Draft: B2OC: add 2 TurCal lines

• add 2 TurCal lines with persistentreco=true and raw_banks=DETECTOR_RAW_BANK_TYPES:
  • Hlt2Calib_BdToDmH_DmToPimPimKp
  • Hlt2Calib_BdToDsmH_DsmToKpKmPim

purpose: provide clean Bd2D(2KPiPi)H and Bs2Ds(2KKPi)H samples for dedicated offline calibrations and new algorithms developement

the Hlt2 output level S/B ratio for the first line is shown by the plot below (from Collision23)

• set the DOCA cut for the 3 body builder to 0.2 mm

patch generated by https://gitlab.cern.ch/lhcb/Moore/-/jobs/32982407

assigned to @abertoli

added RTA label

added 1 commit

• 09dc8e6b - add 2- and 3-body topo. in hlt2_b2oc_all_lines.py

Compare with previous version

added 19 commits

• 09dc8e6b...3bd314f8 - 18 commits from branch master
• 546db6cc - Merge branch 'master' into b2oc_upgrade-abertoli

Compare with previous version

added 1 commit

• 3e095d77 - Fixed formatting

Compare with previous version

added 1 commit

• 18a60b60 - update Hlt/Hlt2Conf/options/examples/b_to_open_charm/hlt2_b2oc_all_lines.py

Compare with previous version

added 1 commit

• 2186dd3d - introduce Hlt2Calib_BdToDmH_DmToPimPimKp and Hlt2Calib_BdToDsmH_DsmToKpKmPim going to Turcal stream

Compare with previous version

changed the description

added 1 commit

• 61d40d17 - Fixed formatting

Compare with previous version

let me ping on this:

• @nskidmor, for the Spruce closure test
• @gligorov @cmarinbe @decianm for the usage of the Turcal stream

@gligorov if I forgot to ping somebody please help me

added PR/Moore_hlt2_bandwidth PR/Moore_spruce_bandwidth labels

removed PR/Moore_spruce_bandwidth label

@mzdybal even if the code is not completely ironed out can I please have a round of CI test here ? I have added the label PR/Moore_hlt2_bandwidth that will provide me a lot of useful informations, many thanks !

added ci-test-triggered label

changed the description

removed ci-test-triggered label

• [2023-10-12 14:56] Validation started with lhcb-master-mr#9440
• [2023-10-23 15:03] Validation started with lhcb-master-mr#9528
• [2023-11-03 09:43] Validation started with lhcb-master-mr#9596
• [2023-11-06 16:42] Validation started with lhcb-master-mr#9619

changed the description

some comments on the numbers:

• the total retention of these Calib lines is small, 1.1 % or 0.5 %, these must correspond only to a few candidates found in the whole min bias sample
• the avg total event size is larger for the Calib lines so indeed it looks like some additional output is saved :-) (273-203)/203 = + 34 %
• the overlap between the two Calib lines is not large so let's say that the extra GB/s is 0.009117+0.003602, assuming 200 days and converting to TB this means (0.009117+0.003602)*200*24*3600/1000 = 220 TB
• let me add that the rates for these Calib lines can be reduced further, by a factor up to 1/2 I am guessing

@gligorov can you please remind me here what footprint on disk you had in mind ?

@poluekt @mvesteri

@poluekt @mvesteri this is the MR I mentioned at the WP 3 Selections meeting

• the basic functionality is there
• the code needs a bit or ironing but retentions will not change due to that
• the footprint on disk need to be defined and the selections updated accordingly but as Bd2DPi and Bs2DsPi/K are rather known lines the update of the selection should not be a major issue
• last but not least the streaming of the new lines should be done properly (Turcal stream) while right now there is no streaming at all

let me copy/paste here the retentions obtained from the .log file of the BW test:

LAZY_AND: Hlt2Calib_BdToDmH_DmToPimPimKp                                                                           #=100000  Sum=11          Eff=|(0.01100000 +- 0.00331644)%|
LAZY_AND: Hlt2Calib_BdToDsmH_DsmToKpKmPim                                                                          #=100000  Sum=5           Eff=|(0.005000000 +- 0.00223601)%|

so the number of candidates left out of 1e+5 min bias events is 11 and 5, respectively

resolved all threads

changed the description

added 1 commit

• 3eddbf72 - adjust F.SDOCA cut in the basic 3-body builder (from 0.5 to 0.2)

Compare with previous version

added 1 commit

• 557bb5c1 - Fixed formatting

Compare with previous version

added PR/Moore_spruce_bandwidth label

@mzdybal can we please run another round of CI tests on this please ?

added ci-test-triggered label

added 1 commit

• 1a091b8a - iron out the Hlt2 and Spruce code for the 2 new Hlt2 and 2 new Spruce calibration lines

Compare with previous version

requested review from @shunan

hello @shunan, so far this MR is:

• adding two B2OC calibration lines in Hlt2 and two corresponding calibration lines in Spruce
• adjusting the F.SDOCA max cut for the 3-body builder to 0.2 mm (was 0.5 mm)

next steps:

• I think we should also be able to tune F.SDOCA max for the 2-body builder by the next 2 weeks
• the appropriate streaming of the Calib lines will be implemented after the presentation at the WP 4 meeting of Nov. 2ond at 15:00

added 1 commit

• b8c63268 - Fixed formatting

Compare with previous version

These calibration lines will also be interesting for FT calibration and future developments (inclusive tagger) and studies.

mentioned in issue #664

Considering that currently only 10% of the TURCAL is on disk with the raw event, is it not more useful to just have lines with a restricted mass window in the turbo stream, with persist reco & the raw event persisted? Naively, I would say you have still a lot of room to make a cleaner selection, such that bandwidth shouldn't be an issue...

added 1 commit

• 77a36820 - test reduction cutting on DOCA for the 2-body builder (0.5 -> 0.2)

Compare with previous version

added 1 commit

• d8be3079 - Fixed formatting

Compare with previous version

added 1 commit

• d23e2d34 - remove SDOCA cut on K0SPi0 builder (builders/d_builder.py), use a list for...

Compare with previous version

resolved all threads

@jagoodin can we please please please run another round of CI in this branch ? thanks !

I need to grab some numbers from the output, meanwhile I will start to stream the Calib lines properly

resolved all threads

added 1 commit

• 537fa567 - updated to __init__.py hlt2_b2oc.py and spruce_b2oc.py

Compare with previous version

added 1 commit

• 9af89606 - stream properly the B2OC calib/turcal lines

Compare with previous version

added 1 commit

• e65679cd - typo in spruce_b2oc.py

Compare with previous version

@jagoodin can we please have the hopefully final round of CI in this branch ? many thanks !

/ci-test

@lgrillo @mfontana can I have your explicit green light on this following my recent presentation at the RTA WP4/5 meeting ?

resolved all threads