Title: I assume this is a working title. I would expect this note to become public with a title along the lines of "Bottom and top quark reconstruction with the ILD detector" Abstract: write the abstract Introduction: Line 6: Heavy quarks may be messengers of new physics of primary importance -> Justify this statement by elaborating further (how do heavy quarks pick up the signals? in which models are the deviations sizeable?). To avoid a lengthy discussion here, it would be good to add a number of references. My favourites for top are: http://inspirehep.net/record/1285486?ln=en http://inspirehep.net/record/1683931?ln=en A good one for bottom is this paper: http://inspirehep.net/record/718059?ln=en ---> DONE. The detection of the onset of new physics require however a superb detector performance in terms of flavor tagging including the event by event determination of the charge of the final state jets. -> Even it is very well motivated, we cannot assume the new physics will be there. I would say that "precise measurements of the EW couplings of third-generation quarks require superb detector performance". ---> Ok Line 8: Work out the need for charge determination more slowly? Say at least that it is needed to distinguish b from bbar and thus reconstruct observables like the charge asymmetry (by the way, it would be good to quantify how much better the constraints are when we use the charge reconstruction; Gauthier Durieux once compared charge-aware and charge-agnostic optimal observables for top and he claims the loss is not as pronounced as I would have expected; not sure how this will work out for bottom production). ---> Well, ignoring optimal observables the first thing to do is to reconstruct correctly the quark and the anti-quark. In case of bbar-events and fully hadronic top decays the quark charge is the only handle to do that. We have added a sentence. Line 18: I would keep the three first bullet points. I would move this sentence to the main text and rewrite it to include also bbar production. ---> Ok Line 22: The analyses are ported to the large, IDR-L, and small, IDR-S, models of the ILD detector~\cite{} for the International Linear Colliders~\cite{}. -> don't assume your readers know that you work on ILD and the ILC. ---> In principle you're right. At this point this note is at the level of an ILD Note. One may thus be a bit more sloppy here. Line 24: Introduce the ILC operating scenarios more gently, with a reference to the operating scenarios paper. ---> See previous comment. Line 25: The results here benefit from a refined analysis strategy for the ILD paper that is under review in ILD. -> not sure which ILD paper you refer to and readers from outside the community will have even less of an idea. ---> Agreed but at the moment this is rather a supporting note for the IDR. Therefore we leave it for the time being. Knowing that soon we will be able to add at least a reference to the EPS proceedings. Currently we use the link to Adrian's talk. Line 28: A few words on the simulation would be helpful (at least to state that this is full simulation on a detailed detector model). ---> Ok, added Line 30: I'm delighted that you would choose the algorithm we developed, but I am not too happy about the "ValenciaVertex". It's already rather confusing with just the Valencia and VLC algorithms. What you describe in the following lines is the VLC algorithm, that is documented in the following paper http://inspirehep.net/record/1476640?ln=en. Is "ValenciaVertex" simply the VLC algorithm, reimplemented in the LCFI environment, or does it do something else? In particular, does it use the secondary vertices to guide jet finding? We have discussed the issue of splitting secondary vertices across jets several times. I believe the most correct solution is to replace the PFOs connected to the vertex with a "B-hadron" PFO. That way it always gets clustered into a single jet, and we maintain a "clean" jet algorithm, for which one can calculate a cross section. ---> These bullets were a relic of a very early version of the note. We have described now better what has been done being aware that even more streamlining seems necessary here. That's what reviews are good for. Line 35: We use the LeptonFinder to identify isolated electrons and muons in semi-leptonic ttbar events. Line 36: remove [maintained by Ryo]. ---> We consider that this note is also (and maybe even mainly) for internal communication. Therefore we prefer to keep it because otherwise this detail may get forgotten. In general you are right sentences like these will have to vanish. Line 39-44: It's not clear how this "restoring" comes about. Do you simply create a "cheated" track? Which d0 and z0 do you assign? ---> Added the word "reconstructed" to the introductory phrase. We think that it reads better now. Line 56: We need a sentence here that says what you are going to do. Something like: "The following method combines the results of the two (nearly independent) charge measurements on the b and bbar jet into a robust charge determination" ---> Couldn't say it better. Thanks! Line 57: I would merge the second and third bullet. Specify which events are accepted and rejected (i.e. consistent charge assignment and contradictory results?) ---> This item has been rewritten and combined into a single item w/o sub-bullets. Line 64: The selection of the bottom and anti-bottom quarks in ttbar events can be straightforward, alright. Just take the two jets with the highest b-likelihood score. It isn't always correct, but I think the performance will be close to that of more sophisticated algorithms. ---> We assume that you refer to the application of the pq-method to the ttbar sample. Be aware that the pq method assumes implicitly that the measurement b and bbar are uncorrelated. This is a priori less true in case of ttbar events where the b-quark pair cannot be expected to be produced always back-to-back and be in different hemispheres. It seems that this assumption is more or less valid is the case of a left-handed electron beam but not in the case of a right handed electron beam. Section 2.1: you have a subsection on MC samples with number 2.1. Create also a subsection "Methods" the first bit of the section. If you have material on event processing it would be good to create a separate subsection. If not, you can drop this from the subsection name. ---> We describe the event processing in great detail now in Secs. 4.1 and 4.2. Line 68 contradicts some of what comes later (you do have the opposite polarization for top quark pairs). Merge the two paragraphs into one. ---> Indeed the text was confusing mixing two phases of the analysis. Text has been corrected. Line 77-100: This description is OK for an internal document. For outsiders you need a few lines: "Samples generated with WHIZARD X.X. Top quark pair production is the dominant process in the e+e- -> bbbar lv qqbar sample, but it contains also single top and WWZ. ---> Ok, added a line as suuggested Line 95: Why do we get a number for the bbar fraction for 250 GeV? Can you get a number for 500 GeV? W ---> Typo. 250 GeV -> 500 Gev. Line 93-97. Indicate whether ttbar is part of the two-quark sample... I guess you simply produce e+e- -> qqbar, where q=u,d,s,c,b? ---> Ok, done. Figure 1: Is this the momentum or the transverse momentum? Can you add a line to explain what causes the broad peak and long low-momentum tail for the b-jets in bbar production? Is it just the luminosity spectrum or do we expect results to be affected by the finite jet area, by the limited acceptance, etc.? In particular, In this respect, it would be good know whether the jets are reconstructed on stable particles or on reconstructed PFOs? Is the result very different for these two? ---> Well the first purpose of the figure is to demonstrate the different momentum ranges of the b quark that are covered by the analyses presented here. At the moment the bbar study is rather a supporting study so we won't put more effort into this for 500 GeV. We will pick this up for the paper of the 250 GeV study Figure 2 and 3: It would be good to have uniform Y-axes for all figures, such that one can compare them more easily. Can you use the same fill style for the ttbar figures as for the bbar figures? We need a more descriptive caption (Something like: The fraction of charged particles missed by the track reconstruction as a function of the charged particle polar angle. The inefficiency is broken down into several sources using the VertexRecovery tool.) ---> TO BE DONE. Figure 2 and 3: There must be a subset of tracks that is reconstructed, has the right impact parameter, but is not included in the vertex fit. I guess some tracks are mismeasured and would have a too large contribution to the chi-squared of the vertex. Are these included in other reasons? Or is that a separate category of losses? ---> They are included in "Other reasons". In fact LCFIPlus is very stringent on which tracks to include or not. It may be worthwhile to resume the study that Sviatoslav started in 2016 when he visited KEK, Tokyo and Kyushu. Line 105: This improvement in vertex reconstruction is achieved by using Monte Carlo truth information, right? How much can you gain without cheating? ---> There is no cheating. The algorithm was developed using the MC Truth information but does use in the end only information that is available in the data. Figure 4: For ttbar events, does |cos theta| refer to the polar angle of the top quark or of the b-jet? While for bbar events these are very close, the direction of the top quark and the b-jet are only somewhat correlated. You should clarify which direction is plotted. It may be useful to compare the two plots in ttbar events (i.e. the purity versus cos theta_t and cos theta_b). I would expect the purity versus b-direction to be similar to figure 5 (modulo effects related to the momentum), while in the purity versus top-direction the dip should be much more smeared out. ---> |cos theta| refers here to the polar angle of the B-hadron. Further studies as suggested are for sure interesting but we would like to move this to a later stage. Line 122. Can you specify the acceptance better? I assume the selection includes a cut on the b-tagging likelihood, such that the poor track-finding efficiency in the forward region propagates into the acceptance? If you make Fig 6 using the B-hadron direction instead of the b-jet direction, does the dip become steeper, with all the loss concentrated in the last bin? (or even the second half of the last bin? Figure 15 seems to show that this is indeed the case.). ---> For the sake of getting this note we would kindly ask to address this question (which is relevant) at a later stage. It is for sure true that a jet has the tendency to dilute acceptance issues (as the top does in comparison to the b). Line 128: How often is a kaon found in charm or light-quark jets? Could we use the presence of the kaon to b-tag the jet? ---> Kaons are of course also produced in c and s-quark processes. Would have to check in the PDG for the exact numbers. Line 130: Explain the relation between the kaon charge and the b-quark charge briefly. ---> We cite from Sviatoslav's thesis "In the generator, about 87% of b-hadrons (NDLR: which agrees with the PDG value) are set to have correlated K± charge in the generator, which makes the K± charge a reliable indication of the initial b-quark charge. The kaon charge was used to determine the b-quark charge at the SLC and the LEP experiments." In principle there was statement on that just two lines above but the the 80\% have become 87\% now. Figure 7: these all correspond to the bbar process. The four figures correspond to three momentum intervals [2,2.5], [5,6] and [10,15]. The last figure is shown twice. ---> Last figure, indeed a mistake, now the correct figure for the interval [30,40] has been included (same for Fig. 8). Line 133. It would be good to give an indication of the separation power here, and of how it evolves with momentum. Somethign like: the pion-kaon separation is XXX sigma for 1 GeV and degrades to YYY sigma for 10 GeV particles. ---> Well in principle we just select a strip in the dE/dx vs. momentum region with a high population of dE/dx values expected from Kaons. For the moment we have added a dE/dx plot and put the details to an appendix. For the sake of making progress we refer the more quantitative argumentation that you require to a later stage (We take however well note of your question). The Fig. 9 (now Fig. 7 right) shows basically the change in purity and efficiency when varying the width of this strip. Figure 8: Is there any reason to believe that the two processes would yield a different result, other than through a slightly different population of higher- versus lower-momentum kaons in each the momentum bins? What do you conclude from this figure? ---> Well in principle the figure include all final state particles. In case of ttbar there also the W that decays to cs or ud and can thus yield kaons. Since it is now in the appendix we would like to keep it for further reference. Figure 9: Explain the difference between ttbar and bbar. This is due, I guess, to the stiffer spectrum for kaons in bbar events? Once you fold in the degradation in separation power at high momentum, you get the effect of Figure 9? ---> We don't have a precise answer on that at the moment but apart from the different momenta please note also that the events are much richer in can of tt as the W adds also hadrons to the final state. Table 1: We need a bit of explanation to understand what B_rad.Z means. I assume these are radiative return events. They have a very large impact? What does your selection do to get rid of them? Is it just the cuts on the invariant mass of the di-jet system and the photon veto? It would help if we had that section on "Evernt processing" to outline the selection. ----> We have added a few sentences to Sec. 4. Figure 10: we need a more descriptive caption. Is the y-axis the same as Figure 13? In that case you should make the figures more uniform and the captions the same. ---> Caption updated for Fig. 10. Fig. 13 NEEDS FURTHER REVISION. Sections 4 and 5: describe what is in the Figures 10 and 13 and give an explanation. It seems the size of the detector does matter for the final purity of the charge determination (even if effects were quite small in the lower-level plots). It would be good to understand what drives this. ---> The difference is mostly visible in the combinations that include Kaons. This speaks for the fact that a smaller TPC puts a penalty on the dE/dx measurement. We note this now more clearly in the text w/o speculating too much. Figure 11 and 12. These figures add very little information. Consider moving them to an appendix. ---> Since we are at the level of an ILD note that may/should contain more detailed information we prefer to leave it here for this time. Tables 3 and 4: is there no double-charge-determination cut in ttbar events? ---> This is just the preselection. Cuts that are needed to distinguish the charges are explained in the text to Fig. 15. Might be useful to present them also in form of a table. Line 153-156: I would be interested to hear how you plan to combine the charge measurements of the kaon, vertex charge and of the isolated lepton. It should be a very powerful system to remove any mismeasurements in-situ. ---> At the moment we do the following and we are not consistent between ttbar and bbbar i) bbbar - In case of bbbar we proceed to the principle first come first serve, i.e. if e.g. the vtx+vtx measurement yield a consistent results we stop looking for other methods and so forth. ii) ttbar a) We build all possible combinations of charge measurements and check whether they lead to consistency results a1) For the isolated lepton + chi2 we take of course only the charge of the lepton. b) We then issue a majority vote, i.e. we add up the charges associated to the hadronic top and the sum of all these charges determine the charge of the top. If the sum is zero then we discard the event. There might be other more intelligent (and maybe less error prone) ways to exploit the information (for example give more weight to the more robust vertex charge measurement or require that the sam exceed a certain value or simply proceed as in the bbbar case). Line 156: The chi-squared cut was intended to remove strong migrations due to incorrect W-b pairings. I had hoped one could remove that cut now that a second handle is available to determine which is the correct pairing (i.e. one can use the sign of the charged lepton, together with the b-bbar separation to understand which is the right combination). ---> There are two philosophies that one can adopt. i) Increase the statistics by re-analysing events that escaped even the chi2 cut. ii) Discard the chi2 cut and use only the other handles (lepton charge, vertex and kaon charge). We wanted to stick as long as possible as close as possible to the old analysis since it serves/served as a reference. We have however also the impression that it's time to say bye-bye to the Chi2 cut. However for the sake of getting this note done we didn't play around too much with the cuts since after all the final result looks quite satisfactory. Let Line 159: This refers to Eq. 3, not 2. In principle I would expect the method to correct the spectrum perfectly (as this is Monte Carlo). Do you understand where the non-closure in the very forward and backward regions comes from? I suspect the method has a prize to pay in terms of statistics. Could you compare the statistical uncertainties on AFB one would obtain with a single method (i.e. the charged lepton in ttbar events, assuming one will rely on MC to fix any migrations) and with the more sophisticated double-method approach? ---> Ok for Eq.2 -> Eq.3, thanks. Note that to correct for residual migration effects we don't make use of MC information by means of Eq. 3. The bb-study at 500 GeV suffers from comparatively small statistics. Therefore one has to apply global p and q values that are fine for the central part of the detector but not anymore at the edges of the acceptance (see also Adrian's talk in the ILD Meeting of the 24/4/19). CHECK WITH ADRIAN (also for 250 GeV) Line 160: To see differences between L and S detectors, it might be more instructive to look at the polar angle distribution before the correction. ---> Ok material can be provided. TO BE DONE. Line 164: I prefer to see the full selection description in one central place, rather than learning about additional cuts in the results section. ---> We have gathered now the details of the selection in Sec. 4. Figure 15: nearly perfect. Can you compare with the pre-correction result, at least by adding a sentence like: the reconstruced AFB has a bias of XXX % before the correction; the correction reduces this to YYY %. It may be worth stating that the distribution is corrected for migrations due to charge mismeasurements, but not for the (polar-angle) dependent acceptance. That explains the dip in the first and last bins in Fig 15b, that remains even after the correction. ---> Not sure which corrections you have in mind here. For the tt-analysis we don't apply (yet) e.g. the pq-method. It is however true that the limited detector acceptance smears out in the top polar angle spectrum but remains visible in the spectrum of the underlying b spectrum. Note also that in case of ttbar the polar angle of the top (and thus also of the b in case of left-handed electron polarisation) is much less steep than in case of bbbar (left). Altthough historically we have 'discovered' the migration effect in ttbar it can be easier dealt with in this case. Figure 15 and 16: the y-axis states "entries/0.1 rad", but it says cos theta on the x-axis. ---> Indeed, will be corrected (TO BE DONE) Line 169: this sentence should refer to figure 16. ---> Indeed -> Corrected Line 174: I suspect the acceptance depends on polar angle. Your procedure does not correct for acceptance. Can you check whether this effect could be responsible for the non-closure in Fig. 16b? ---> Meanwhile replaced by a new (and much nicer!) plot. The b-quark spectrum in the right-handed case deserves further studies. Table 5: The ttbar analysis has a trade-off between statistical uncertainty and possible systematics. Cutting hard on the chi-squared is good for the latter, but bad for the former. We had no alternative when we chose this cut value of 15, but now that there is a second handle on migrations, you should be able to relax this cut. ---> So far we stick relatively close to the reference analysis. Take into account that the analysis is/has been carried out by master students and between the DBD and the IDR so many things have changed that it was good to have a solid rock in the storm. We agree that the cuts have to be revised. Line 190: electro-magnetic ---> Isn't electromagnetic also ok? Line 191: Figure 17 ---> Yes Line 191: as already indicated in my previous comment, you should specify what you mean by "precision". I understand that it is half of the 68% C.L. interval. You should also specify how the fit was performed. Did you extract all 5 form factors from the 4 measurements simultaneously? Did FCCee do the same? For the HL-LHC prospects, did you use individual or marginalized bounds? ---> The text has been changed since then taking into account your for remark. Figure 17: It would be interesting to see the current LHC+LEP/SLC bounds. The actual results are much more solid than the S2 extrapolation, which rest on an ad hoc extrapolation. ---> Can they give a better picture than the extrapolation. If we do this one may trigger a question how we compare to 3ab-1. But yes be can discuss (maybe also with ILD) to what to compare. Figure 17: The caption needs a comment on the missing HL-LHC bars. The EFT does not include operators that map onto the F1V and F2V form factors. ---> Text corrected accordingly. Figure 17: the caption is missing the word "top" Figure 17: It should be possible to make a similar figure for the bottom quark. The fit in arXiv:1907.10619 includes the relevant operators. It would be interesting to see the effect of 250 GeV and 500 GeV runs of the ILC. No FCCee results are available, as far as I know. ---> We still consider (also due to the lack of statistics) to be in a state for a serious attempt to produce such a plot. We agree however this is interesting. At the moment in another study we discuss the evolution from the pole to 250 GeV. Figure 17: Where does the 3200/fb come from? I thought we integrated 4000/fb in the official operating scenario? 800 fb-1 will be spent on equal polarities. We haven't studied these samples by now. Line 199: Ref.[1] was for 500/fb. You envisage a much larger data set. When you say the statistical uncertainties are compatible, I think you mean after scaling to the same integrated luminosity? ---> Yes. You are right. Summary: I suggest to rewrite the summary, indicating in first paragraph what has been done, before drawing conclusions on the value of these techniques and the differences (if any) between the two detector models. ---> TO BE DONE.