Track reconstruction challenge at the HL-LHC

• Andreas Salzburger (CERN)

The upgrade of the Large Hadron Collider to the HL-LHC will bring unprecedented challenging experimental conditions: the increased event pile-up will lead to event complexities in term of particle multiplicities that have never been seen before. While the detectors are prepared, upgraded and - in the case of the Inner Tracking devices mostly fully replaced by new systems - also the event reconstruction has to be prepared accordingly. Due to the combinatorial complexity track reconstruction is arguable the most compute intense and certainly amongst the most complex problems to solve. This upgrade of the track reconstruction software comes at an interesting point of time: the rise of artificial intelligence in almost all areas of life (including and spearheading in science) brought new, alternative concepts to track reconstruction. At the same time, and partly catalyzed by the AI revolution, the extended availability of accelerated compute hardware brings both a chance to new resources on the one hand, but also raises the problem of supported heterogenous computing approaches. And finally, the timescale of the HL-LHC era comes with the requirement to establish sustainable and supported software for the next two decades. We present the algorithmic, physics and logistic challenges of track reconstruction for the HL-LHC era and beyond, together with our attempt to establish a community driven software project that aims to bind the aforementioned aspects under one common umbrella.