At the Large Hadron Collider, protons collide, and from the flash dozens of particles are born. Among them may be traces of rare processes — for example, the decay of a Standard Model Higgs boson into two W bosons. But these particles fly apart so quickly that their tracks merge into a single wide jet. Picking out such a jet among thousands of similar ones is like finding the right branch in a noisy forest.
The new algorithm, named PaRT, acts like an attentive observer. It looks at each particle inside the jet and decides for itself which ones matter more. This approach is called the attention mechanism: it underpins modern neural networks. PaRT was trained on many examples: ordinary jets and jets from decays of different particles. For the first time, an algorithm reliably recognizes jets from a Higgs boson that decayed into W bosons.
The physicists tested the algorithm on data collected by CMS over several years of operation. It turned out that in real collisions it performs just as well as in simulations, with a correction factor of about 0.9–1.0. This means it can be trusted. Now PaRT will help measure the Higgs boson's properties more precisely and search for new particles that decay in similar ways.
The forest of particles becomes clearer when you have a tool that can pick out what matters. PaRT is such a tool: it doesn't just see jets: it understands what they are made of.
as written by the authors arXiv:2604.09809
A novel deep neural network classifier, a ``Particle transformer'' (PaRT), is introduced for the identification of highly Lorentz-boosted resonances reconstructed as single, multipronged jets in measurements and searches performed by the CMS Collaboration at the CERN LHC. Based on a self-attention mechanism that allows the model to weigh the importance of different particles, PaRT is trained on a wide variety of topologies, notably demonstrating strong performance for the first time on jets originating from boosted Higgs boson decays to W bosons. The PaRT algorithm achieves a tagging efficiency of more than 50\% for such jets at a background efficiency of 1%, while maintaining decorrelation from the jet mass. A calibration is performed in proton-proton collision data collected by CMS at a center-of-mass energy of 13 TeV, with a data set corresponding to a total luminosity of 138 fb$$^{-1}$$. Data-to-simulation selection efficiency scale factors are measured to be in the 0.9$$-$$1.0 range, with relative uncertainties between 7 and 23%. The tagging capability of PaRT enhances the sensitivity of standard model measurements and searches for beyond-the-standard-model resonances decaying to hadronic diboson systems.
The jet from a fast Higgs boson is so narrow that all its particles fly within a cone smaller than one degree, like hitting a coin placed in the goal from the other end of the field with a football.
In the series 'Star Trek', the ship's sensors instantly identify particle types and their energy — PaRT takes a step toward that level of precision, but for an earthly collider.