search

cernanalysispreservation / analysispreservation.cern.ch

Source code for the CERN Analysis Preservation portal
https://analysispreservation.cern.ch/
GNU General Public License v3.0
43 stars 30 forks source link

CMS Analysis preservation demo input form feedback #132

Closed gbenelli closed 7 years ago

gbenelli commented 8 years ago

Comments are based on the form at: https://analysis-preservation-qa.cern.ch/records/CMSAnalysis/create Note: Not sure whether this is intended or not, but I was not able (permission issue) to access the 3 CMS Records:

CADI-NUM 8 (CMSAnalysis) Untitled 10 (CMSAnalysis) B2G-12-015 11 (CMSAnalysis) listed in https://analysis-preservation-qa.cern.ch/CMS/ I think that viewing (by CMS people) should always be allowed, while editing could be restricted. Should there be a setting allowing the person filling the form to identify other CMS users allowed to edit the given form?

Basic Information

Support idea by Wei (issue #129 ) of adding some text field for the abstract/description. Also support idea by Kati of adding the 3 use cases:

  1. CADI line (AN/PAS/Paper number)
  2. CMS DAS (or workbook/tutorial) exercise (like Z-> tau tau) exercise used by Rishika)
  3. Release Validation

    AOD Processing

Autocomplete using full DAS querying for data and MC datasets would be very useful. Perhaps you want to ask the Center of Mass Energy (7, 8, 13 TeV) to restrict the autocomplete range? I think it would make sense to have dataset/trigger pairs (i.e. user could put a set of datasets with 1 set of triggers, but also create a second set of datasets with another trigger, for example when having multiple final states combinations). I may be wrong but I think that the comments by Wei (issue #129 ) that he has under AOD processing really should go under the next section (Physics Information) where the selection should be described.

Physics Information

Synthesizing the comments I made at out last meeting:

  1. I think that it's fine to put down all the particle object selection (currently Final State Particles), but these are just the atomic components of the selection, not the final state itself.
    • The list needs to be completed, e.g. electron, muon, tau, but very quickly one needs to consider "objects" rather than particles, e.g. jets (here one needs to think whether the type of jet is a sub-variable of the jet, e.g. AK4, AK8 or a different object, e.g. b-jet vs. other type of jets), MET etc.
    • It would be great if given the dataset, the form could automatically get the object collections available in miniAOD, and possibly the available variables (I understand this could be enormously complicated).
    • For each object one has to have the option to define both a "tight" and a "loose" object (based on how tight or loose are the selection criteria).
    • In the selection criteria, transverse momentum (p_T) and eta are valid variables, but also the various "isolation" definitions.
      1. I would then separately define final state (allowing the possibility of having multiple of them, e.g. in dilepton final state analysis, one would generally have slightly different selections for ee, mumu and emu final states)
    • In each final state one could define a signal region (in general more than one) where the actual final state is described (e.g. 2 tight muons, veto on extra tight muons or tight electrons, with p_T>X, miniIsolation>0.2 etc) including cuts that imply a specific object reconstruction (for example a leptonic W decay, where the MET is identified with the neutrino p_T, or an hadronic W decay, where jets are combined with a W hadronic decay hypothesis, or a Z decay in two leptons where generally vetos in the region around the Z mass are applied) and signal hypothesis. In the same final state one could define several control regions (using loose leptons, or relaxing isolation cuts, or inverting Z mass veto cuts) that are used in the analysis some times to validate the data/MC agreement, other times to estimate data driven backgrounds (so they are defined only in data and possibly a few MC samples)

Unfortunately I do not have my notes with me and I do not remember further details on this part, but I think it would be good to possibly include a sample "ntuple" file produced from the processing of data, MC signal and MC background. Maybe indicate the size of these ntuples? Ideally reading out one of these files automatically would give all the object variables and the event variables that are used. I would also probably ask here for the "ntuple" production code used. Most analysis have a "framework" in common which is used to produce ntuples out of miniAOD, then a set of (py)ROOT/C++ programs to analyze further (estimate efficiencies, do systematics studies, extract data.MC scale factors, extract signal, by cut and count or sophisticated fits, extract cross-sections, total or differential, masses, spin/angular distributions/asymmetries measurements or cross-section limits). Those common frameworks generally live in some concurrent version system (GitHub, CVS or SVN). Also the common ntuples are generally stored in some common (Tier-2?) space so that people can easily run on those via Grid jobs.

Post AOD processing

I think that here OS and CMSSW version should be automatically detected (at least a sub-list of all possible CMSSW and SLC versions). I would use a CMSSW field rather than an Analysis software field, this last one should be reserved to a GitHub/SVN/CVS repository of the user(s)/group code used to do the second step (from the ntuple to the final production of tables/plots/estimates/systematics/results/limits).

As mentioned at the meeting, I think it would make sense to get a full description on how to run the various components to reproduce the analysis and to capture in a comprehensive archive all the analysis code used to go from miniAOD to the final set of public plots/tables/results.

Hope this helps! Gabriele

suenjedt commented 8 years ago
ghost commented 8 years ago

I also support the ideas given by all. Just have one suggestion.

In the form: https://analysis-preservation-qa.cern.ch/records/CMSAnalysis/create

Post-AOD Processing has options for Input Data Files and Output Data. I think it would be good if other options can also be added there like eos area or something else. As some people dont have Web area for storage.

Thanks Rishika

suenjedt commented 8 years ago

possibly superseed by #155 @Kjili to check overlap and close this

Kjili commented 8 years ago

@rishika123 Could you please have a look at the new https://analysis-preservation-qa.cern.ch/records/CMSAnalysis/create and see if this is still an issue? Generally we support every given link for output data and propagate the input data from the previous output. What we have as placeholder is just an example for what can be inserted there. But if you have a suggestion on improving this example or you require some additional feature/field please tell us.

ghost commented 8 years ago

Hi Kjili

I shall look into the new form and will let you know if the problem still exists or not.

Thanks Rishika

On Tue, Aug 9, 2016 at 9:46 AM, Kjili notifications@github.com wrote:

@rishika123 https://github.com/rishika123 Could you please have a look at the new https://analysis-preservation-qa.cern.ch/records/ CMSAnalysis/create and see if this is still an issue? Generally we support every given link for output data and propagate the input data from the previous output. What we have as placeholder is just an example for what can be inserted there. But if you have a suggestion on improving this example or you require some additional feature/field please tell us.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/cernanalysispreservation/analysis-preservation.cern.ch/issues/132#issuecomment-238576616, or mute the thread https://github.com/notifications/unsubscribe-auth/AHvNVEDnabOn0a5xW48kGNE3zZ33zadKks5qeJLRgaJpZM4IctR8 .

Kjili commented 8 years ago

Great, thanks!

ghost commented 8 years ago

Hi Kjili

I am not able to open the form from the web link:

https://analysis-preservation-qa.cern.ch/records/CMSAnalysis/create

Everytime I sign in from my CERN account and click on CMS it again asks for sign in. And this is repeating everytime. Please suggest.

Thanks Rishika

On Tue, Aug 9, 2016 at 10:07 AM, Kjili notifications@github.com wrote:

Great, thanks!

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/cernanalysispreservation/analysis-preservation.cern.ch/issues/132#issuecomment-238583575, or mute the thread https://github.com/notifications/unsubscribe-auth/AHvNVCVOLAu4_Wp04J9mhX9UZOLgZvsUks5qeJe0gaJpZM4IctR8 .

Kjili commented 8 years ago

Hi Rishika,

this should not happen and I am not sure why it does. When I click on the link and sign in I am currently taken to a page for linking accounts but if I click on the name in the top left or open the link again I can use the page as usual. Are you taken back to the exact same sign-in page are you only asked to sign in again when you try to access the CMS landing page? Also, if you are signed in your email adress should appear at the top right where the sign-in button was before. Does it do that or do you still see the button?

Thanks, Kjili

ghost commented 8 years ago

Hi Kjili

I am able to open the link now and create a new form. I added some information to it and saved that. It got saved. But then I tried editing that and clicked on save button (bottom left) of the form. It doesnt get saved.

Also when i tried signing up again it says:

Internal Server Error.

I cant open the form again.

Thanks Rishika

On Wed, Aug 10, 2016 at 3:15 AM, Kjili notifications@github.com wrote:

Hi Rishika,

this should not happen and I am not sure why it does. When I click on the link and sign in I am currently taken to a page for linking accounts but if I click on the name in the top left or open the link again I can use the page as usual. Are you taken back to the exact same sign-in page are you only asked to sign in again when you try to access the CMS landing page? Also, if you are signed in your email adress should appear at the top right where the sign-in button was before. Does it do that or do you still see the button?

Thanks, Kjili

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/cernanalysispreservation/analysis-preservation.cern.ch/issues/132#issuecomment-238796797, or mute the thread https://github.com/notifications/unsubscribe-auth/AHvNVGfxpeI40PXAT44sI05LddjncAK1ks5qeYimgaJpZM4IctR8 .

Kjili commented 8 years ago

Hi Rishika,

can't you sign in at all any more or is it just a problem related to your record and you can sign into the main page?

Thanks, Kjili

ghost commented 8 years ago

I can sign in now. But I see the changes I made to the form doesn't get saved.

On Wed, Aug 10, 2016 at 10:45 AM, Kjili notifications@github.com wrote:

Hi Rishika,

can't you sign in at all any more or is it just a problem related to your record and you can sign into the main page?

Thanks, Kjili

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/cernanalysispreservation/analysis-preservation.cern.ch/issues/132#issuecomment-238909397, or mute the thread https://github.com/notifications/unsubscribe-auth/AHvNVNjbntM8gycVbqH1s8nbHah54pSzks5qefIkgaJpZM4IctR8 .

Kjili commented 8 years ago

Thanks for reporting back. This should not be the case and we will investigate what happens and check back with you as soon as we know more or the problem is resolved.

Kjili commented 7 years ago

Thank you for the feedback and sorry for the delay. This should be fixed now, the new page https://analysispreservation.cern.ch will be accessible from outside CERN soon. If any problems occur with this, please create a new issue.