Allen merge requestshttps://gitlab.cern.ch/lhcb/Allen/-/merge_requests2024-03-21T18:04:20+01:00https://gitlab.cern.ch/lhcb/Allen/-/merge_requests/1496Closes Allen ci cleanup issue https://gitlab.cern.ch/lhcb/Allen/-/issues/5182024-03-21T18:04:20+01:00Saverio MarianiCloses Allen ci cleanup issue https://gitlab.cern.ch/lhcb/Allen/-/issues/518Third part of the Allen ci cleanup, to hopefully speed-up testing and ease merging. This MR removes tests on outdated MC samples, except a throughput and an efficiency one on minbias_v5 and moves testing to expected 2024 samples.
~~Bef...Third part of the Allen ci cleanup, to hopefully speed-up testing and ease merging. This MR removes tests on outdated MC samples, except a throughput and an efficiency one on minbias_v5 and moves testing to expected 2024 samples.
~~Before updating references and moving on, would like @cagapopo or @dovombru to review this pls~~
As this only touches the Allen ci, there is no need of integration ci
Closes https://gitlab.cern.ch/lhcb/Allen/-/issues/518Andre GuntherAndre Guntherhttps://gitlab.cern.ch/lhcb/Allen/-/merge_requests/1491Enable SMOG2 flag in all sequences2024-03-22T16:00:45+01:00Saverio MarianiEnable SMOG2 flag in all sequencesAs also mentioned in https://gitlab.cern.ch/lhcb/Allen/-/issues/518, `withSMOG2 flag` is required to be enabled by default for all physics sequences, so that change of TCK is not needed when we decide to inject gas. Reference changes are...As also mentioned in https://gitlab.cern.ch/lhcb/Allen/-/issues/518, `withSMOG2 flag` is required to be enabled by default for all physics sequences, so that change of TCK is not needed when we decide to inject gas. Reference changes are expected due to the addition of the SMOG2 lines.
As a consequence, some tests from the Allen ci are removed, which also contributed in speeding up the testing sequence.
cc @oboenteg to take it into account when testing downstream
cc @cagapopoRTA/2024.03.27Christina AgapopoulouAndre GuntherYingao TangChristina Agapopoulouhttps://gitlab.cern.ch/lhcb/Allen/-/merge_requests/1468Removes MuonFiltered algorithms that were never used2024-03-18T07:42:09+01:00Adrian Casais VidalRemoves MuonFiltered algorithms that were never usedThe code removed was never used. These algorithms were written copy pasting from FilterTracks and VertexFitter kernel source files. The purpose was to reconfigure them to run on a different set of events. This was a dirty solution before...The code removed was never used. These algorithms were written copy pasting from FilterTracks and VertexFitter kernel source files. The purpose was to reconfigure them to run on a different set of events. This was a dirty solution before the current MultiEventScheduler was release.
Turns out this code is useless and should be removed from the codebase.https://gitlab.cern.ch/lhcb/Allen/-/merge_requests/1459PV reconstruction : Homogenizing Allen and Moore PV finders. Change of rho cu...2024-03-22T16:00:44+01:00Bogdan KutsenkoPV reconstruction : Homogenizing Allen and Moore PV finders. Change of rho cut and Kalman Filter covariance matrix initialisation- Rho cut (`Maximum distance squared of vertex to beam line`) is changed from 0.3f to 0.09f. The distribution on [2024 MC sample is attached](/uploads/e924b789dd705647575a79291302e454/Rho_distr_MC_2024.pdf)
- Kalman Filter covariance ma...- Rho cut (`Maximum distance squared of vertex to beam line`) is changed from 0.3f to 0.09f. The distribution on [2024 MC sample is attached](/uploads/e924b789dd705647575a79291302e454/Rho_distr_MC_2024.pdf)
- Kalman Filter covariance matrix initialization is changed. Here are two motivations for that:
1) To match the initialisation in REC : lhcb/Rec!3464
2) New initialisation shows slightly better performance in PV efficiencies than the current default based on 2024 MC sample :
| | Default covariance matrix initialisation Velo::Tracking::param\_w\_inverted , 0.f, 0.01f Velo::Tracking::param\_w\_inverted, 0.f, 0.01f | | Initialize the covariance matrix with 100.0f, 0.f, 0.01f, 100.0f, 0.f, 0.01f | |
| --- | --- | --- | --- | --- |
| | Efficiencies, % | False PVs, % | Efficiencies, % | False PVs, % |
| all | 93.91 +- 0.03 | 1.94 +- 0.02 | 94.06 +- 0.03 | 1.68 +- 0.02 |
| isolated | 97.48 +- 0.05 | 1.06 +- 0.04 | 97.63 +- 0.05 | 0.95 +- 0.04 |
| close | 89.92 +- 0.05 | 2.99 +- 0.04 | 90.08 +- 0.05 | 2.56 +- 0.04 |
| ntracks<10 | 67.98 +- 0.23 | 21.35 +- 0.22 | 68.74 +- 0.23 | 18.89 +- 0.23 |
| ntracks>=10 | 96.82 +- 0.03 | 0.00 +- 0.02 | 96.90 +- 0.03 | 0.00 +- 0.02 |
| z<=-50.0 | 95.02 +- 0.13 | 1.72 +- 0.10 | 95.20 +- 0.13 | 1.34 +- 0.10 |
| z in (-50.0,50.0) | 92.76 +- 0.04 | 2.02 +- 0.03 | 92.92 +- 0.04 | 1.88 +- 0.03 |
| z>=-50.0 | 96.08 +- 0.13 | 1.96 +- 0.09 | 96.17 +- 0.13 | 1.50 +- 0.09 |
| decayBeauty | 98.51 +- 2.10 | 1.83 +- 0.61 | 98.51 +- 2.10 | 2.65 +- 0.66 |
| decayCharm | 98.22 +- 0.19 | 1.47 +- 0.12 | 98.36 +- 0.19 | 1.12 +- 0.12 |
| decayStrange | 94.10 +- 0.03 | 1.65 +- 0.02 | 94.25 +- 0.03 | 1.47 +- 0.02 |
| other | 52.17 +- 4.90 | 13.64 +- 0.92 | 52.17 +- 4.90 | 11.34 +- 1.04 |
| 1MCPV | 99.70 +- 0.14 | 1.39 +- 0.10 | 99.80 +- 0.14 | 1.09 +- 0.10 |
| 2MCPV | 98.16 +- 0.14 | 1.24 +- 0.10 | 98.16 +- 0.14 | 1.03 +- 0.10 |
| 3MCPV | 95.74 +- 0.15 | 2.12 +- 0.11 | 95.74 +- 0.15 | 1.57 +- 0.11 |
| 4MCPV | 93.63 +- 0.17 | 1.47 +- 0.13 | 93.89 +- 0.17 | 1.73 +- 0.13 |
| 5MCPV | 92.62 +- 0.22 | 2.08 +- 0.17 | 92.79 +- 0.22 | 1.74 +- 0.17 |RTA/2024.03.27Andre GuntherYingao TangAndre Guntherhttps://gitlab.cern.ch/lhcb/Allen/-/merge_requests/1404Remove obsolete options files2024-02-12T11:49:40+01:00Roel AaijRemove obsolete options filescloses #276closes #276Christina AgapopoulouChristina Agapopoulouhttps://gitlab.cern.ch/lhcb/Allen/-/merge_requests/1396Drop FittedForward track type and use FitHistory instead2024-02-09T11:22:09+01:00Andre GuntherDrop FittedForward track type and use FitHistory insteadGoes with Rec!3722 and LHCb!4419Goes with Rec!3722 and LHCb!4419Andre GuntherAndre Guntherhttps://gitlab.cern.ch/lhcb/Allen/-/merge_requests/1391Use ODIN standalone from DAQEvent instead of including it in Allen2024-02-09T11:22:09+01:00Roel AaijUse ODIN standalone from DAQEvent instead of including it in AllenAdresses https://gitlab.cern.ch/lhcb/Allen/-/issues/449
Depends on https://gitlab.cern.ch/lhcb/LHCb/-/merge_requests/4412Adresses https://gitlab.cern.ch/lhcb/Allen/-/issues/449
Depends on https://gitlab.cern.ch/lhcb/LHCb/-/merge_requests/4412https://gitlab.cern.ch/lhcb/Allen/-/merge_requests/1386remove spurious dependencies on DAQKernel2023-12-23T16:28:21+01:00Gerhard Ravenremove spurious dependencies on DAQKernelRequired for LHCb!4399 which removes the DAQKernel package, but can be applied by itself, as nothing in Allen actually uses DAQKernel...Required for LHCb!4399 which removes the DAQKernel package, but can be applied by itself, as nothing in Allen actually uses DAQKernel...https://gitlab.cern.ch/lhcb/Allen/-/merge_requests/1362Fixes for gcc 13, clang 16 and C++202023-12-16T11:53:04+01:00Marco Clemencicmarco.clemencic@cern.chFixes for gcc 13, clang 16 and C++20- Avoid use of deprecated `is_pod`
- Fix implicit capture of this (deprecated in C++20)
- Remove unused variables
Note that implicit capture of `this` via `[=]` causes a warning in C++20, but the new way `[=,this]` causes a warning in C...- Avoid use of deprecated `is_pod`
- Fix implicit capture of this (deprecated in C++20)
- Remove unused variables
Note that implicit capture of `this` via `[=]` causes a warning in C++20, but the new way `[=,this]` causes a warning in C++17, so we hide the C++17 warning (at some point we will stop supporting C++17).Sebastien PonceSebastien Poncehttps://gitlab.cern.ch/lhcb/Allen/-/merge_requests/1358Fix standalone builds of public headers2023-12-12T15:23:34+01:00Marco Clemencicmarco.clemencic@cern.chFix standalone builds of public headersGaudi CMake provides a mechanism to test the build of public headers in standalone mode, meant to ensure that the code doesn't have hidden dependencies on the particular context in which it is used. This test builds have been running for...Gaudi CMake provides a mechanism to test the build of public headers in standalone mode, meant to ensure that the code doesn't have hidden dependencies on the particular context in which it is used. This test builds have been running for a while in ~"lhcb-gaudi-head" (see lhcb-core/LHCbNightlyConf!1027) where problems were sorted out with the goal of enabling the test builds in `lhcb-master`.
The header `PyAllenHelper.h` fails compilation in standalone mode because it requires a header that has a different path at build time than when installed (during build we use `InputProvider.h`, but it gets installed as `Allen/InputProvider.h`).
The problem can be fixed in a number ways:
- make Allen public headers accessible with the same name at build time and when installed *(my favorite option)*
- install `PyAllenHelper.h` outside of `gaudi_add_library` so that it is not seen as a public header
- skip all public headers standalone test builds for BinaryDumpers
- since `PyAllenHelper.h` is only needed in `allen.py`, replace the public header as text in `allen.py` *(the option implemented in this MR)*
and probably I can come up with another few options.
In the attempt to clean up ~"lhcb-gaudi-head" I decided to give this a shot, but I'd prefer that Allen maintainers decide which approach they want before we enable standalone builds of public headers in `lhcb-master`.https://gitlab.cern.ch/lhcb/Allen/-/merge_requests/1348Fix CI job for formatting2023-10-18T21:48:47+02:00Rosen MatevFix CI job for formattingRosen MatevRosen Matevhttps://gitlab.cern.ch/lhcb/Allen/-/merge_requests/1345PbPb BGI lines and (disabled) particle based lines for alignment+monitoring2023-12-20T09:52:56+01:00Christina AgapopoulouPbPb BGI lines and (disabled) particle based lines for alignment+monitoring- Fix `reco_particles` flag that enables track/muon alignment and a few other key physics lines (Ks2PiPi, D2hh, Lambda2PPi) for the PbPb sequences.
- For the mini sequence, only the alignment lines are turned on for events below the UP...- Fix `reco_particles` flag that enables track/muon alignment and a few other key physics lines (Ks2PiPi, D2hh, Lambda2PPi) for the PbPb sequences.
- For the mini sequence, only the alignment lines are turned on for events below the UPC GEC.
- Right now the default for PbPb sequences is `reco_particles=False` until https://gitlab.cern.ch/lhcb/Allen/-/merge_requests/1350 is in.
- Enables by default BGI lines that will be used during VDM/BGI data taking periods (including to measure ghost charge for ATLAS/CMS in parallel to taking physics data for us).
- Tweak minimum velo clusters (2->5) for activity line to take into account the higher baseline of clusters (due to background in PbPb conditions).
- First prescaled to 0 and then completely removed Hlt1BGIPlumeActivity line since the Plume v2 decoding is not ready.
- Prescale Hlt1VeloClustersMicroBias by 0.01 for PbPb (to be picked up after the VDM scans).
- Removed the Hlt1ODINNoBias line that was selecting every bb crossing.
- Cleanup from the summer:
- Adapted cuts for activity-based BGI lines (as already used in the September VDM scans).
- Picked up some line prescales for the pp sequence as already deployed online, see https://gitlab.cern.ch/lhcb/Allen/-/merge_requests/1316.
@rmatev @baudurie @samarian @mfontanaRosen MatevRosen Matevhttps://gitlab.cern.ch/lhcb/Allen/-/merge_requests/1316Draft: Remove HLT1LowPtMuon and prescale lines for pp sequence2024-01-29T18:22:00+01:00Rosen MatevDraft: Remove HLT1LowPtMuon and prescale lines for pp sequenceThis applies the changes that were done for the majority of the 2023 pp data. See https://lblogbook.cern.ch/HLT/292
TODO
- [ ] make sure that the velo_mapping sequence is not affected (lines should not be prescaled there)This applies the changes that were done for the majority of the 2023 pp data. See https://lblogbook.cern.ch/HLT/292
TODO
- [ ] make sure that the velo_mapping sequence is not affected (lines should not be prescaled there)Andrew George MorrisAndrew George Morrishttps://gitlab.cern.ch/lhcb/Allen/-/merge_requests/1312Protect makeTriplets against division by zero2023-09-04T13:30:17+02:00Christina AgapopoulouProtect makeTriplets against division by zeroA division by zero was caught in https://gitlab.cern.ch/lhcb/Allen/-/blob/master/device/SciFi/hybridseeding/src/seed_xz.cu#L159 in tests on data with a `ubsan` build. This MR protects against divisions by zero in the `makeTriplets` algor...A division by zero was caught in https://gitlab.cern.ch/lhcb/Allen/-/blob/master/device/SciFi/hybridseeding/src/seed_xz.cu#L159 in tests on data with a `ubsan` build. This MR protects against divisions by zero in the `makeTriplets` algorithm. Repeating locally the [tests](https://lhcb-nightlies.web.cern.ch/nightly/lhcb-sanitizers/1071/MooreOnline/x86_64_v2-centos7-gcc12-dbg+ubsan/tests#AllenOnline_mep_real_data), no other zero-divison errors appear.
Closes issue https://gitlab.cern.ch/lhcb/MooreOnline/-/issues/29
@lohenry @rmatevRosen MatevRosen Matevhttps://gitlab.cern.ch/lhcb/Allen/-/merge_requests/1296Simplify lines by removing the input_size function and use offset instead.2023-08-14T10:26:55+02:00Daniel Campora PerezSimplify lines by removing the input_size function and use offset instead.This MR simplifies some lines by removing the need to define `input_size`, reusing `offset` for that purpose instead.This MR simplifies some lines by removing the need to define `input_size`, reusing `offset` for that purpose instead.Rosen MatevRosen Matevhttps://gitlab.cern.ch/lhcb/Allen/-/merge_requests/1291Adapted to drop of GaudiAlgorithm in FT2023-10-03T22:33:00+02:00Sebastien PonceAdapted to drop of GaudiAlgorithm in FTGoes together with LHCb!4233Goes together with LHCb!4233https://gitlab.cern.ch/lhcb/Allen/-/merge_requests/1251Use MULTI_ALLOC engine for sanitizer binary tags2023-06-28T18:42:43+02:00Rosen MatevUse MULTI_ALLOC engine for sanitizer binary tagsRosen MatevRosen Matevhttps://gitlab.cern.ch/lhcb/Allen/-/merge_requests/1218One more fix in preparation for GaudiAlg refactoring2023-06-26T07:38:19+02:00Marco Clemencicmarco.clemencic@cern.chOne more fix in preparation for GaudiAlg refactoringThis fixes an issue after a number of rebases of !1193This fixes an issue after a number of rebases of !1193https://gitlab.cern.ch/lhcb/Allen/-/merge_requests/1203Update docs and diag_suppress preprocessor directives2023-05-15T20:04:23+02:00Da Yu TouUpdate docs and diag_suppress preprocessor directives- Updated preprocessor directives to remove diag_suppress deprecation warning during compilation with CUDA v11.5.
- Updated docs for CUDA builds outside CERN network.- Updated preprocessor directives to remove diag_suppress deprecation warning during compilation with CUDA v11.5.
- Updated docs for CUDA builds outside CERN network.Roel AaijRoel Aaijhttps://gitlab.cern.ch/lhcb/Allen/-/merge_requests/1189Wait for popen process to avoid zombies2023-05-15T20:04:23+02:00Rosen MatevWait for popen process to avoid zombiesDaniel Campora PerezDaniel Campora Perez