sPHENIX discussion of software

["Perepelitsa, Dennis" <dvp AT bnl.gov>]

Hi all (CC: to simulations team),

A few things: 

I’ve regenerated input HEPMC files with more appropriate filtering in truth jet |eta| for the narrowed EMCal acceptance we are supposed to consider for these simulations. Events are generated in five jet-pT slices, with 10k events / slice for now:

/sphenix/user/dvp/gen/QCD7/hepmc_QCD7GeV_R04jet15GeVeta0p45_seq* (generator-level trigger on presence of R=0.4 truth jet with pT > 15 GeV within |eta|<0.45)

/sphenix/user/dvp/gen/QCD10/hepmc_QCD10GeV_R04jet20GeVeta0p45_seq* ( “" pT > 20 GeV within |eta|<0.45 “” )

/sphenix/user/dvp/gen/QCD15/hepmc_QCD15GeV_R04jet30GeVeta0p45_seq* ( “" pT > 30 GeV within |eta|<0.45 “” )

/sphenix/user/dvp/gen/QCD25/hepmc_QCD25GeV_R04jet40GeVeta0p45_seq* ( “" pT > 40 GeV within |eta|<0.45 “” )

/sphenix/user/dvp/gen/QCD35/hepmc_QCD35GeV_R04jet50GeVeta0p45_seq* ( “" pT > 50 GeV within |eta|<0.45 “” )

(however, one of Aaron’s nice suggestions below was to run an higher-pT slice to understand the pT-dependence of the JER — I could do that)

Using these, I’ve kicked off some initial signal-only (p+p) simulations using the official CD-1 tag of the geometry, and am creating some summary output trees (recording R=0.2, 0.3, 0.4, 0.5 truth and subtracted-reco jets, seed jets, summary of estimated UE backgrounds, truth particles, etc.).

Unfortunately it is going slower than I was hoping so there are only partial-statistics tree outputs available right now: 

/sphenix/user/dvp/sims/new-jet/pp-QCD7-7k-test.root (to be used for 15-20 GeV jets)

/sphenix/user/dvp/sims/new-jet/pp-QCD10-7k-test.root (to be used for 20-30 GeV jets)

/sphenix/user/dvp/sims/new-jet/pp-QCD15-7k-test.root (to be used for 30-40 GeV jets)

/sphenix/user/dvp/sims/new-jet/pp-QCD25-7k-test.root (to be used for 40-50 GeV jets)

/sphenix/user/dvp/sims/new-jet/pp-QCD35-7k-test.root (to be used for 50-60 GeV jets)

I’ve asked the Colorado group to start taking a look but I post these here so that others can as well.

I’m continuing to push through the rest of the pp simulations, and am about to start a full round of embedding simulations (b=0-4fm and b=4-8fm).

Aaron, thanks for taking a close look and providing this feedback. Let me reply to your comments inline below...

Dennis

On Apr 10, 2018, at 10:41 AM, Aaron Angerami <angerami AT cern.ch> wrote:

Hi Dennis,

Just looking through the list of proposed plots, I have a couple of comments.

-The specific charge of the CD-1 review:

Is the conceptual design technically sound and likely to meet the objectives of its scientific case?

Some of the plots towards the end probably go beyond this. I think this is fine and its always a good opportunity to show how far along the collaboration is, but we should make sure the emphasis is in the proper place: the relationship between the specifics of the CD-1 design and experimental capabilities.

I agree with you that the latter plots are significantly more “derived” indications performance rather than “core" metrics. However, we have a specific request to include plots like this, in part because we had versions of these plots in the old MIE proposal document and review committees were requesting updated versions.

-I think its really important that all figures be made with some kind of label that indicates what configuration of the detector and simulation is being used, e.g. “sPHENIX CD-1 projection” or some such thing. Probably should coordinate this across different TGs, etc. Not only would it make it look like we have our act together, but is essential for plot archaeology.

That’s a great idea. I like "sPHENIX CD-1 projection”.

-Perhaps this is suggestion is coming from the spokespersons, or was discussed elsewhere, but should we be assuming a 5-year running plan? Or is this too optimistic with what is being officially proposed in the context of CD-1?

Yes, I believe we should. Jamie Nagle has outlined a proposed 5-year running plan with projected rates & narrow vertex fractions from CA-D (as part of the Colorado group’s responsibilities on trigger, rates, etc.), and we want to synchronize all projections with these across the topical groups.

-Do we have a version of plot 2 that compares different centralities and pp but at fixed R? It would be very nice to demonstrate that the behavior of the JER is understood. If these exist, it would be nice to actually do some fits of the JER to compare the different parameters. It may also help to go to higher pT, even if we don’t expect data there, to better fix the constant term. I’m willing to help with the fits for this.

Yes, we should certainly make plots like this (for plot #2, what’s shown on my slide was intended to be just an indication). I think we can coordinate either giving you some trees directly, or at least the extracted JER values to do this kind of study.

-For plots 3-6, I think this exercise is only meaningful if we choose binning consistent with what we would actually use in a measurement. The right figure has the potential to do more harm than good if a reviewer has any experience with jet measurements. 

Given that I think there is a pretty strong desire to have analogues of these, we should definitely discuss how to do these “right”.

-If we have the capabilities to make those plots, can we also make them showing including some dummy uncertainties? e.g. uncertainty on an unfolded RAA assuming an uncertainty on the JES or JER motivated by the design? If this is too large a task, we could do the unfolding with a reweighed spectrum. Anything to show how well the design holds up when the data doesn’t exactly match the simulation, as this is the practical limitation in most cases.

That’s a good idea. We should discuss how to motivate and defend reasonable uncertainties, perhaps in the Jet Structure meeting on Monday (making sure to draw in John Lajoie & Songkyo, etc.) 

Aaron

On Apr 9, 2018, at 10:48 PM, Perepelitsa, Dennis <dvp AT bnl.gov> wrote:

Hi Jet Structure members,

As you have heard in the General & Simulations Meetings, we are organizing a simulations and analysis effort to produce “basic” and “derived” jet performance plots, and updated projection plots, for the CD-1 Review. 

We have a working list of proposed plots here: https://indico.bnl.gov/event/4433/contributions/20632/attachments/17937/22158/dvp-sPHENIX-proposed-plots-3-22-18.pdf , and a major goal is to converge on draft versions of these by the GM on Friday, April 20th.

We’d like to call a meeting in one week to review the status and get input from Jet Structure members, at: 

Monday, April 16th, 4pm BNL time / 2pm Mountain time

If you are available to work on simulations directly, please get in contact with myself and Rosi before then (Skype, email, Slack) so we can coordinate. If not, we would still welcome your input as plots are being drafted.

— 

Below, I am collecting some useful information regarding the simulations effort:

Slack channel #cd1-jet-simulation : https://sphenix-software.slack.com/messages/CA1CYEKRB/ 

Official tag of the detector to be used for all CD-1 review targeted MC production: https://github.com/sPHENIX-Collaboration/macros/tree/CD-1-reference-jet-production 

Output of b=0-4fm and b=4-8fm hits files: /sphenix/sim/sim01/cd1_review/sHijing/fm_0-4/ and /sphenix/sim/sim01/cd1_review/sHijing/fm_4-8/

Working example of G4 HITS read back + signal embedding: https://github.com/sPHENIX-Collaboration/macros/tree/CD-1-reference-jet-production-readback-embed

Dennis 

Dennis V. Perepelitsa

Assistant Professor, Physics Department

University of Colorado Boulder

_______________________________________________
sPHENIX-jet-structure-l mailing list
sPHENIX-jet-structure-l AT lists.bnl.gov
https://lists.bnl.gov/mailman/listinfo/sphenix-jet-structure-l

Dennis V. Perepelitsa

Assistant Professor, Physics Department

University of Colorado Boulder