For the Loose Muon Selection (L207-210), do you require the muon to be a tracker muon or global muon, or anything?

COMMENTS (RED: Not Dealt With. YELLOW: In Progress. GREEN: Finished)

Sal's Comments:

1. Top-tagging : Since many analyses will now use the CA8 CMS top
tagger, perhaps instead of outlining this in many different analysis
notes, it may be useful to have one that everyone refers to. Perhaps
this is something to organize with the JME POG?

- Ok, nothing much we can do about this at the moment.

2. As I've mentioned in the HN, please be sure to organize and present
your BTV-related studies to the BTV POG during the approval process.

- Kevin gave a a presentation today to the B-tagging POG and our method was approved (on the basis that it was "conservative nad therefore acceptable).

3. You can probably use the high-mttbar MC instead of the standard
TTJets MC to flesh out your MC statistics.

- We will not add this right now.

4. Your uncertainty on your QCD estimate is described as a
normalization uncertainty only. However, shouldn't there be a
shape-based component there? Or is this what is already done (in which
case, please describe it well in Section 7.7)

- This has been fixed now.

Specific comments:

Line 64: You are using a PFJet trigger at 320 GeV, not CaloJets,
please indicate this here.

- Revised the sentence. (This was an artifact from the 7 TeV paper.)

Line 67 : What Parton shower MC (including tune) is interfaced to
CompHep here? (See below).

- Pythia TuneZ2star

Line 84: Aren't you using the same prescription as B2G-12-005? IIRC
they are using Z' signal MC, not ttbar MC, so what exactly are you
using here? shouldn't this be done on the W' MC? Your cuts on the jet
pt are pt > 450 which is far below the trigger turnon here, but you do
a trigger weighting which is okay, but please describe the
uncertainties you use here as well as in the systematics section.

- We have reworked the trigger section.

Line 105: As in B2G-12-005, please update to the new AK7 jet
corrections instead of using AK5. These will be closer to the "truth"
for the CA8 jets and will probably reduce the uncertainty a bit.

- This was a typo. Fixed to R 0.7.

Line 111-121: You should motivate where all of the following numbers come from :
- pt1 > 450
- pt2 > 370
- (You do the deltaY < 1.6 in Section 4.5, so also say "The
description of the optimization is given in Section 4.5" or

- Fixed: We now refer to the sections where this information is presented.

Figure 3: Remove the markers on the black histogram, or put the marker
in the legend so that they are consistent, otherwise it "seems" like
these are data

Line 157 : Add the reference for this information.

- Fixed.

Line 169 : What about the other 82000 CA8 jets that *don't* match to
an AK5 jet? The deltaR < 0.005 is far too artificially small, please
use deltaR < 0.5 for this.

- this has been changed to reflect the presentation to the BTV meeting today.

Figure 6,7 : Needs a legend.

Section 5: Please make sure that this is tied to the B2G-12-005
approval also (see above)

Section 5 : What is the parton shower MC used for the signal MC? The
SF determination uses the PYTHIA generator, so you need to be
consistent between the two. Otherwise see EXO-11-095 for the correct

- We are using PYTHIA.

Line 262: As in B2G-12-005, please use the top-tagging SF instead of
the W-tagging SF in the end. Note what the value is here, it will be
useful to see this ASAP.

Figures 21-22 : Legends

Line 340: The JEC uncertainty is not 5%, it's eta-phi dependent and
then you include a flat additional uncertainty as in B2G-12-005.
Please ameliorate the description to coincide with that analysis.

Line 400: Are you really still using CLs? If so, switch to Bayesian
and it will go faster. If not, please change this to what you are

- Fixed.


1) You are only setting limits on a right-handed W'. Are you actually sensitive to the right vs left-handedness? And would you be able to set limits on the a^R from Eq 1 as well? (or even better: would you be able to present a limit in the a^R vs a^L plane)?

- We haven't studied the reach for the left-handed W'. We wanted to

establish the procedure first, and get the right-handed W' result
out as a proof of principle. For the paper, we would like to do
both left-handed W', as well as the generalized coupling analysis
a la EXO-12-001. If possible, we'd also like to improve the top
tagging (in collaboration with JME jet algorithms group).

2) Single tops: your signal is of course essentially equivalent to a (s-channel) single top decaying all hadronically. As such, I am very surprised that there is no mention of this contribution at all in your note. As a back-of-the-envelope calculation I just did no the plane (so please check me, no guarantees) suggests that the xsec times BR (of the order of 10 pb) for s-channel single top would be substantially larger than the signal cross sections in Table 2 and only a factor 10 smaller than the ttbar contribution, I am wondering the following:
2a): are you able to set a competitive limit on the cross section limit for s-channel single top production using the low mass region? Can you see it in some distributions? (figure 20 for example).
2b): what happens whe you actually start including single top production in your (MC) background estimates? Does this change your limits drastically?
2c) I would minimally (if 2a and 2b are indeed completely negligible) expect a short sentence mentioning this in section 4.1, where you (very) shortly discuss the possible interference between standard model single top and possible signal. I would like to request that you expand this explanation.

- We have added single-top to the analysis.

3) Triggers: In Sec 4.2 you mention you make the trigger turn-on in MC match the data, which is of course the right thing to do. What I am missing in the documentation is a turnon-curve for the data (or even better: some form of a closure test that you didi the right thing, but minimally the turn-on curve for both data and MC, not just only MC).

- This has been dealt with. We just need to add the new plots to the note.

4) B-tagging: I second the feedback you already received from Ivan and Devdatta regarding the presentation (in sleep-inducing detail if that is what it takes) in the BTV POG. Actually I explicitly request that you make such a presentation before we call the preapproval and get your fake rate parameterizations approved by the BTV POG. Some more questions regarding those parameterizations and the efficiency:
4a) as you have made the cross checks on the b-tagging efficiency anyway, please give some more quantitative detail, such as the sample composition (or fraction of tagged/untagged jets) discussed in Sec. 4.6. Specifically, I would like to have more elaboration on whether this cross check on ttbar MC would still hold for the signal.

4b) Looking at your tag rate functions in section 6 I am wondering if you could not choose some more optimized functions for the fits, as the functions seem to diverge from the data at very low and high values and undershoot at jet pT values around 500-700 (where you still might expect some signal so having a good description there is worth taking the effort). The same comment actually holds for the undershoot of the fit at low values in Figure 18. These should be improved or minimally the effect of the discrepancy on the background estimation and expected limits should be discussed in substantial detail.

5) are you sensitive to the step5 tracking reco bug? I saw you made the Delta y cut (sec 4.5) but as this is a CMS-internal document I would also like to see if this cut manages to actually cut away any tracking features or really just only cleans up the S/sqrt(B) distribution.

6) there is still a very bad agreement between data and MC in the subjet mass drop variable (I recognize these plots from B2G-12-005 and guarantee you that I will expect this question to be answered during the approval of that analysis so beware): What is the cause of this disagreement?

- This is something that is being dealt with. No changes to the note need to be made at this time.

7) when do you intend to include the full 2012 dataset?

Ulysses's Comments:

Where did the tight muon selection come from? It's similar, but not identical to the tight muon selection recommendedby the Muon POG.

-The muon selection here is from the SHyFT measurement of the ttbar cross section. The same selection is used in B2G-12-005

L194: What does "Not contained in a sub-leading primary vertex" mean, exactly? Does it have to be contained in the leading primary vertex? And what defines 'contained'? Same for L209.

-The wording here is a bit confusing. The muon is from the highest Pt primary vertex. The next version of the note will make this clear

Do you require muon to be isPFMuon()?


L201: For the Isolation, do you use the delta beta corrections? Same for L210.

-yes. The next version of the note will make this clear

For the Loose Muon Selection (L207-210), do you require the muon to be a tracker muon or global muon, or anything?

-Either tracker OR global. The next version of the note will make this clear


1. AN line 136, figures 3 and 4: Do you refer to the ttbar MC sample with "QCD Monte Carlo" or are you comparing the signal to multijet QCD production (as listed in table 8)?

2. Table 6: The W+jets cross section has changed from 36257 pb to 37509 pb on the cross section twiki page just recently.

3. Chapter 5: How do you treat single top and Z+jets processes in the semi-leptonic sideband study? Are you including MC simulations for these processes? Do you assume that their contributions are covered by the ttbar and non-W templates in the Ht fit?

4. PAS: I would mention how the signal MC sample was produced and add a reference to CompHep


In the introduction, it would be good to add some references to the specific models predicting W'->tb.

Section 2 / Line 34-35 are confusing -- the highest pT jet is used to define the hemispheres, correct? Any other jet within dPhi < pi/2 is assigned to this hemisphere, while jets with dPhi > pi/2 are assigned to the second hemisphere? Please clarify the text here.

- Clarified in text.

Section 4 / Line 63 -- do you use a CaloJet or PFJet trigger?

There's a typo in table 5 (signal cutflow/, mW' = 1500, Nsubjets > 2 entry).

Have you investigated other functions for the b-tagging rate parametrization? It seems in some eta bins, the fit does not always describe the data adequately, nor are all the data points within the errors of the fit parametrization.

Can you show plots of the eta distributions of the top and b candidates? It would be good to confirm there is not an obvious bias due to the TOBTEC tracking problem.

-- We can produe these plots to check, though it seems strange to put them in the note since we will not be bringing up the TOBTEC error. We will have this available as a backup slide for our talk.

Please change to the eta-dependent smearing factors for the jet energy resolution systematic. This will also affect the default smearing applied to the nominal templates.

Is the jet energy scale systematic always symmetric? If not, please report the up and down variations in table 9.

Some of your systematics have a very small effect (jet angular resolution especially). It might save CPU time when calculating limits to just neglect this systematic.-

- We can do this, but for the moment we will leave it in rather than change our workflow at the last minute.


l19: please update to a more recent CMS published result: (the 2012 result is about to be submitted as well)

l63-64: I assume you are talking about the trigger here. But the trigger uses PF jets - please fix.

-this is not about the trigger

l66: since you estimate ttbar from MC, why not use the larger statistic POWHEG (NLO) sample ? In any case, have you evaluated if NLO effects impact your analysis ?

somewhere in Sec. 4 you should mention that your samples are PU re-weighted (are they?), add a plot showing the primary vertices' agreement in data/background after re-weighting. BTW, I did not see a systematic for pileup in your list - that should be added.

Table 1: you should mention here and wherever x-sections are listed if they are NLO or LO. Especially since here they are NLO while in Table 2, they're probably LO if straight out of COMPHEP ?

Table 2: assuming these are LO, are you using a k-factor in the analysis ?

l82: when you talk about implementing the generalized coupling analysis in the future, do you mean post-Moriond for the paper or even later ?

l87: can you please add a plot which shows the matching i.e. the turn-on in data and the trigger efficiency in signal MC.

- this is being dealt with from another comment.

l136: this should be Figure 3.

- Yes. This is fixed now.

Figure 3: can you please use dashes for some of the lines with very similar shades of blue or green - otherwise, it is difficult to distinguish even in a color printout. Also, this color scheme is different from that of Fig. 5. Please fix and use only one. Also, remove the dots from the QCD. And explain the second peak at low min pairwise mass for some of the mass points.

Section 4.5, Table 5: the improvement in expected limit - is it in the entire mass range ? Can you add a plot ? What does this cut do for the masses around where you are setting limits ? Is it worth checking if you should use different cuts for low/high mass ? (Fix typo in the 1500 GeV row)

- I don't understand the question. Table 5 is multiple cuts, some of them without which a limit is impossible.

Fig. 6: what is the difference between some of the plots here (say min. pairwise mass, top candidate mass) and those in Fig, 3 [aside from QCD not being shown]. The normalization looks different for one.

l214 -- l220: please justify/motivate your cuts

Fig. 8: what is the lumi used here ? Please add the usual CMS Preliminary, cms energy etc.

l224: do you really mean HLT_IsoMu40_eta2p1 here ? There is a lower unprescaled HLT_IsoMu24.

- Yes, it is the eta2p1.

Fig. 9, 10: please add lumi information to the plots

Fig. 11: the agreement is not good - especially for mu < 0.4. You need to add an explanation/comment.

Fig 13: the agreement is not good - especially in the region between 200-300 GeV. Please comment.

Fig. 18: for low eta and transition eta, can the fit for the tag rate be improved for pT between 400-500 GeV ? In any case, you should comment on its impact.

- We are fixing a bug in this section. Most of these comments will (hopefully go away).

Fig 21, 22: you should probably add the background uncertainties to the plot. The agreement is not very good for some of the plots - eg. top candidate pT, b candidate pT, pT of the tb system etc. For the latter the peak in data appears to be shifted wrt to the background,

Systematics: you should add systematic due to PU, update prescription for jet pT smearing as per the latest recommendations.

- We are dealing with this.

Also the lumi uncertainty is 4.4% and not 5% (and add reference for it)!

Fig. 31: please add description of theory line to the legend. Also, mention if k-factors are included (you should use them if not included already). And use standard CMS color scheme (green, yellow).

Topic attachments
I Attachment History Action Size Date Who Comment
PDFpdf AN-13-004_temp.pdf r1 manage 1331.7 K 2013-01-15 - 18:03 UnknownUser  
PDFpdf AN2013_004_v2.pdf r1 manage 1328.6 K 2013-01-05 - 07:14 UnknownUser  
Edit | Attach | Watch | Print version | History: r13 < r12 < r11 < r10 < r9 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r13 - 2013-01-15 - unknown
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    Sandbox All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2020 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
or Ideas, requests, problems regarding TWiki? use Discourse or Send feedback