Improved Isolation Variables (-10% BW reduction in charm)
Improved isolation variables following work done in b2noc group [1].
Expected significant reduction in BandWidth:
- -64% for rare_charm_lines
- -44% for d0_to_hhpi0
Affecting 8+4 charm sub-modules saving extra_outputs in the old way (RAREandNEUTRAL) and now for the first time (DetectionAsymmetry), respectively.
Impact of changes in RAREandNEUTRAL lines. Tested on 1e4 evts on the full charm module with changes from charm_2025_reopt(@27/03/2025) for the RAREandNEUTRAL lines already implemented before.
- BEFORE: BW = 1290 kHz * 4.6 MB /1e4 evts = 0.59 GB/s
- AFTER: BW = 1290 kHz * 4.1 MB /1e4 evts = 0.53 GB/s --> ~11% BW reduction
Impact of changes in DetectionAsymmetry lines. Tested on 1e4 evts on the full charm module with changes from charm_2025_reopt(@27/03/2025) for the DetectionAsymmetry lines already implemented before and RAREandNEUTRAL after all the improvements.
- BEFORE: BW = 1290 kHz * 4.0 MB /1e4 evts = 0.52 GB/s
- AFTER: BW = 1290 kHz * 4.0 MB /1e4 evts = 0.52 GB/s --> <1% BW increase
Overall impact of the changes of this MR --> ~10% BW reduction
Tested tupling with DaVinci:
- rare_charm_lines
- d0_to_hhpi0
Comparison with Run2 variables: ongoing but no showstopper
Dependencies: