Presentation on Mask Clustering for Velo by Daniel Campora
Need to define clustering efficiency
Standard clustering algorithm for Velo finds all clusters
Other algorithms:
Should find the same clusters as baseline algorithm
In addition compare distance to center of true cluster
Or compare all LHCb IDs of the clusters
Presentation on EB options by Rainer Schwemmer
Different options: normal density nodes / high density nodes / very high density nodes
Depending on chosen option, accelerators fit in EB nodes or in filter farm nodes or in separate box in filter farm
Presentation by Dorothea vom Bruch on cuda_hlt repository
Place to collect all GPU code and connect different pieces
Ongoing work on PrChecker implementation: should be part of main framework
The final output should include p_T, chi2 and other quantities used for the selection + the LHCb IDs of all objects used for selection (to be turned on / off by compiler flag depending on type of running: checking / performance)