Thomas-Boettcher-WG-identified pi/K/p in pPb @ 5 TeV

Thomas-Boettcher-WG-identified pi/K/p in pPb @ 5 TeV =========================================== General: I made the same comment for the 8 TeV analysis, but I really think for this important measurement we should try to get as much physics out as possible. For this analysis I think this means two things: 1. Can you provide covariance matrices, at least for R_pA? These will be necessary if these measurements are going to be used in nPDF fits. Added the covariance matrix study in section 9 :::success Done. ::: 2. Is it possible to extend the pion measurement down to the pion RICH threshold? And the kaon measurement to the kaon threshold? No, it is not possible to extend the pion measurement down to the pion RICH threshold, nor the kaon measurement to the kaon threshold. This is because the calibration samples used in this analysis do not cover the full momentum range needed to reach those thresholds. In particular, all three calibration samples (for pions, kaons, and protons) are required to overlap in the kinematic region of interest to properly build the efficiency and misidentification matrices. The lowest momentum range where a reliable and precise efficiency estimation can be performed corresponds to the proton threshold, which sets the lower limit for all species in this analysis. :::success Done. ::: l. 296: Do you check the PV position distributions for stability over time? Yes, the stability of the PV position distributions over time is checked. This study is added to Section 4.3.2, where the primary vertex projections along the $x$, $y$, and $z$ axes are evaluated as a function of the number of events. The distributions of $x_{\text{PV}}$, $y_{\text{PV}}$, and $z_{\text{PV}}$ are shown for the pPb, Pbp, and pp datasets. :::success Done. ::: l. 309: I guess you maximize this because you want to consider the tradeoff between statistical and systematic uncertainties? Is there an easy argument for why this is the best choice? Maximizing the product of significance and purity helps balance statistical and systematic uncertainties. The significance term reduces statistical uncertainties by increasing the signal-to-background ratio. The purity term minimizes background contamination from fake or secondary tracks, reducing systematic uncertainties. :::success Done. ::: l. 324: Is it really true that the DLL variables don't rely on simulation at all? This has been clarified in the section by specifying that the variable is defined based on the detector response and that simulation is only used for calibration and validation. :::success Done. ::: Table 9: Your PID requirements don't necessarily create disjoint samples for the different particle species. I think it's unlikely that there's any overlap, but can you confirm that this is true? Even if there is no overlap, it might be good to tweak the selections to make the samples disjoint by definition. The table has been corrected with the correct values. The potential overlap between the different particle species samples was checked using a Venn diagram, and no overlap was found, ensuring that the samples are disjoint by definition. :::success Done. ::: Figure 5: It might be nice to draw your cut values on these plots. Added. :::success Done. ::: Figure 12: Are the vertical axis labels correct here? Isn't this the corrected f^h_fake defined in eq. 18? Corrected f^{h}_{ghost} to f^{h}_{fake}. The systematic uncertainty in pp is also corrected. :::success Done. ::: Section 5.3.3: It would be good to see this study repeated in MC to make sure the efficiencies you measure here actually correspond to the "true" efficiency. Since the RICH is not properly simulated in the MC, we cannot directly apply the efficiencies obtained from the calibration samples to the MC. Instead, we opted to reweight the PID variables and the event multiplicity to better match the data before performing the closure test. We are currently finalizing the reweighting and will include the results in the second version of the analysis note.. Corrected f^{h}_{ghost} to f^{h}_{fake}. The systematic uncertainty in pp is also corrected. Figure 25c and similar: Didn't you say you fit DeltaM? Can you include that fit projection? Added. :::success Done. ::: The procedure laid out here seems reasonable to me, but I think we really need to see a closure test in MC to be sure that it works as expected. As mentioned, we are finalizing the reweighting of PID variables and event multiplicity to better match the data. Once this is completed, we will perform the closure test and include the results in the second version of the analysis note. l. 755: Do you include the efficiency of the p > 10 GeV cut here? No. The efficiency is calculated for the ghost probability and the(pseudoIP) criteria within the momentum range of interest. :::success Done. ::: Section 6.1: Of course this needs to be filled in. This section will be removed it doesnt make any sense since the systematic uncertainty on the occupancy is included in the reconstuction efficiency. :::success Done. ::: l. 912-913 and referenced figures: It'd be good to see a breakdown of the different contributions to these uncertainties. It's not clear which Added. :::success Done. ::: Section 6.4: Inverting a noisy matrix like this can cause major numerical issues. This is why people use e.g. Bayesian unfolding. It would be nice to see the distributions of the elements of the inverted matrices you get from the toy procedure to make sure there aren't major numerical issues. I've added a few η-pT bins. Please let us know if you require specific bins. :::success Done. ::: l. 924: Each row of the non-inverted matrix will be highly (anti)correlated. Do you take this into account for your toy procedure? Have you estimated these correlations? Added the covariance matrix study in section 9 Section 6: I don't think I saw any mention of the luminosity uncertainty. This should be added for completeness. Added :::success Done. ::: Section 6.8: I think you should discuss how you take into account systematic correlations between data sets when constructing ratios (R_pA, R_FB, particle ratios). In particular the statistical uncertainty, along with the uncertainty on the PID efficiencies, will be non-trivially correlated between the different hadron species. How do you handle this in the hadron ratio measurements? The hadron-hadron correlations are added in section 7.3. If the methodology is correct, we will complete the results for Pbp and pp and propagate them to the final measurement. Figures 54 and 58: I worry that there are unphysical features here. In particular, the proton cross section always seems to be much lower in bins near the RICH threshold. The fact that this behavior occurs at different pT depending on rapidity makes me think that this is a detector effect and not actually physical. So I have a couple of requests: 1. As I mentioned, the entire PID efficiency/unfolding procedure needs a closure test in MC. This could reveal if there are any pathologies leading to these effects. we are currently studying the closure test in MC. We will send the results as soon as possible. :::success Done. ::: 2. You should try aligning your bin edges so that the entire bin is above the RICH threshold. For example, I'd guess most of the lowest pT bin in the 1.6-2.0 eta_CM bin is outside of the p > 10 GeV cut. bin edges Corrected. :::success Done. :::