AN link: http://cms.cern.ch:80/iCMS/jsp/openfile.jsp?tp=draft&files=AN2016_364_v4.pdf

Comments:

Color code for answers to reviewer questions:

• Green -- we agree, changes to analysis/documentations implemented.
• Lime -- we agree, but the item hasn't been done yet. (Open item.)
• Red -- we disagree, changes to analysis/documentation is not implemented.
• Teal -- we agree, but we don't think any change to analysis/documentation is needed.
• Blue-- authors/ARC/conveners need to discuss. (Open item.)
• Orange-- not longer applicable.

$3 Dataset and MC Samples
. Can you add a table reporting the names and xSec used of the signal and MC samples?
. No need to say, but please add V+gamma samples when available

$4 Object Reconstruction and Event Selection
$4.1 Trigger
. Which event selection do you use for the trigger efficiency evaluation?
Does it resemble the signal region phace space?
Could you show the trigger efficiency VS the mass used for the signal extraction?
. Which correction factor do you apply to the MC?
It is a flat number or a function of photon pT?
There seems to be a trend vs the photon-pT and the trigger efficiency becomes 1 at high-pT values.
I understand the major background is estimated from data, even though it affects the signal.
Can you show photon-pT distribution for some of the signal points to which you are sensitive?
. Fig. 5
I am not sure I understand the error bar. It seems constant going at higher pT values.
Should not it decrease?

$4.2 Photons
. Why photons with only |eta|<1.44 and not higher |eta|?
. Do you apply the photon overlap removal in these samples?
There seems to be higher bkg prediction in these plots.

$4.3 Lepton Veto
. Which isolation do you use for muon and tau?
Could you specify it?
. Do you apply correction factors for the veto to leptons?

$4.4 Jets
. Line 228: do you mean Fig. 11?

$4.5 Final Event Selection
. If you have more than one jet passing the selection you mention, which one do you take?

$4.5.1 Top Control Region
It would be better to call this sub-section Ssignal and top control regions"

$5 Decorrelation of the N2 Variable
. Can you fix the ideas on Eq. 11 with some numerical values that you use for N2(5%) to decorrelate N2?
As I understand phi is function of jet pT, rho, even though this is not clear from the text.
. Fig. 15 you are missing the labels (jet pT and rho, I think).
. Line 268: 5/95 comes by construction (ideally).
Which error would you assign to it, if you used this factor in the final estimation?
. Line 268: in which region is performed the closure test?
. Fig. 18 Would it be possible to add a legend in the plot?

$6 Background Estimate
. Could you comment on the fact that QCD and gamma+jet background have the same shape?
Have you checked this in MC?
. I am not sure why you need to decorrelated N2 if N2DDT requires anyway that the transfer factor is pT,rho dependent.
Could you comment on it?
. Do you include any signal hypothesis in the simultaneous fit?
Which is the signal contamination in the failing region?
How are the fit results affected by the presence of a signal?
Could you comment it when using a Z' of 10 GeV (where the effect would be higher)?
. Could you also clarify about the role of V+gamma processes?
Could you remind how you consider them at the moment?
How much do they count in the fit?

Francesco Santanastasio

## main comments:

L67-L71: is this an official recipe? is there some pT cut on the photons?

What do you mean by outgoing parton? I imagine the overlap also depends on

the minimum pT of the photons generated in gamma+jet Madgraph sample.

Please provide a reference where the procedure is described.

L75: does gB=1 correspond to a narrow resonance? IS this in agreement with other

boosted dijet searches?

Figure 4:

- on x axis you have the reconstructed photon pT after full ID and energy

corrections applied? please specify this in the AN as well.

- do you understand the reason for the slow turn-on? Is it due to photon ID or energy resolution,

or trigger issues?

- extend the plot above pT of 450 GeV, there seems to be a drop at 400-450 GeV?

L112:

- why restricting to the barrel only? Did you optimize this cut? Should be discussed.

- where does the 14 GeV threshold comes from?

Selection:

please provide a cutflow table for signal events for all the selection indicated in Section 4.

In particular I would like to understand the impact of lepton veto on the signal (electron and tau cuts

could be loose and killing some signal).

Section 4.3:

how did you decide the pT thresholds for the lepton vetoes? [10,10 and 18 GeV ]

L160: what is the logic to decide the boundaries -2.0 and -7.0?

They are different from what used in other boosted dijet searches.

Motivate the choice in AN as well.

Figure 10: please explain why at 20 GeV the tau21 looses discrimination power.

Is it because the decay products of the resonance are too boosted to be resolved?

The jet mass in Figure 9 instead still keep a good discrimination power.

L243:

- Is it DeltaR <1.5 (instead >1.5) ? Would be good to see this DeltaR

distribution for data, signal, QCD, gamma+jet and ttbar background.

- Do you allow that the jet from the resonance decay can be b-tagged?

(this possibility should be kept since scalars might decay preferentially to b quarks)

L255: missing figure

L256:

- missing figure

- is the MET cut 75 GeV or 60 GeV (see L242)?

L260 (about Figures 28 and 35): not sure these are the Figures you wanted to show.

After L265 and Figure 15:

- “We will use ρ instead of mass as it leads to smoother distributions for N2(5% efficiency).”

Sentence unclear.

- What is shown on x and y axis? rho and pT?

Figure 15:

if the problem is the limited statistics, why don’t you use larger bins?

Figure 17 vs 21:

can you explain the difference between these two plots?

Why can’t you use Figure 17?

Section 5.1:

- what is the difference between the scale factors discussed in Section 5.1 (0.847 +/- 0.19)

and those discussed in Section 5.1.1 (84.8 +/- 0.05)?

- Are these scale factors applied to signal or ttbar background? Not clear in AN.

Section 5.1.1:

is this exactly the same of EXO-17-001? I would specify this in the AN.

L342: the choice of the function (and the number of parameters)

should be motivated. Did you perform F-tests and looked at

chi2 distributions as done in other boosted dijet searches?

Maybe you should describe here that part or move this section later.

Figure 23:

- not clear what is this fit: for example

which are the different components for signal and background, which is the likelihood used.

- What is the composition of this sample? What is considered background in this fit?

Is the N2DDT conversion still valid here?

- In the end it is unclear if you are using this fit in the final results.

Figure 34:

- what is the blu line in the pre-fit? Just data in the failing region multiplied by the MC-based

transfer factor?

- there is a peak at 60 GeV in the failing category? This gets reflected also in the background estimation for the signal region (despite some smoothening due to the fit). Why the mass distribution

is not smooth in the failing region?

- what is the expected signal contamination in the failing region?

Section 6.5:

could you please clarify better the content and explain the plots shown later?

I do not understand why 3x3 is a good choice. Also the chi2 of the fit in Figure 39

doesn’t look very good for this small dataset. With the full dataset it might be worse.

Figure 40:

the bias study should be performed in a more realistic way, using for example the MC

to generate events instead of generating and fitting using the same functional form.

Figure 42: what is the observed line here?

========================

## minor comments:

- run spell check.

Title:

sqrt = 13 TeV

L1:

This is actually the version v4. Please update.

Figure 1: use the most recent plot from EXO-17-001 with full 2016 dataset

L50: below 200 GeV -> above 200 GeV?

L63: add table with all background MC samples used

L65: provide references to EWK and NLO corrections

Table 2: this is no data; fix the caption.

L84: missing table.

Explain here what you are doing at the moment to account for these missing MC samples.

How do you know that these backgrounds are small?

L89: missing table

L98: what is HE? add in the AN.

L114: add reference to ID and energy corrections applied

L142: add reference to JEC applied (probably lines 166 and 168 should go here, or

you can move completely the PUPPI part in this section when you first introduce jets)

L153: add reference to btag scale factors used

L192 and Figure 9: extend the plot up to 250 GeV then.

L228: Figure 13 -> Figure 11 ?

Figure 14: caption says blue lines.. I see red. Z axis is also missing.

Section 5.1: add a /clearpage in the latex file

L373: not sure these are the Figures you wanted to show

-- MarcOsherson - 2017-12-20