athena merge requestshttps://gitlab.cern.ch/atlas/athena/-/merge_requests2023-05-23T15:42:18+02:00https://gitlab.cern.ch/atlas/athena/-/merge_requests/63226Revert "Merge branch 'MemoryLeak' into 'master'"2023-05-23T15:42:18+02:00Carlo Varnicarlo.varni@cern.chRevert "Merge branch 'MemoryLeak' into 'master'"This reverts merge request !63210This reverts merge request !63210https://gitlab.cern.ch/atlas/athena/-/merge_requests/61424Allow for customization of dynamic attribute split-level2023-03-21T21:43:23+01:00Peter Van GemmerenAllow for customization of dynamic attribute split-levelIf needed, this MR will allow to set split-level for dynamic attributes.If needed, this MR will allow to set split-level for dynamic attributes.https://gitlab.cern.ch/atlas/athena/-/merge_requests/61141Pass context to ROBDataProviderSvc function2023-03-01T21:42:13+01:00Stewart Martin-HaughPass context to ROBDataProviderSvc functionPass EventContext down and prevent currentContext lookup.Pass EventContext down and prevent currentContext lookup.https://gitlab.cern.ch/atlas/athena/-/merge_requests/58950PowhegControl: adding feature for MadSpin decays - AGENE-21652022-12-03T03:45:06+01:00Timothee Theveneaux-PelzerPowhegControl: adding feature for MadSpin decays - AGENE-2165This MR does a few modifications in PowhegControl related to MadSpin, as mentioned in AGENE-2165.
If the decays are done with MadSpin, using a model that contains "ckm" or "CKM" in its name, then the decay branching ratios for top and W...This MR does a few modifications in PowhegControl related to MadSpin, as mentioned in AGENE-2165.
If the decays are done with MadSpin, using a model that contains "ckm" or "CKM" in its name, then the decay branching ratios for top and W are removed from the param_card before running MadSpin, so that MadSpin will recalculate them at LO using the model, possibly with special ckm matrix parameters.
If such a model is used, the Wolfenstein parameters block is written in the param_card. The four parameters can be set by the user in the joboption.
The default values are:
```
PowhegConfig.Wolfenstein_lambda = 2.257000e-01
PowhegConfig.Wolfenstein_A = 8.140000e-01
PowhegConfig.Wolfenstein_rho = 1.350000e-01
PowhegConfig.Wolfenstein_eta = 3.490000e-01
```
These correspond to the default values in the `sm-zeromass_ckm` model.
Finally, I removed the setting of widths and BRs for -6 and -24 in the param_card (whatever the model is), because MadSpin anyway always does the same for particle and anti-particle. So having two values for top and anti-top is confusing (even if we were always setting the same value), and useless.
@ewelina I will do another MR to master, for 22.6 - I think it's quicker than if I wait for the automatic sweep.
Tagging @dhirschhttps://gitlab.cern.ch/atlas/athena/-/merge_requests/58830Ensure that GenEvent barcodes attribute is filled after GenEvent copy constru...2022-12-01T00:24:18+01:00John Derek ChapmanEnsure that GenEvent barcodes attribute is filled after GenEvent copy constructor is called.This fixes MC Overlay in the `master--HepMC3` nightlies. It is urgently needed for the HepMC3 validation for MC23.
Fix for ATLASSIM-6152.
We are working on a follow up MR with a more (CPU) efficient implementation.
Tagging @averbyts, @...This fixes MC Overlay in the `master--HepMC3` nightlies. It is urgently needed for the HepMC3 validation for MC23.
Fix for ATLASSIM-6152.
We are working on a follow up MR with a more (CPU) efficient implementation.
Tagging @averbyts, @tsulaia, @tadej.https://gitlab.cern.ch/atlas/athena/-/merge_requests/57340remove trigID ART test, fsjet_actsvtx_pu552022-10-09T18:29:26+02:00Tim Adyeremove trigID ART test, fsjet_actsvtx_pu55Remove trigID ART test, fsjet_actsvtx_pu55. The acts vertexing is now default for all jobs, so we no longer need a special test.
This is the same as !54883 for 22.0.Remove trigID ART test, fsjet_actsvtx_pu55. The acts vertexing is now default for all jobs, so we no longer need a special test.
This is the same as !54883 for 22.0.https://gitlab.cern.ch/atlas/athena/-/merge_requests/56478CaloTrackingGeometryBuilder: remove remaining mutables2022-09-05T17:37:37+02:00Frank WinklmeierCaloTrackingGeometryBuilder: remove remaining mutables- Make local copies of property values to avoid `mutable`. Modifying
property values is bad to begin with.
- Make `bpCutouts` a local variable and pass it explicitly
- protect garbage bin with mutex and mark thread-safe
Once the `trac...- Make local copies of property values to avoid `mutable`. Modifying
property values is bad to begin with.
- Make `bpCutouts` a local variable and pass it explicitly
- protect garbage bin with mutex and mark thread-safe
Once the `trackingGeometry()` interface is declare thread-safe, this class should pass the thread-checker.
Relates to ATLASRECTS-7252.https://gitlab.cern.ch/atlas/athena/-/merge_requests/56013Fix wrongly implemented protection against divergence2022-08-18T21:43:08+02:00Vadim KostyukhinFix wrongly implemented protection against divergenceThis protection in some cases kills normal events in a wrong way producing a crash.This protection in some cases kills normal events in a wrong way producing a crash.https://gitlab.cern.ch/atlas/athena/-/merge_requests/53966continuous2D implementation in xAODBTaggingEfficiencyTool2022-08-20T12:15:48+02:00Ilaria Luisecontinuous2D implementation in xAODBTaggingEfficiencyToolRestructuring of the xAODBTaggingEfficiencyTool tools in order to:
* remove some string comparisons
* remove tag+cveto implementation and substitute with a real Continuous2D implementation
* implement the continuous2D in truth tagging
*...Restructuring of the xAODBTaggingEfficiencyTool tools in order to:
* remove some string comparisons
* remove tag+cveto implementation and substitute with a real Continuous2D implementation
* implement the continuous2D in truth tagging
* change the way the SFs are retrieved in truth tagging. Now they are retrieved only after the permutation has been chosen (more similar to the direct tagging)
* validated in the VHbb workflow for fixed cut/continuous and continuous2d separately running on ttbar pwpy8 before and after the changes. slides in preparation.
IMPORTANT: `useVetoWP = False` argument now replaced with `useCTag = False`. This affects only the analyses actually using the b-veto implementation.
tagging @nkakati @fdibello @vdao @biliu @yama @eschopf @hanar @robouquehttps://gitlab.cern.ch/atlas/athena/-/merge_requests/53307Fix to L1Topo multiplicity algorithms2022-06-06T15:35:27+02:00Paula Martinez SuarezFix to L1Topo multiplicity algorithms- Fix eta rounding to retrieve energy threshold in eEmMultiplicity and eTauMultiplicity
- Add iso cut to jTauMultiplicity
- Write template for jTauNoSort (iso cut can be added in the future)
ATR-24829
Tagging @orlando @iriu @asonay @jm...- Fix eta rounding to retrieve energy threshold in eEmMultiplicity and eTauMultiplicity
- Add iso cut to jTauMultiplicity
- Write template for jTauNoSort (iso cut can be added in the future)
ATR-24829
Tagging @orlando @iriu @asonay @jmharris.https://gitlab.cern.ch/atlas/athena/-/merge_requests/52601Express stream prescaling2022-05-20T18:12:09+02:00Mark StocktonExpress stream prescalingChanges to read and using the express stream prescale ATR-21324,ATR-20870
* duplicate name parameter removed from HLTPrescale json ATR-20877
* add express stream maps to HLTPrescaleSet DataStructure and functions to access
* includes ...Changes to read and using the express stream prescale ATR-21324,ATR-20870
* duplicate name parameter removed from HLTPrescale json ATR-20877
* add express stream maps to HLTPrescaleSet DataStructure and functions to access
* includes returning default if chain not in express
* Update PrescalingTool to read express prescale [see getPrescale](https://acode-browser1.usatlas.bnl.gov/lxr/source/athena/Trigger/TrigSteer/HLTSeeding/src/PrescalingTool.cxx#0137)
* Update StreamTagMakerTool to check express stream prescale for relevant chains
cc @rbielskihttps://gitlab.cern.ch/atlas/athena/-/merge_requests/51758Prevent config to schedule multiple InDetUsedInVertexFitTrackDecorator doing ...2022-04-01T18:48:47+02:00Pierre Antoine DelsartPrevent config to schedule multiple InDetUsedInVertexFitTrackDecorator doing the same thingThis is an attempt to fix
https://its.cern.ch/jira/browse/ATLASRECTS-6925
The configuration is changed in various places
- A config module dedicated to UsedInVertexFitTrackDecorato is introduce and allows to build unique alg names ac...This is an attempt to fix
https://its.cern.ch/jira/browse/ATLASRECTS-6925
The configuration is changed in various places
- A config module dedicated to UsedInVertexFitTrackDecorato is introduce and allows to build unique alg names according to trk,vtx and decoration (thus avoiding duplication)
- The jet config is made to use the above module and not it's own private alg (thus temporarily breaking compatibility with AnalysisBase release)
- Some run2 style remnants of pflow config are removed.
Tagging the experts I know on the jet & pflow side, don't hesitate to add more.
@mhodgkin @cdelitzs @khoo @sawyer @schaarscPierre Antoine DelsartPierre Antoine Delsarthttps://gitlab.cern.ch/atlas/athena/-/merge_requests/51535update Powheg version to 05-032022-03-24T09:56:48+01:00Ewelina Maria Lobodzinskaupdate Powheg version to 05-03update Powheg version to 05-03update Powheg version to 05-03https://gitlab.cern.ch/atlas/athena/-/merge_requests/51394PowhegControl: fixes for MadSpin errors, and for openloops V2 processes2022-04-06T13:29:34+02:00Timothee Theveneaux-PelzerPowhegControl: fixes for MadSpin errors, and for openloops V2 processesThis MR should fix the following things in PowhegControl:
- one MadSpin error message was missed in !50660, see AGENE-2132
- ttbarj should work: there were problems related to OpenLoops, which should now be solved; I also added the possi...This MR should fix the following things in PowhegControl:
- one MadSpin error message was missed in !50660, see AGENE-2132
- ttbarj should work: there were problems related to OpenLoops, which should now be solved; I also added the possibility to do decays with MadSpin - AGENE-1849
- fourtops should work, except the decays: one can still do the decays with MadSpin, or generate lhe files with undecayed tops; also I think there were issues with OpenLoops which would have made this process crash once deployed on cvmfs - AGENE-2128
- ttj_MiNNLO should work: similar problems related to OpenLoops (though a bit different because of the structure of the MakeFile for this process) should now be solved - AGENE-2055
For the last three things, this will require the latest afs test installation for these processes to be installed on cvmfs. Once this is done I'll make the relevant change in the cmake config file, then remove the Draft. This will also make the process Wt_DR_modified accessible on cvmfs, cf. AGENE-2131.
In addition, I now added two new possible parameters, `MadSpin_max_weight_ps_point` and `Nevents_for_max_weight`, which allow to set the corresponding two parameters in MadSpin run cards (e.g. `PowhegConfig.MadSpin_max_weight_ps_point=1000` in the joboption would be translated as `set max_weight_ps_point 1000` in the madspin card). This could help to improve the integration for the madspin decays when needed, as we do for MadGraph samples.
I just tested fourtops, ttj and ttj_MiNNLO using the newly deployed POWHEG 05-03, and they seem to work both on machines which have access to afs, and on machine which don't (this was a big problem before). So I'm removing the Draft and we can go ahead with this MR for the next release (and of course with !51535 which @ewelina just created).https://gitlab.cern.ch/atlas/athena/-/merge_requests/51274Ensure all duplicated proxies are reset and newly read2022-03-31T15:42:43+02:00Peter Van GemmerenEnsure all duplicated proxies are reset and newly readFor metadata, the SharedWriter may have to merge multiple streams at once, which can result in conflicts of having the same type#key combination from different streams. Typically (as in derivation) those are identical metadata objects wr...For metadata, the SharedWriter may have to merge multiple streams at once, which can result in conflicts of having the same type#key combination from different streams. Typically (as in derivation) those are identical metadata objects written to more than one stream. This change will ensure that in case of such a conflict, the existing proxy is reset so the new data object can be loaded from shared memory ant not cause the worker/writer to get stuck (as seen in FAFE mode with multiple streams).
For workflows with multi-level streams (such as RAWtoALL) it still needs to be checked whether objects with the same type#key are identical (potential subject to future MR).https://gitlab.cern.ch/atlas/athena/-/merge_requests/49871Flavor Tagging GNN: bug fix + cleanup2022-01-27T03:42:16+01:00Nilotpal KakatiFlavor Tagging GNN: bug fix + cleanupThis MR includes -
- Bugfix - When `ORT Tensor` objects are created, the tensor only has a memory reference to the data (not a deepcopy), so the previous approach of flattening the vectors in a loop and creating `ORT tensors` from it wa...This MR includes -
- Bugfix - When `ORT Tensor` objects are created, the tensor only has a memory reference to the data (not a deepcopy), so the previous approach of flattening the vectors in a loop and creating `ORT tensors` from it was faulty (in `OnnxUtil.cxx`). Now the inputs to the GNN are created as 1D vectors directly (in `GNNTool.cxx`)
- Typo fix following the MR [!48584](https://gitlab.cern.ch/atlas/athena/-/merge_requests/48584#note_5038375)
- Clean up to improve readability
Tagging @dguesthttps://gitlab.cern.ch/atlas/athena/-/merge_requests/47749ATLASRECTS-6655 : Returning a huge value (1.997kT) for the mag field , when...2021-11-03T01:17:03+01:00Christos Anastopouloschristos.anastopoulos@cern.chATLASRECTS-6655 : Returning a huge value (1.997kT) for the mag field , when...ATLASRECTS-6655 : Returning a huge value (1.997kT) for the mag field , when none is present , from the getFieldZR creates issues. Return 0.1 Gauss in these cases
Ping @schaffer and @goblirsc . This is a fresh one hopefully cleanerATLASRECTS-6655 : Returning a huge value (1.997kT) for the mag field , when none is present , from the getFieldZR creates issues. Return 0.1 Gauss in these cases
Ping @schaffer and @goblirsc . This is a fresh one hopefully cleanerhttps://gitlab.cern.ch/atlas/athena/-/merge_requests/47686Trigger Navigation conversion - fix recursive seed2021-10-29T15:48:55+02:00Tomasz BoldTrigger Navigation conversion - fix recursive seedThis MR protects merging of nodes that are in parent child relation. Mering those results in recursive seed.
Tagging @tamartin @wprzygodThis MR protects merging of nodes that are in parent child relation. Mering those results in recursive seed.
Tagging @tamartin @wprzygodhttps://gitlab.cern.ch/atlas/athena/-/merge_requests/47682WorkDir: temporary fix for finding CORAL/COOL2021-10-30T18:45:55+02:00Frank WinklmeierWorkDir: temporary fix for finding CORAL/COOLTemporary hack to find CORAL/COOL for projects where it is built as part
of tdaq-common (ATLINFR-4302).Temporary hack to find CORAL/COOL for projects where it is built as part
of tdaq-common (ATLINFR-4302).https://gitlab.cern.ch/atlas/athena/-/merge_requests/47575Fix TriggerMenuMT EventBuildingSequences unit test2021-10-26T16:02:22+02:00Rafal Bielskirafal.bielski@cern.chFix TriggerMenuMT EventBuildingSequences unit testQuick fix for ATR-24406
* add missing import
* use `tool.getType()` which works for old and conf2 configurables instead of `__cpp_type__` which is conf2 only
Might want to ensure that all tools are same configurable type later on, but ...Quick fix for ATR-24406
* add missing import
* use `tool.getType()` which works for old and conf2 configurables instead of `__cpp_type__` which is conf2 only
Might want to ensure that all tools are same configurable type later on, but this at least fixes the unit test which fails in CI.