Skip to content

Draft: Single prong b to s tau tau lines

Alex Chen Hu requested to merge ic_tau_one_prong into rd-devel-march25

As of Friday 21/03 afternoon, this MR is ready. As discussed with Fionn, we will be working on Saturday to try to lower the bandwidths even more without a big reduction in our signal efficiency.

This MR aims to add new lines for b -> s tau tau decays (with one prong tau decays), and to fine-tune existing one-prong tau decays.

  1. New lines for b -> s tau tau decays, with one-prong tau decays.
  • Currently, only leptonic one-prong tau decays are available. This MR adds decays with tau -> pi (pi0) (pi0) nu, combined with another tau decay to pi (pi0 pi0) or to a lepton.
  • Some missing same-sign lines are added.

  1. Fine-tuning of existing b -> s tau tau with one-prong tau decay is investigated, in particular through the inclusion of a cut on the combined mass of the dihadron and one of the products of the tau decay.

  2. Increase the background prescales which were heavily reduced in a previous MR. To keep the bandwidth reasonable, this has led to the following changes

  • Removal of B+ -> K+ tau tau HLT2 lines
  • Removal of Bs -> Kst Kst tau tau HLT2 lines in most of the files
  • Removal of LFV tau decays with leptonic tau decays

Items to be completed:

  • Refactoring
  • Rebasing
  • Get rates / bandwidth
  • Investigate more mass cuts
  • Fine-tuning
  • Used ParticleContainerMerger to combine some of the lines together
  • Match prescale change in !4531 for spruce lines

Minimum bias rates tests

Tests were at first carried using the x86_64_v2-el9-gcc13+detdesc-opt build and the HltEfficiencyChecker “wizard” script (DaVinci/HltEfficiencyChecker/scripts/hlt_eff_checker.py) running on expected_2024_min_bias_hlt1_filtered_v2. However, large differences in the rates were observed when switching to the snippet provided in the main RD MR: with the former, the rates between master and this MR agree for the leptonic tau decays, whereas with the latter the rates from the MR are 4 times as large as the ones in master. This is in the process of being corrected.

The bandwidth is calculated as follows: 1289.692 kHz (input rate) x file size in MB / 100000 (number of events we run over).

LFV

Removing the lines in rd.btosmutau_tau_to_e_exclusive_hlt2, rd.btosetau_tau_to_e_exclusive_hlt2, rd.btosetau_exclusive_hlt2, and rd.btosmutau_exclusive_hlt2.

The decrease in BW with respect to the RD build in master is 106.6 KiB (109,138) -> bandwidth of 0.001 GB/s.

Leptonic tau lines

Branch File size (KiB) File Size (MB) Bandwidth (GB/s)
This MR 850.7 KiB 0.8711168 MB 0.01123472368
master 396.8 KiB 0.4063232 MB 0.0052403178
rd-devel-march25 397.2 KiB 0.4067328 MB 0.00524560038

The increase with respect to rd-devel-march25 (mainly due to putting the prescales back at a reasonable level) is 0.0059891233 GB/s

Pionic tau lines

Branch File size (KiB) File Size (MB) Bandwidth (GB/s)
This MR 666.1 KiB 0.6820864 MB 0.00879681373
No corresponding lines in master. New lines 😄
Edited by Alex Chen Hu

Merge request reports