Response to comments for AN-20-100

General comments:

Many plots, descriptions of systematic uncertainties, etc. appear to only be shown for 2016 and 2017, please add the 2018 equivalents as well, since that is what this iteration of the analysis adds. This also leads to the question : are the 2016/2017 plots corresponding to the re-analysis (with UL dataset) or are they the old plots from the previous iteration of the analysis?

• For distribution, limits plots, 2018 dataset is added and UL dataset is used in all years.

Some section/table/figure references appeared to be broken. Some examples are L61, L108, L110, L213-214, L225-226, L261, and L283. Certain parts of the text also appear to be broken (eg. L210-211, L281), please correct.

• Fixed broken references.

Some symbols appeared to be missing, maybe due to failed macros. Examples include L107 ("of is"), Tables 7-10, L210-211 (presumably should all be Z'?), L260 "()", Table 15 (for QCD, PDF, and MC stat. Maybe these should be qq->Z?), Table 16 ("80 << 100 GeV"), and L307.

• Fixed missing symbols.

Please be sure to label figures as "CMS Work in Progress" as described in https://twiki.cern.ch/twiki/bin/viewauth/CMS/PhysicsApprovals#Thesis_endorsement

• Label added for distribution and limits plots.

Detailed comments:

Abstract: "These are the first dedicated limits on the Lμ − Lτ model at the LHC." - what about EXO-18-008? Actually, are the upper limits listed here for the full Run 2, or are they from EXO-18-008?

• Sentence fixed.

Section 2: Please add more information about the samples used, e.g. which global tags, which lumi JSON file is used, etc.

• lumi JSON file is added.

Table 1: It looks like the period H datasets are listed along with period F (or should the "H" have said "F" there?). Please double check the datasets being used for 2016 based on https://twiki.cern.ch/twiki/bin/view/CMS/PdmVDatasetsUL2016, e.g. Run2016B ver1 and Run2016B ver2 should both be used.

• Typo fixed.

Section 2.1.2: For your tag-and-probe method, what selection is used? You mention that events with four muons are used, but are any other requirements applied, e.g. Z mass constraints, muon pT, etc.? It sounds as though the same probe muon can added to the denominator of your efficiency multiple times (i.e. once for each tag muon). If so, this may be a source of double-counting that you don't actually want. Please clarify the text.

• I didn't reproduced this part.

Please add description for 2018 as well, and and figures as in Figure 2. Are the plots in Fig 2 corresponding to the updated UltraLegacy datasets?

• I didn't reproduced this part.

L81-82: "for 2016, the last 6% of the dataset was not used for the trigger efficiency measurement" - please motivate why in the text

• I didn't reproduced this part.

L86: "ggH" - Is this ggH->ZZ*->4l, or some other process?

• Yes.

L89: "applies the efficiency" - applies the efficiency itself, or efficiency correction factors?

• Apply correction factors.

L97: Please motivate these uncertainty values and how they are determined. Are they fixed across years?

• I didn't reproduced the uncertainty part.

Figure 2 and elsewhere: please double check the integrated luminosities you quote against those from the Lumi POG (https://twiki.cern.ch/twiki/bin/view/CMS/TWikiLUM#TabLum)

• Figure 2 is directly from AN-17-321

Section 2.2: Please add the to the Table 10-12 the remaining path of the MC sets as part of captions as done in tables 7-9. Are all the samples used produced using UL reconstruction corresponding to the datasets used?

• Path added. Background samples are using UL.

Section 3.1.2 Muon Isolation: Is there a reason that you use a PF isolation cut value of 0.35 rather than one of the values corresponding to the WPs from https://twiki.cern.ch/twiki/bin/view/CMS/SWGuideMuonSelection#Particle_Flow_isolation? Do you derive data-to-MC scale factors for the muon isolation cut that you use?

• Just used the same isolation cut as AN-17-321. I didn't derive the scale factors.

L163: Which Loose ID do you refer to here? The "loose muons" from L123, or the muon POG loose WP? Please clarify.

• Loose muons from L131.

L210-211 : some of the text seems to be missing, please add the missing text in the brackets to complete the description.

• Missing text added.

Table 13: What does "total events" correspond to? i.e. which luminosity does it correspond to, does it only correspond to a particular background process, etc.? Table 13: Please describe the mass window size of 2% in the analysis strategy section; currently it doesn't appear until Section 6.2, so it isn't quite obvious from this table currently.

• This table is from AN-17-321. Deleted from the update version.

L213-214: "KNNLO" - to be sure, is this a ratio from NLO to NNLO, or LO to NNLO? It should be the former, since you already apply KNLO. Please specify in the text.

• It is NLO to NNLO. Added explaination text.

Figure 3: Text in captions (of subplots as well as for the plot) is missing, please update after adding missing text.

• Missing text added.

Figure4 : 2018 plot is missing. Is the bottom plot cross section, or cross section x branching ratio? The axis titles and captions are in contradiction, please clarify in the text captions / figure axes, Figure 4: 2018 seems to be missing? The bottom plot is labeled as 35.9 fb^-1, but presumably applies equally well to all 13 TeV datasets. For the top two plots: How do you do this interpolation between 40 and 50 GeV, where the Z1 vs. Z2 choice changes? It might be safer to generate masses near the threshold rather than to extrapolate here. Figure 4 "Left: central line is shown in black while the red and blue color shows interpolation ..." - is this referring to the red and green lines in the top plots, or something else? Is the interpolation in the top two plots linear, or based on some fit function? The systematic variations described in this caption are not apparent.

• Plots of three years are added.

Section 5 Systematics: Please describe the uncertainties relevant for 2018 as well.

• I didn't reproduced the uncertainty part.

L232-233: "7% if the lowest pT muons if less than 7 GeV" - how is this used? It sounds like a flat 4.9% uncertainty is used, which is smaller than this value?

• I didn't reproduced the uncertainty part.

L236: Please provide more information, as well as the impact of these uncertainties. Or are L240-246 also intended to describe how you assess this uncertainty?

• I didn't reproduced the uncertainty part.

L238 "We neglect the PileUp uncertainty of 2%" - are you sure it is negligible? It's of a similar size as your luminosity uncertainty, and many of the uncertainties are fairly small.

• I didn't reproduced the uncertainty part.

L252: "root mean square" - The difference is likely small, but I believe the current PDF4LHC recommendation (Section 6.3.1 of https://arxiv.org/pdf/2203.05506.pdf) is to take the central 68% interval over the variations.

• I didn't reproduced the uncertainty part.

Figure 6: It is hard to tell if there are differences between the nominal and the variations; it might be easier if the three fits are overlaid (e.g. in different colors) on the same canvas.

• I didn't reproduced the uncertainty part.

Table 15: "Interference effect" - please describe this in the text as well.

• Added a interference section.

L264-267: Although the impact is small, it would be good to treat the correlations properly.

• I didn't reproduced the uncertainty part.

L281-283: It doesn't sound like a single "bin", given the windows used. Please clarify. Certain parts of the text seem to be missing, please fix.

• Missing text fixed.

Table 16: Which dataset is this? It would be good to see for each of 2016, 2017, 2018, and combined. Table 16: "uncertainty" - no uncertainties appear to be shown?

• Full RunII dataset. And uncertainty is shown.

Figure 9: Seeing 2018 would also be interesting here, since that is what this analysis is adding on top of the previous result (at least plots of the background estimate and predicted signal).

• A new set of plots are added, including all three years.

L322-325: As mentioned in the meeting on March 16th, it would be useful to overlay the limits from the 2016+2017 analysis with those from full Run 2, to better see the gains that 2018 brings.

• A comparison on Figure 30 and Table 17.

Figure 14: Please explain the large drops in the observed limits and 95% bands in the right-hand plot for many mass hypotheses.

• That was using full CL method and is removed from update version.

Figure 21: Does this include the 2018 dataset? It would be interesting to see the expected limits in the right-hand plot for full Run 2 as well.

• This one is only 2016-2017 and removed.

Editorial comments:

L52: "during 2016 and 2017" - 2016--2018

• Typo fixed.

L53: "(2017) 59.7 fb−1 and (2018)" - typo, should be "(2017), and 59.7 fb−1 (2018)"

• Typo fixed.

Table 3: Missing dashes between run numbers

• Dashes fixed.

L100: "Production of via" - missing a word between "of" and "via"

• Missing word fixed.

L107: "leptophilic coupling to muons (gμ = 0.1)" - since you discuss alternative coupling values in the results section, it might be useful to already mention here that alternative couplings will be discussed later, so that the reader is aware that they will be considered.

• Added a sentence to mention it.

L123: Please add units for the pT, dxy, dz

• Units added.

L146: "subtraced" - typo

• Typo fixed.

L209: subscripts 1 and 2 are in the line but the text before the subscript (Z ?) does not appear.

• Symbols fixed.

L230: "taste" -> "years"?

• Typo fixed.

L235: This line may be a mistake?

• Line deleted.

L309: "FEWZ and " - missing something after the "and"?

• Typo fixed.

-- KunShi - 2022-04-19

Comments from slides:

Discussed during the talk: Limit plots say cross section, it should be cross section X branching ratio. Please fix in the AN.

• Fixed the label.

From your talk: Slide 14: is there a point at 45 GeV? If so, please note it in the figure caption, since it is hard to see in the AN. In the long term, can go for more granular mass samples near the transition area.

• Yes, there is a point at 45 GeV, it is completely covered by the fitted green line.

Discussed during the talk: limits comparison: please provide an overlay plot and table for expected limits. Both for full Run-2 vs previous limits and as a cross check for re-analysis of 2016/17 vs previous limits.

• Added cross check plots. Provided the comparison side by side. Since I don't have the datacard from previous analysis, cannot plot the overlay plots.

AN Section 5: There are still descriptions missing for 2018, for e.g. (but not limited to) L305,L315, Table14, 15. Please add more description of 2018, even if it is just that the same values are used as e.g. 2017.

• Added some description of 2018.

AN Figures 12-14: What drives the phi distribution for the ZZ background events here? (i.e. why isn't it uniform?)

• Plots changed. Now are the real phi distribution.

Long term comments. (will be addressed in longer term)

General: For the full result, please study the uncertainties assigned and the trigger efficiency for 2018 as well as UL. Discussed during the talk: Slide 15: Please check if the interference has any dependence on the mass.

L95 : Was any check/study performed for 2018 also?

Figure 5: This plot and the corresponding text are a bit confusing. The SM Higgs width is 4 MeV, so what is the difference between the top-left and top-right plots? Is it that the width varies with off-shell mass? The cross sections between the two plots differ by four orders of magnitude, is that correct?

L189: In general the MET filters can be used even for analyses that do not use MET explicitly since the function of the MET filters is to remove mis-reconstructed events.

Figure 6: Ratios might be useful here to better understand the difference between the two distributions. Showing this on a linear scale may also be useful.

L279-280: "4.9% for 2016 dataset and 4.6% for 2017 dataset" - what about 2018?

Figure 7: Is the impact of the interference at all mass-dependent as well?

Figure 10: Which dataset is this? We suggested this previously, but overlays would be useful.

Figure 11: Hard to be sure looking by eye, but it seems like there are ~10 data events for pT > 140 GeV with a prediction of only 1 or 2 background events? Is this understood?

Fig 19: At the peak, the difference looks larger for 2018 compared to 2017/2016 -> is this understood?

Figures 21 and 22: Did these change at all with respect to the published result for 2016+2017?

Table 17: Difference at 40 GeV is ~30%, but much smaller (~10-15%) for the other masses. Is there a reason for this?

Fig 30: Including both sets of lines in one figure (overlaying past and current limits) would be helpful to estimate the overall improvement

Editorial comments

Abstract: "These are the update limits" -> These improve upon the limits / These are the updated limits

• Fixed

L154: "substracted" - typo

• Fixed

L187: Npv -> ndof

• Fixed

L211: May be useful here to mention that the Ref. [28] is for H->4L search (as one would naively have guessed that it was the AN for the 2016+2017 Z->4mu search)

• Description added

Figure 3: Labeling the axes (or at least mentioning in the caption) here would be useful, so we know which is m(Z1) and which is m(Z2).

• Added description in caption

L264-265: "Because the windows have to be narrow at low masses" - this seems to be the first mention of the window sizes in the AN? It would be good to mention this sooner.

• Mention the window at beginning of the AN

Table 14: Please label these uncertainties as percentages (assuming that is the case). Do you have the same table for 2018? -Table 15 caption: Please add description of what the Table contents placed in brackets refer to (presumably 2017)

• Label added for 14. I don't have the same table for 2018. Description added for 15.

L290-291: Please add extra information into the text to clarify this. Since this is the first mention of a dimuon selection, the reader may initially be confused since you've selected events with four muons everywhere else!

• Added description

L304-305: What about for 2018 samples?

• Added 2018 samples

Figure 20: Very fine bins shown on the right-hand plot at low mass; hard to see them very well. Could imagine an additional version of this plot with a log-x scale? Unsure if it would help...

• Fixed binning.

Fig 19 caption: 20162017 -> 2016, 2017

• Fixed

L473: "CMS-AN-16-2442" - should be "AN-16-442" (including the hyperlink here and for other references might be nice)

• Fixed bib.

-- KunShi - 2022-05-08