The Large Hadron Collider (LHC) at CERN is the highest energy collider ever constructed. It consists of two counter-circulating proton beams made to interact in four locations around a 27 kilometer ring straddling the border between Switzerland and France. It is by some measures the largest man-made scientific device on the planet. The goal of the LHC is to probe the basic building blocks of matter and their interactions. For example, in 2012, the Higgs boson was discovered by the CMS and ATLAS collaborations.
The LHC collides proton beams at the center of our detectors. By measuring the energy and momentum of the escaping particles, we infer the existence of massive particles that were created in the collisions and measure the massive particles’ properties based on their decay products. The determination of the trajectories of charged particles ("tracks") from a set of positions of energy deposits from the various layers in our detector ("hits") plays a key role in identifying particles and measuring their charge and momentum. This pattern recognition problem—known as "track reconstruction" or simply "tracking"—is as a whole the most computationally complex and time-consuming step in the measurement process, as well as the most sensitive to increased activity in the detector, and traditionally, the least amenable to parallelized processing.
This project aims to develop tracking algorithms based on the Kalman Filter for use in a collider experiment that are fully vectorized and parallelized. These will be usable with parallel processor architectures such as Intel's Xeon Phi and GPUs, but yet maintain and extend the physics performance required for the challenges for the High Luminosity LHC (HL-LHC) planned for the 2020s.
The project also provides training for young researchers through a dedicated school on tools, techniques and methods for Computational and Data Science for High Energy Physics (CoDaS-HEP). The CoDaS-HEP school provides a broad introduction to these critical skills as well as an overview of applications High Energy Physics. Specific topics to be covered at the school include: Parallel Programming, Big Data Tools and Techniques, Machine Learning Technology and Methods as well as practical skills. The first CoDaS-HEP school took place on 10-13 July, 2017, at Princeton University. The second CoDaS-HEP school is being planned for 2018.