IRIS-HEP Fellow: Rohith Karur
Fellowship dates: Jun – Sep, 2021
Home Institution: UC Berkeley/LBNL
Project: Implement hashing-based particle track reconstruction in ACTSWe will work on implementing similarity hashing techniques using the Approximate Nearest Neighbors (ANN) search method using C++ and Python into the ACTS project at CERN. The minimization of search complexity in identifying track hits is a tool which will be invaluable to data collection at the HL-LHC. After identifying particle track clusters using the ANN method, we will then use existing Kalman Filters to focus on these clusters to comprehensively perform track reconstruction. We will then plan on tuning parameters to optimize both the complexity of the algorithm as well as the track reconstruction performance, and maximizing the extent to which our code can be parallelized. We will also implement an extension to this project in which the track reconstruction is performed with neural networks instead of Kalman filter.
More information: My project proposal
Louis-Guillaume Gagnon (UC Berkeley)
Heather Gray (UC Berkeley, LBNL)
- 18 Oct 2021 - "ANNs for ACTS", Rohith Karur, Recording: ANNs for ACTS