. Neural networks come to the rescue, and EveNet is one of the most ambitious: it was trained on 500 million simulated collisions.
EveNet views each event as a cloud of particles, each with known direction, energy, and type. This approach lets the model grasp overall patterns rather than memorize specific decays. As a result, it outperforms the best previous algorithms in searching for heavy resonances and rare Higgs decays, and it shines especially when data are scarce.
But the most important thing is that the model transfers knowledge from simulations to real data. It doesn't just find particles; it also measures subtle quantum correlations between them, robust against systematic uncertainties. That means EveNet can be trusted for precision measurements where every hundredth of a percent matters.
Such a versatile tool saves time and computational resources. Instead of training a new model for each task, physicists take EveNet and fine-tune it on a small dataset. For future colliders, where data rates will multiply, this is not just a convenience: it's a necessity.
as written by the authors arXiv:2601.17126
While deep learning is transforming data analysis in high-energy physics, computational challenges limit its potential. We address these challenges in the context of collider physics by introducing EveNet, an event-level foundation model pretrained on 500 million simulated collision events using a hybrid objective of self-supervised learning and physics-informed supervision. By leveraging a shared particle-cloud representation, EveNet outperforms state-of-the-art baselines across diverse tasks, including searches for heavy resonances and exotic Higgs decays, and demonstrates exceptional data efficiency in low-statistics regimes. Crucially, we validate the transferability of the model to experimental data by rediscovering the $$\Upsilon$$ meson in CMS Open Data and show its capacity for precision physics through the robust extraction of quantum correlation observables stable against systematic uncertainties. These results indicate that EveNet can successfully encode the fundamental physical structure of particle interactions, which offers a unified and resource-efficient framework to accelerate discovery at current and future colliders.
EveNet was trained on 500 million simulated events — roughly as many as the Large Hadron Collider will produce in a few days of running at full luminosity.
The idea of a universal model that understands particle physics echoes the AI in Liu Cixin's "The Three-Body Problem," which models proton behavior to fool Earth's detectors.