Presentation is loading. Please wait.

Presentation is loading. Please wait.

Diphoton + MET Status Update

Similar presentations


Presentation on theme: "Diphoton + MET Status Update"— Presentation transcript:

1 Diphoton + MET Status Update
Daniel Damiani Editorial Board Meeting 18 April 11

2 Photon + MET Editorial Board Meeting
Outline Updates/Changes Bug fix Extra material systematic treatment* Genuine ETmiss background model systematics* PDF uncertainties* Official Production Grid Status Numbers and Plots to Present for the Paper CDS Comments Status of Note/Paper *Indicates this was a request from approval meeting 18/04/11 Photon + MET Editorial Board Meeting

3 Photon + MET Editorial Board Meeting
Bug Fix There was a bug found in the code that was calculating the signal MC acceptance. Effectively no pt cut was being applied to photons as a result of the bug. Change in acceptance was not drastic because of the pT spectrum of the photons. Acceptances decreased by <1%. Affected only the MC acceptance numbers. 18/04/11 Photon + MET Editorial Board Meeting

4 Extra Material Systematics [1/2]
UED has dedicated extra material samples. GGM grids do not – the extra material systematic is extrapolated from number from the direct photon analysis. Uncertainties from direct photon group quoted for photons as a function of pT, η, and if it is converted or not. Objections about this difference were raised at the SUSY approval talk Quoted uncertainty for each is an average. 18/04/11 Photon + MET Editorial Board Meeting

5 Extra Material Systematics [2/2]
Numbers from the different methods UED method on UED samples – uncertainty of 2.1% GGM method on GGM samples – uncertainty of 2.5% GGM method was applied to two of the UED samples. Gave an average uncertainty of 2.5% Close to the UED dedicated extra material samples Note the numbers for UED that averaged to 2.1% had a large variance (0.77%, 2.26%, and 3.4%). Two methods seem to be consistent. Will quote 2.5% for UED to unify methodology. 18/04/11 Photon + MET Editorial Board Meeting

6 Background Estimation [1/3]
Objects were raised to our method for estimating the uncertainty on the background contribution of the genuine ETmiss sample. Used loosened electron-photon sample as central value and quoted systematic as the difference between loosened and 'nominal' sample – 300%. To address this: Use the nominal electron-photon sample from data as the central value. The ETmiss distribution from ttbar and W+X MC was normalized to the number of events in the nominal electron-photon sample and compare. 18/04/11 Photon + MET Editorial Board Meeting

7 Background Estimation [2/3]
From this quote a new total uncertainty on the genuine ETmiss of 33% for ETmiss > 125 GeV cut. Scale factor uncertainty: 10% Template shape for subtracting QCD: 6% Nominal eγ to MC comparison: 31% (was 300%) 18/04/11 Photon + MET Editorial Board Meeting

8 Background Estimation [3/3]
Updated total expected background with new treatment of the genuine ETmiss sample of 0.16±0.08(stat)±0.07(syst) events. The previous number: 0.089±0.051(stat)±0.075(syst) events. See backup for old version of table below. 18/04/11 Photon + MET Editorial Board Meeting

9 PDF Uncertainties Treatment [1/2]
Objections raised at approval meeting to the different treatments of PDF uncertainties. Both sets of samples generated in Pythia with MRST2007LOmod PDF. Previous method: GGM: The central value of the NLO cross section is calculated by Prospino with the CTEQ6.6m PDF and used as the x-sec. Use CTEQ6.6 error sets to get an uncertainty. UED: Generated CTEQ6.6 event samples for all error sets. Used these to derive an uncertainty relative to the nominal CTEQ6.6 cross section. Additionally take the difference between MRST2007LOmod and CTEQ6.6 as an additional uncertainty. 18/04/11 Photon + MET Editorial Board Meeting

10 PDF Uncertainties Treatment [2/2]
Proposed new way: GGM Points: Use only the CTEQ6.6 NLO cross section value from Prospino, Use CTEQ6.6 error sets to get an uncertainty. UED Points: Use error sets from MSTW2008lo90cl compared to the nominal MSTW to obtain an uncertainty on the cross section. This is the recommend error set to use for cross sections from MRST2007LOmod. New uncertainty independent of 1/R: 4%. Still waiting on feedback from the experts. 18/04/11 Photon + MET Editorial Board Meeting

11 Grid Production Status
Gluino vs. Bino grid: Official production of events down to a bino mass of 50 GeV is more or less complete. CMS like gluino vs. squark grid: Official production done for some points (~1/3). A total of 187 points were requested. Bino mass of 50, 150, and 500 GeV. Generated as to directly compare to CMS grid. SLHA files fed to Pythia have identical theory assumptions to CMS Worked CMS + theorists to confirm this. Some odd theory assumptions chosen for this gird. May be possible to recast grid without these assumptions. 18/04/11 Photon + MET Editorial Board Meeting

12 Photon + MET Editorial Board Meeting
Results to Present Results we plan to present in the paper: Raw number of excluded events. Excluded cross-section ranges (with no theoretical uncertainties applied). Limit contour plots – Methods to show: Frequentist (CLs and PCL) – SUSY suggested. Bayesian plus one frequentist method – Exotics suggested. Updated limits not yet in the note/paper. Updates to be made in the next few days. Numbers on the following three slides are not the latest numbers. 18/04/11 Photon + MET Editorial Board Meeting

13 Raw Event Number Limits
The limit on raw number of events excluded at 95% confidence that we plan to quote is ~2.2 events. Includes background uncertainties only. This limit is independent of the chosen models. The raw number of events excluded at 95% confidence using a Bayesian method is ~3.3 events. Do not plan to quote this in paper (int. note only). 18/04/11 Photon + MET Editorial Board Meeting

14 Photon + MET Editorial Board Meeting
Cross Section Limits For the cross section limits currently quoted in the paper is a range of excluded cross sections that fall along the exclusion limit. This makes the more sense to us than quoting the range including all points, even those excluded points far from the limit. That would just give an arbitrarily high cross section based on merely the lowest mass point included in the grid. Cross section limits currently in paper: GGM – σ = 0.25 – 0.51 pb UED – σ = 0.15 – 0.50 pb Note: CMS quoted their cross section exclusion range for the mbino = 150 GeV points only in their 3d grid. (σ = 0.3 – 1.1 pb). 18/04/11 Photon + MET Editorial Board Meeting

15 CMS Cross Section Limits
From CMS's preprint: 18/04/11 Photon + MET Editorial Board Meeting

16 Photon + MET Editorial Board Meeting
Limit Contours [1/3] It appears there was still some disagreement from the physics groups on what types of limits to put in publication at the Stats Forum meeting on Friday . My understanding of Physics group recommendations: SUSY Group Preference: PCL and CLs. Exotics Group Preference: One Bayesian + One frequentist method. Current limit plots in the paper draft: GGM – CLs+b; UED – Bayesian. These need to be unified plus updated for the latest numbers. In the case of nobs=0 the Bayesian limits and CLs limits will be exactly the same (nsignal < 3.0 neglecting systematics). 18/04/11 Photon + MET Editorial Board Meeting

17 Photon + MET Editorial Board Meeting
Limit Contours [2/3] 18/04/11 Photon + MET Editorial Board Meeting

18 Photon + MET Editorial Board Meeting
Limit Contours [3/3] Note: This plot is with a Bayesian limit. Excludes 1/R ≤ 960 GeV. A frequentist limit has been calculated (1/R ≤ 990 GeV), but the plot has not yet been generated. 18/04/11 Photon + MET Editorial Board Meeting

19 Status of the Note and Paper
Current draft of the paper is here: All limit numbers need to be update still. (E.T.A. next day or so). Limits still need to be updated in the internal note as well. Same time frame as paper. Some of the changes due to comments from the approval meetings still need to be incorporated in the note. Documentation of the analysis code in the appendix needs to be completely. Should be done in a few days. 18/04/11 Photon + MET Editorial Board Meeting

20 Photon + MET Editorial Board Meeting
Paper Length Target The current draft of the paper is 6.5 pages not including the bibliography. About 34 lines can be saved by dropping one plot and reducing the height on two plots to make them match that of the GGM limit plot. 18/04/11 Photon + MET Editorial Board Meeting

21 Photon + MET Editorial Board Meeting
CDS Comments Blue: Comments from CDS Red: Response currently in CDS Black: Proposed response not yet in CDS 18/04/11 Photon + MET Editorial Board Meeting

22 Photon + MET Editorial Board Meeting
CDS Comments [1/7] 1) I am concenred that the pdf systematics are treated differented for UED and GGM models. I understand that the are reasons for this, but I think you should compute the uncertainties in a similar way. We have now tried to treat the PDFs in a consistent way between the two theory interpretations. [See slides 9 and 10 in this talk] 2) The statistical treatment of the limits is different btw the SUSY and the Exotics part. During the approval meeting I understood that you have decided to use CLs+b. Did you agree to use PCL or without PCL? If it is without PCL we will have difficulties with it, since the statistics forum recommendations are to use PCL with Cls+b. We are still iterating on this. We do plan to show the same limit types for both theory interpretations. A CLs+b limit has been calculated for the UED case, the limit plot still has to be generated. On the PCL vs. no PCL I am unsure. From the Stat Forum meeting on Friday it seems SUSY wants PCL 18/04/11 Photon + MET Editorial Board Meeting

23 Photon + MET Editorial Board Meeting
CDS Comments [2/7] 3) The bias/uncertainties due to material effects are estimated in a different way btw GGM and UED. Again I see a potential problem here which should be discussed with the Edboard. This has been unified. [Slides 4-5 in this talk] 4) If you have checked the debug stream, you should mention it in the backup materials. The debug stream has been checked, and no candidate events were found. The note will be modified accordingly. 5) The analysis code needs to be in a public svn area and well documented (not just a core dump). Please add an appendix to the backup material where you tell the reader where to find the code and its documentation. This is currently being worked on. E.T.A. of a few days. 18/04/11 Photon + MET Editorial Board Meeting

24 Photon + MET Editorial Board Meeting
CDS Comments [3/7] 1a) Were the GGM signal points produced with mass spectra consistent with the sum rules shown on pg 3? If no, why not? I recall CMS set all other sparticle masses to 1.5 TeV, and then allowed the gluino, squark, and chi10 masses to vary. I believe you do something similar (but just allow the gluino and chi10 masses to vary). The 2-d gluino versus bino grid obeys the GGM sum rules. All sleptons and squarks are set to 1.5 TeV which satisfies the sum rules. 1b) Do I understand correctly that you consider only gluino pair production? If so, is the only decay chain under consideration then the 3-body decay gluino -> qq chi10? Why do you not consider squark pair production? Correct, only gluino production with that decay is considered. Squark pair production is not included to avoid introducing a strong dependence of the limits on the squark mass. Goal was to decouple the squark masses. 18/04/11 Photon + MET Editorial Board Meeting

25 Photon + MET Editorial Board Meeting
CDS Comments [4/7] 1c) Given that final state signature (i.e. multiplicity of jets and other objects, and their kinematic distributions) is determined by the production mechanism and the details of the decay chains, have you investigated how robust your limit is to different scenarios? For example, the efficiency to pass the photon isolation requirement may vary significantly between the cases of gluino pair production and squark pair production. I believe this explains the ridge in the CMS plot of 95% xsec upper limit vs. squark and gluino masses. We have investigated this with some privately produced points that replicate the grid of the CMS analysis. Applying our cuts we did not see a great difference in the acceptances between the two production schemes. We saw some difference in the jet multiplicities between the squark vs. gluino production, but even when we implemented our best approximation of the CMS cuts the difference in acceptance between squark vs. gluino production was no where near as large as seen by CMS. Stats of the private samples were small. We should have official produced points to mimic CMS soon that could allow us to investigate this more. 18/04/11 Photon + MET Editorial Board Meeting

26 Photon + MET Editorial Board Meeting
CDS Comments [5/7] 2) Could the UED limit be presented in a plane of 1/R and N? N = 6 represents a favorable case in terms of the fraction of events that yield the gg+MET final state. However, one could still say something about the cases N = 2, 4, for example, but the reach in 1/R would be diminished. I raised this point in the 3 pb-1 UED analysis (comment in CDS), and the response was that this would be considered in the next update (i.e. this analysis). we use N=6 because it has the highest BR (~100%) for the region of phase space we are looking at. N=2,4 would give less stringent limits. > However, ... analysis). well... I do not remember saying this myself. what we would rather like to consider is a more general UED model, where the KK gluon and quark masses would be set free. also what I do remember discussion with the ed board is to have a 1/R versus Lambda plane for the exclusion. but in the end, discussing with theorists leaded to the previous suggestion. this suggesstion of having the discovery limit as a function of the KK mass and KK gluon. but in this case I would have to modify Pythia. this is in fact work in progress. 18/04/11 Photon + MET Editorial Board Meeting

27 Photon + MET Editorial Board Meeting
CDS Comments [6/7] 3) At the end of Jovan's presentation I asked why the GGM and UED limits are so different. For the record I was forwarded an by Helenka over the weekend from Martin Schmaltz saying that the difference can be associated with spin multiplicity factors. 18/04/11 Photon + MET Editorial Board Meeting

28 Photon + MET Editorial Board Meeting
CDS Comments [7/7] 4) I was a bit confused by the use of the eg sample with a loosened isolation criterion. Jovan stated this was done to increase the statistics. Can you take the MET template from MC (which should be well modeled as it is true MET from well understood processes), but normalize according to the eg tight isolation data sample and corresponding e->gamma fake rate? This would make the main plot on pg. 19 look a bit less statistically limited and not introduce any potential isolation bias. This has been addressed. [Note this is on slides 6-8 in this talk] 18/04/11 Photon + MET Editorial Board Meeting

29 Photon + MET Editorial Board Meeting
Backup Slides 18/04/11 Photon + MET Editorial Board Meeting

30 Old Background Estimate
Old Numbers: Updated Numbers: 18/04/11 Photon + MET Editorial Board Meeting


Download ppt "Diphoton + MET Status Update"

Similar presentations


Ads by Google