Update Run 3 geometry DB payloads #31529
Conversation
|
test parameters |
|
The code-checks are being triggered in jenkins. |
|
+code-checks Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-31529/18505
|
|
A new Pull Request was created by @christopheralanwest for master. It involves the following packages: Configuration/AlCa @cmsbuild, @pohsun, @christopheralanwest, @tocheng, @tlampen can you please review it and eventually sign? Thanks. cms-bot commands are listed here |
|
please test |
|
The tests are being triggered in jenkins.
|
|
+1 |
|
Comparison job queued. |
|
Comparison is ready @slava77 comparisons for the following workflows were not done due to missing matrix map:
Comparison Summary:
|
| # GlobalTag for MC production with realistic conditions for Phase1 2021 | ||
| 'phase1_2021_realistic' : '112X_mcRun3_2021_realistic_v7', # GT containing realistic conditions for Phase1 2021 | ||
| 'phase1_2021_realistic' : '112X_mcRun3_2021_realistic_v8', # GT containing realistic conditions for Phase1 2021 |
There was a problem hiding this comment.
I was expecting to see also a new ExtentedZeroMaterial labeled tag for GeometryFileRcd, based on the Extended2021ZeroMaterial scenario introduced in #31082. Will that follow in another PR?
There was a problem hiding this comment.
@cvuosalo - IMHO, these scenarios should be uploaded together. Otherwise, we should request to delete ExtentedZeroMaterial from this GT.
There was a problem hiding this comment.
at the moment there is no such label in that GT:
$ conddb list 112X_mcRun3_2021_realistic_v8 | grep GeometryFileRcd
[2020-09-22 12:23:13,561] INFO: Connecting to pro [frontier://PromptProd/CMS_CONDITIONS]
GeometryFileRcd CTPPS XMLFILE_CTPPS_Geometry_2018_101YV6
GeometryFileRcd Extended XMLFILE_Geometry_112YV2_Extended2021_mc
MFGeometryFileRcd 120812 MFGeometry_120812
MFGeometryFileRcd 130503 MFGeometry_130503
MFGeometryFileRcd 160812 MFGeometry_160812
MFGeometryFileRcd 90322 MFGeometry_90322
There was a problem hiding this comment.
Does a tag for the ExtendedZeroMaterial scenario exist? I understood from @cvuosalo comments at yesterday's AlCaDB meeting that it did not and that we should not wait for the zero material scenario to converge. Can the payload corresponding to #31082 be generated now or is additional work required?
@cvuosalo @ianna Please also see and respond to @fwyzard's recent HN post regarding use of old MC samples in releases built after this PR is merged.
There was a problem hiding this comment.
I understood from @cvuosalo comments at yesterday's AlCaDB meeting that it did not and that we should not wait for the zero material scenario to converge.
As far as I can see there are some samples for EGM that need that to fall in place in order to be injected [*], so some priority would be needed. Alternatively the driver command for those samples could be tweaked in order to deliver the simulation geometry from XML.
[*] https://hypernews.cern.ch/HyperNews/CMS/get/prep-ops/7302.html
There was a problem hiding this comment.
Does a tag for the
ExtendedZeroMaterialscenario exist? I understood from @cvuosalo comments at yesterday's AlCaDB meeting that it did not and that we should not wait for the zero material scenario to converge. Can the payload corresponding to #31082 be generated now or is additional work required?@cvuosalo @ianna Please also see and respond to @fwyzard's recent HN post regarding use of old MC samples in releases built after this PR is merged.
I'll check with @cvuosalo about the Zero material payload. The scripts used to produce all Sim geometry variations at once.
I've noticed there is a GeometryFileRcd with a CTPPS label. Is it with iovs? e.g. is it their Reco geometry? It should eventually go away, I think, they have decoupled their Reco geometry from the XML now.
The last point - the Magnetic field geometry. How old the payloads are? The change in Muon yoke thickness should be identical for both Sim geometry and MF geometry.
Sorry, I would have checked it myself, but I'm having trouble connecting to CondDB...
There was a problem hiding this comment.
The real changes proposed in #31082 involve only the densities of the non-active volumes and IIUC are not coupled with the other (small) tracker changes (solution of overlaps, etc.) for Run-3. I would say the payload could be derived right away
|
Currently, before this PR, for GT 112X_mcRun3_2021_realistic_v7 there is no Zero Material Geometry File Record. This PR does not change that. |
|
Concerning the |
|
The @namapane FYI: The Run 3 Extended 2021 geometry is being uploaded to the Conditions DB. It contains minor changes in CSC geometry. You may want to upload new versions of the |
|
The comparison tests may be showing larger changes than might be expected in Run 3 workflows. The largest I've seen are in workflow 11634 where Tracker hits seem to move from one TOB to another in the "alternative comparisons". For example, @vargasa Could you please help to interpret what these plots tell us?
|
|
@vargasa Could you please comment on whether 10 MC events in the baseline and 10 independent MC events in the PR would be expected to produce the plots I show in my comment (#31529 (comment)), that is, certain slices of eta have no hits while other slices have thousands of hits? |
|
@cvuosalo I see those differences not only within the outer barrel but also in end-cap, inner disks, etc. It may seem reasonable to see those differences when using different MC events. Is it possible to get more statistics (or perhaps rerun the workflows)? i.e check to see if the same regions remain empty (which may indicate there's a problem). @kpedro88, from previous PRs I have seen you working with these comparisons. Can you shine some light in here? please |
|
@vargasa currently there is no option for the bot to rerun comparisons with more statistics. If you want to make those comparisons, you have to run them yourself. |
|
unhold |
|
The issues in #31519 go back to Run 2 and are not a reason to stop this PR, so I removed the hold. |
|
I think the differences I posted #31529 (comment) are probably due to low statistics of comparing 10 events in the baseline to 10 independent events for the PR. I've generated 700 events each for baseline and PR geometry, and I will check them to see the differences. |
|
With 700 MC events generated each for baseline and PR here are similar plots to those above.
|
|
@cvuosalo Do you see any empty regions in the new plots? |
|
@vargasa With 700 events, I didn't find any of these Tracker Hit plots that are empty for baseline or PR. They only show the kind of differences that I posted. |
|
@christopheralanwest Can this PR go forward so that it gets into the pre-release on the 29th? |
|
To conclude this is due to some statistical effect you should look at the baseline done with two different random number seeds. I don’t see how to conclude it from a comparison of runs with two different geometries.
David
On 25 Sep 2020, at 02:07, Carl Vuosalo <notifications@github.com> wrote:
With 700 MC events generated each for baseline and PR here are similar plots to those above.
I think the plots show there is substantial statistical fluctuation in these Tracker Hits plots, so differences should be expected.
I think it is reasonable to conclude that there is no evidence of problems in the Run 3 geometry payloads, so this PR should go ahead.
[image]<https://user-images.githubusercontent.com/5736159/94211910-52736600-fe98-11ea-91ea-3264bf53bcfd.png>
[image]<https://user-images.githubusercontent.com/5736159/94212006-99615b80-fe98-11ea-83ae-987737c0d918.png>
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub<#31529 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ABGPFQ56OUKEGD6FSRW22S3SHPNMBANCNFSM4RU5B5XA>.
|
|
@davidlange6 Could you please suggest an easy way to get different random seeds? I'm getting the same events for 11634.0 when I do runTheMatrix.py twice. |
|
you can put this in the configs generated by from IOMC.RandomEngine.RandomServiceHelper import RandomNumberServiceHelper
randHelper = RandomNumberServiceHelper(process.RandomNumberGeneratorService)
randHelper.resetSeeds(120) |
|
In a baseline vs. baseline comparison (both using the current Global Tag) with 10 events each with different random seeds, the plots show similar statistical fluctuations as seen in the PR comparison tests. I think we can conclude that the comparison differences in the PR tests are due to fluctuations from the low statistics (only 10 events) and not due to any error in the new geometry. The black in the plots below is the same standard 10 events for 11634.0 with the same random seed as in the PR test. The red is 10 events with a different random seed.
|
|
@cvuosalo Are you satisfied with your checks so that I may sign off on this PR? If the zero material payload becomes available, I can open a new PR to add it. I would rather not delay this PR unnecessarily so as to ensure that it goes into the next pre-release. |
|
@christopheralanwest Yes, I am satisfied. Please approve this PR so it can be merged and get into pre7. The zero material scenario is not urgent and can come later. |
|
+alca |
cmsbuild
added
alca-approved
fully-signed
and removed
alca-pending
pending-signatures
labels
|
This pull request is fully signed and it will be integrated in one of the next master IBs (tests are also fine). This pull request will now be reviewed by the release team before it's merged. @silviodonato, @dpiparo, @qliphy (and backports should be raised in the release meeting by the corresponding L2) |
|
+1 |
PR description:
This PR updates the geometry DB payloads in the Run 3 MC global tags. The changes and their validation are described in the 21 Sep 2020 AlCaDB presentation by @cvuosalo.
The Run 3 data GTs will be updated once release validation of the payloads has been completed.
The global tag diffs are the same for all six scenarios:
Attn: @cvuosalo @ianna
2021 design
https://cms-conddb.cern.ch/cmsDbBrowser/diff/Prod/gts/112X_mcRun3_2021_design_v7/112X_mcRun3_2021_design_v8
2021 realistic
https://cms-conddb.cern.ch/cmsDbBrowser/diff/Prod/gts/112X_mcRun3_2021_realistic_v7/112X_mcRun3_2021_realistic_v8
2021 cosmics
https://cms-conddb.cern.ch/cmsDbBrowser/diff/Prod/gts/112X_mcRun3_2021cosmics_realistic_deco_v7/112X_mcRun3_2021cosmics_realistic_deco_v8
2021 heavy ion
https://cms-conddb.cern.ch/cmsDbBrowser/diff/Prod/gts/112X_mcRun3_2021_realistic_HI_v8/112X_mcRun3_2021_realistic_HI_v9
2023 realistic
https://cms-conddb.cern.ch/cmsDbBrowser/diff/Prod/gts/112X_mcRun3_2023_realistic_v7/112X_mcRun3_2023_realistic_v8
2024 realistic
https://cms-conddb.cern.ch/cmsDbBrowser/diff/Prod/gts/112X_mcRun3_2024_realistic_v7/112X_mcRun3_2024_realistic_v8
PR validation:
Please see the 21 Sep 2020 AlCaDB presentation by @cvuosalo for details.
In addition, a technical test was performed:
runTheMatrix.py -l limited,12024.0,7.23,159.0,12834.0,7.24 --ibeosif this PR is a backport please specify the original PR and why you need to backport that PR:
This PR is not a backport and should not be backported.