Comments by M. Missiroli ------------------------------------------------- Type-A ------------------------------------------------- l9: scalars -> scalar l16: "O( TeV)" -> remove spacing after first parenthesis l137: "jets [..] dominated by [..] reconstruction failures" -> 'affected' might read better than 'dominated' here Table 1, caption: "full cut-based" -> suggest to drop "full", as you refer simply to "the cut-based analysis" throughtout the document l274: 1 -> 1% l338: "the SM cross section" -> "its SM cross section" l442: "the tttt from" -> "the tttt signal from" l452: fix the newline in between the inequality "H < 0.12" ------------------------------------------------- Type-B ------------------------------------------------- General: in most of the document, the cut-based and BDT analyses are presented on equal footing (for example, in the discussion of the results and the first part of the summary). At the same time, it seems that the BDT-based one is considered to be the best result, based on the slightly higher sensitivity wrt cut-based and the fact that only the BDT results are used for the BSM intepretations. It would help to state clearly in the document what is considered as the main result (BDT) and what is the cross-check analysis (cut-based). l35: not sure what is meant here by "The analysis [..] improves [..] by taking into account [..] running conditions"; might be clearer to write something along the lines of "the analysis [..] improves upon [..] thanks to upgrades to the CMS detector and the large increase in integrated luminosity". l85-100: not sure this discussion belongs in this initial section; it's not a "standard" CMS correction, it is based on Data in a phase space specific to this analysis and only applies to some of the simulated backgrounds; this paragraph could be moved to backgrounds (Section 5). l183: for the HT-based triggers in 2016, why is an online-HT cut at 300 GeV "fully efficient" for your offline selection, given that the same threshold of 300 GeV is used offline? l222: "for the purposes of plotting and extracting the results in the analysis.": suggest to drop if, indeed, the backgrounds in each of the 3 categories are considered as one for all intents and purposes (in general, it sounds strange to mention 'plotting' explicitly) l253-257: you go into detail on what are the values of these corrections in the different years, but there isn't any explanation of why the values are different in 2017/18 with respect to 2016. Suggest to either give less details, or add an explanation of the difference. l279: are the ISR/FSR corrections derived separately for 2016, 2017 and 2018? This could be specified in the text. l284: is it entirely justified to apply this ttbb/ttjj correction to ttV (ttW, ttZ, ttH)? does that improve Data/MC agreement in certain distributions, or CR-yields, of this analysis? l294: you write "allow [..] to constrain using data" but you also use a prior uncertainty of 40%? How's the latter value picked? Were other priors (or even free-floating parameters) tested, i.e. is it known how much the value of these priors affects the final sensitivity to tttt? l324: did you test how the choice of correlating the ISR/FSR correction across years affects the final results? This partly goes back to the question of whether the corrections were determined independently in the 3 years or not. The ttbar MC Tune in the latest 2016 MC (TuneCUETP8M2T4) was different than in 2017 and 2018 (TuneCP5), which would suggest decorrelating the ISR/FSR corrections at least between 2016 and 2017+2018; are the correction factors measured in 2016 different from the ones in 2017/2018? l343: "affected" -> "significantly affected" l346: is the pre-fit ttH normalization taken from the SM prediction? For the prior, you used the uncertainty from the CMS ttH observation result (l296) and it's not clear enough here what the 0.3 refers to: is it ~30% on the normalization of the ttH background (so similarly to the prior you use), or is it the post-fit uncertainty of the nuisance parameter relative to the prior (so 0.3 of the prior unc.)? l456: "followed by a decay to ttbar" -> if this decay refers to the scalar particles, would move it into the first part of the sentence for clarity "For a new scalar [..] with m [..] and decaying to ttbar, [..] is probed."