Berkeley Lab physicist Ben Nachman and Chamberlain Fellow Mariel Pettee, with colleagues in the ATLAS Collaboration, have released a new result that uses machine learning to measure particle collision reaction rates in many parameters all at once. In particular, this is the first such measurement to be published that is both unbinned and high-dimensional, in an interactive public access format, enabling a wide range of scientific and educational applications.

Using traditional techniques, it has been possible to make precise measurements of only one or two collision properties at a time, because measurements of many quantities at one time require exorbitantly large amounts of data to process. With the machine learning tool OmniFold, also co-developed by Berkeley Lab researchers, the team has been able to measure 24 simultaneous quantities – i.e. a measurement in 24 dimensions – and then they developed a new scheme, Z-jets Omnifold 2024, for publishing the complex result using modern data science tools, so that users can interact with the data directly.

The magnitude of information contained in this result allows researchers to make unprecedented particle physics measurements and explore lesser-known regimes of the Standard Model.

Learn more: