From: SMTP%"JELys@LBL.GOV" 31-OCT-1997 17:21:07.62 To: TOBACK CC: Subj: LBL comments on draft PRL cdf4149 Date: Fri, 31 Oct 97 15:20:47 PST From: lys@Csa.LBL.Gov Message-Id: <971031152047.3d204d7d@csa6.lbl.gov> Subject: LBL comments on draft PRL cdf4149 To: toback@uccdf.uchicago.edu, frisch@uccdf.uchicago.edu, reb@hep.anl.gov, stk@sgi6.hep.anl.gov, yagil@fnald.fnal.gov, mdshapiro@lbl.gov Reply-To: JELys@LBL.GOV X-ST-Vmsmail-To: @GGPRL.DIS X-ST-Vmsmail-Cc: LYS Berkeley comments on the first draft of the paper "Searches for new physics in diphoton events in ppbar collisions at sqrt(s) = 1.8 TeV". We congratulate the authors (and godparents) for producing a quite reasonable draft in the unusual circumstances of refinding an event that is already out there. We appreciate the hard work that goes into calculating various "expected" rates. We agree with the idea of publishing this work. ** Thank you!! We have some comments, given below. 1. The title says "New Physics", the abstract "anomalous production..." and "consistent with s m expectations". Is it reasonable to expect a reader to infer that new physics = anomalous production = inconsistency with sm? (See also point 19 below). ** We think so...These words are frequently used in papers which search for new particles. 2. p 4 para 1 line 3: "generate gg signatures involving" seems very awkward. Maybe something like "can yield events with two energetic photons plus one or more of the following: extra missing ET, extra jets, leptons, gauge bosons (W, Z, g), b-quarks." 3. same para: "In the run 1 data" is CDF jargon, not useful information to most readers. Does this search use 1a+1b or just 1b? Whichever, maybe something like: "One 'eeggMET' event recorded with the CDF detector has been reported and has been the subject of some theoretical speculation. We have now performed a systematic search ........ . This search... collected with the CDF detector between April 1993 and June 1995 [3]". (Use appropriate dates). 4. Same para: maybe "..a systematic search for anomalous gg events ...", i.e., omit "other". 5. Same para last word: "it" -> "that event". ** We have tried very hard to try and make this paragraph better as many people had comments on it. After many iterations of incorporating comments 2-5 (as well as other group's comments) we found that indeed each sentence read slightly better. Unfortunately, the paragraph as a whole was worse for it. Eventually, we wound up with a version which was much closer to the original, but fixes most of the problems you (correctly) point out. For example your comment about Run 1 being jargon has been fixed. We believe that the new version is clearer and hope that it meets with your approval. In either case, we feel that the paragraph is clear and does a good job of setting up the paper. We have used authors perogative for the stylistic issues. Just so you know, here is the version with everyone's comments: In many models involving physics beyond the standard model (SM), cascade decays of heavy new particles generate $\gamma\gamma$ signatures involving missing $\Et$ ($\mett$), jets, leptons, gauge bosons ($W$, $\Z$, $\gamma$), and possibly $b$-quarks~\cite{interest}. For example, in supersymmetric models with a light gravitino, pair-production of selectrons which decay via $\selectron\goes e\NONE\goes e\gamma\Gravitino$ can produce the $\gamma\gamma$ final state along with $\mett$ and electrons. In the data taken during 1992-1995, an `$\eeggmett$' candidate event~\cite{Park} was recorded with the CDF Detector~\cite{detector}. We have performed a systematic search for other anomalous $\gamma\gamma$ events by examining events with two isolated, central \mbox{($|\eta|<1.0$)} photons with \mbox{$\Et>12$~GeV} which contain $\mett$, jets, leptons ($e, \mu, \tau$), $b$-quarks, or additional photons~\cite{Future PRD}. This search is based on 85~pb$^{-1}$ of data from $\pbarp$ collisions at $\roots = 1.8$~TeV collected with the CDF detector. In this Letter we describe the results of the search, including the $\eeggmett$ candidate event, and proceed to set limits on two SUSY models that have arisen to explain it. Here is what we decided is best: In many models involving physics beyond the standard model (SM), cascade decays of heavy new particles generate events with two photons plus missing $\Et$ ($\mett$), jets, leptons, gauge bosons ($W$, $\Z$, $\gamma$), and/or possibly $b$-quarks~\cite{interest}. In the Run 1 data taking period of the Fermilab Tevatron an `$\eeggmett$' candidate event~\cite{Park} was recorded with the CDF Detector~\cite{detector}. We have performed a systematic search for other anomalous $\gamma\gamma$ events by examining events with two isolated, central \mbox{($|\eta|<1.0$)} photons with \mbox{$\Et>12$~GeV} which contain $\mett$, jets, leptons ($e, \mu, \tau$), $b$-quarks, or additional photons~\cite{Future PRD}. This search is based on 85~pb$^{-1}$ of data from $\pbarp$ collisions at $\roots = 1.8$~TeV collected with the CDF detector. In this Letter we describe the results of our search, including the $\eeggmett$ candidate event, and proceed to set limits on two SUSY models which have arisen to explain it. 6. p 5, para 2: point b) is ambiguous. Presumably "(i) no track with PT >1 GeV, and (ii) at most 1 track with PT< 1GeV". Also, does rule (ii) apply separately to each cluster or is it the sum of PT<1 pointing tracks that is at most 1? ** We think the sentence is clear. Each photon is treated separately. Only one track is allowed to point at a candidate, and only if it has less than 1 GeV of P_T. 7. same para: twice "to remove pizeros" (compare "to remove electrons or jets"). ** We thought it was so important to point out the backgrounds so we left it. 8. (deleted). ** That one was easy. 9. p 5 last para 1st sentence: maybe "and" -> "or". ** We think "and" is better. You can have combinations. 10. p 5 same sentence: There is an awkwardness in the paper in that we do not really say what we are doing. In this sentence (i) the abstract's "anomalous production of MET etc" has become "anomalous MET etc". (ii) does the word anomalous qualify all that follows or only MET? (iii) We do not really "search for the presence of jets", we set up rules for calling entities "jets", and in each two-photon event we list the jets and their ET values. Then we look at the quantity 'number of jets per event with ET > ET_cut'. Then we make some judicious cut point and count 'number of two-photon events with njet>njet_cut, where njet = number of jets with ET>ET_cut.' Similarly for leptons and photons and b's. MET is a little different. So, perhaps "We search the diphoton [or, two-photon] event sample for events with large MET, and for events with at least one electron, muon, tau, b-jet, or additional photon, or at least four jets." ** This point is well taken. The text of the paper has been fixed to reflect this point. 11. same para, line 8-9: "remove events", does that mean remove such events from the gg sample, or just from the "anomalous MET" sample? The question is answered (confusingly and too late) at the end of the para. Maybe should say "we remove events from the anomalous MET sample if they have a jet...". ** As you point out, the text is clear. We think the order is appropriate. 12. p 6 para 1 (= same para): It seems strange that after requiring two photons with ET > 12 GeV, we now look for a third with ET > 25 GeV - more stringent than the initial search. Some of us have to think for a while here. If an event has photons with 30, 16, 14 GeV, is it allowed that the "third photon" is the 30 GeV one (where all pass low-threshold cuts)? And presumably, if we have 30, 23, 14 GeV and the 23 GeV one passes only high threshold cuts we do not have a third photon passing the cuts? ** You are correct that this is a subtle point, however we weren't able to spend a lot of text on the issue (it's done in more detail in the PRD). The point is that you have to have two photons which pass either the high threshold or the low threshold to get into the sample. A third is required to be in the sample with Et>25. This method is done mostly for consistency, but also makes the fake rate estimate simple (as done in CDF Note 4183, P37-39). We believe the paper is correct. However, any (reasonable) misinterpretation does not change the results; i.e, there are no events in the data any way you slice it. There is no space in this paper to deal with this subtlety in any more detail. (To answer your questions from the examples, if an event has photons with 30, 16 and 14 GeV, and the 16 and 14 GeV photons pass the low threshold criteria, then the event is in the sample. If the 30 GeV photon passes the low threshold cuts, but not the high-threshold cuts (as stated) then the event does NOT pass the selection criteria. If the 30 GeV photon passes the high threshold cuts, then this would pass the selection criteria. In the second example with 30, 23 and 14 GeV and the 23 GeV only passes the high-threshold cuts, then the 30 and the 23 get the event into the sample and the 14 GeV does not pass the 25 GeV requirement. Thus the event would NOT pass the selection criteria. If the 23 GeV photon passes the low threshold cuts, and the 30 GeV pass the high threshold cuts, then it would pass the selection criteria.) 13. same para: there is an overall problem in the way this para is written. Two different notions are very mixed up, (i) how to define a jet, tau, etc, (ii) what "anomalous" means. See point 10 above. The ET cuts on jets and additional photons, but not for leptons or b-jets, are given in the text. It would be better to give them all, and better to give none and refer to Table 1 than randomly give some. ** We think that between the text, the table and the references, the cuts as well as the results are clear. We don't have space to add more and we wanted to highlight what we feel are the important results. This is addressed in more detail in the PRD. Again we don't find the word anomalous to be problematic. 14. p 6 para 2: We come to observed numbers of events, meaning events that pass certain cuts involving signatures or additional objects. Some of the cuts have been mentioned in the text, some have not. No explanation has been given to a reader on the "general principles" of the chosen cuts. Is the only guideline to make the "expected" number < 2 or so? - or did SUSY or other models have some input? Etc. We think that at least a sentence on this point is necessary. ** Another reasonable request. However, there simply isn't enough room in the PRL for any real description that doesn't raise more questions than it answers. The basic answer is that the cuts aren't optimized, they were designed to be simple, reasonable and standard. A more complete description would take a lot of space and we're still over budget (again, this will be done in the PRD). For your personal edification these issues are described in great detail in CDF Note 4183. 15. same para: "expected" numbers of events - only fragmentary information is given to a reader on what expected numbers mean and (sometimes) how they were calculated. The MET expected number is "a standard model expectation", are the others also standard model expectations? Actually, it appears that the MET is one step removed from "sm expectation" in that it is "derived from ... the Z0 control sample" (Fig 1 caption), i.e., without regard to whether the Z0 are strictly all sm. And the njet>4 expected number seems to come from an exponential extrapolation (is that right?), with the reader given no reason to believe or disbelieve such an extrapolation (the poor reader is referred to top papers to find njet>=4 background estimation methods; that is NOT a good place to send a reader for that information). For some of the signatures/objects, we (and a reader) can think of possible sources. For example WZ events with a missed e in Z->ee. Presumably such sources have been considered? We think readers should be reassured on this point. An interesting point can arise in estimating "expected" numbers far out on tails of distributions - how to deal with small numbers and Poisson statistics? In the MET case, e.g., presumably 2663 events gave 0 cases of MET>40 so 2239 events might give <2.3 *2239/2663 = 1.9 or similar. We think readers should be given some idea (or a reasonable reference if one exists) of what is actually done in this situation, how to get from 1.9 to x +- y ? ** You've packed a lot into one comment, but again the primary response is that there simply isn't enough room to discuss how each background is calculated in detail in a manner which doesn't raise more questions than it answers in the PRL. The reader can figure out most of them. For example, the met resolution is given, so I think its fairly obvious that we used the resolution and its uncertainty to estimate the number of events expected. There are of course small backgrounds to any ee sample, but these are tiny and will not affect a resolution measurement, especially within the errors given. There is no room to reassure the reader we haven't done anything stupid here (as opposed to any other place). Also, the njet is from an extrapolation (I think the paper is clear about this). I'm afraid I don't understand your statistics question. We have a resolution distribution and a high statistics sample used to measure the resolution. Your calculation of 1.9 events is not what we do. For more details on the background calculations see CDF Note 4183. 16. p 6 para 3, last sentence: "we find agreement", but for central leptons we see 4 and expect 0.5, at best very marginal agreement. To say that one is a tau candidate with an expectation of 0.2 does not help at all. And presumably radiative and double-radiative Z decays are included in the expected number. There is a real quandary here, which has to be dealt with. Maybe the paper could say (1) observed and expected numbers are given in table 1; (2) there is reasonable agreement between observed and predicted, except in the central lepton case, where obs=4, exp=0.5, prob=0.002 (whatever); (3) detailed examination of the 4 shows one consistent with double-radiative Z and one with radiative Z decay..; (4) such Z decays are included in the expected number, contributing a sum of 0.1 (whatever); (5) to be conservative we must assume that the number of such Z decays has fluctuated upwards; (6) remove these Z decays, then obs=2, exp=0.4, prob=0.02 (whatever); (7) that prob. is still low, but not sufficiently low to constitute a positive signal; (8) however one of the 2 remaining events is a somewhat exceptional event, as is discussed below. ** You are quite correct that we have observed what appears to be an upward fluctuation of a number of sources, which when summed (probably inappropriately) give the impression that we are claiming an excess. We are not claiming an excess. As stated in the text, we believe this to be a fluctuation since the events do not look particularly unusual, and they have very different sources. The intent was to not make too much of this point. Again, there is no space to discuss these issues in more detail. From your comments you have clearly understood the issues. 17. same sentence: "with one possible exception" - given the Table (and Fig 1) the one possible exception is the central lepton number. It is a separate point to claim that one event is exceptional - we are changing the rules. That is allowed, but the paper should say what is happening. We have now looked in detail at the observed events in the Table, and find one that is exceptional. It occurs in more than one place in the Table (presumably - as presently written the paper does not explicitly say so), and the expectation for that is very small (again presumably, since a reader is never explicitly told why this event is a "possible exception"). So, maybe this possible exception should be in a new para, with explanation. If this exceptional event is the same as the one mentioned in the paper's first para, a reader should be told so. ** We are looking for anomalous events. By any standard this event is `exceptional.' We think that the 3 observed events vs. 0.3 expected is clearly not the exception. We think the text is clear. 18. p 7 line 9-13: We think this sentence would be very difficult for a reader to understand. Does "elsewhere in the detector" mean "not close to the four observed clusters"? - if so, why not look and see if there is energy there? ** We have fixed this sentance. 19. p 8, para 3: does the word "anomalous" mean something different here to what it means earlier? ** No it doesn't. 20. same para, line 9: "no such excess of events" - there are no events, so there is no need to raise the question of how to decide what the excess is. So try "No events with 3 or more jets...". ** Done. 21. p 8 last para, first line: "no obvious signal" ?? Signal for what? For anomalous WWgg? - that was the last item discussed. ** That part of the sentence has been removed. 22. Ref [14]: 2627 -> 2626. ** Yes...2626. (I'm impressed!) 23. If space is a problem: (1) much of the detector description could be removed; e.g., the length of the CTC and how many measurements it may provide are not crucial here. (2) there is no need to restate in words what is given in a Table (as above Table I). ** (1) We have removed a good deal of the detector description. (2) We have left the event counting results as we feel they are part of the central results of the paper. 24. Is it possible to give a qualified order-of-magnitude cross section for "eeggMET" events, defined, for example, as on p 8 para 2 1st sentence? As it is, the best a reader can do is sig_tot(pbar.p)* 1/(3*10**12)*guesses at several efficiencies. Does/will ref [19] give some sort of cross section? (We understand that there are problems here.) ** No. There is no plan to estimate the cross section based on one event. Considering the problem the tt->taus had, we weren't even going to try. Jeremy Lys, for the Berkely CDF group.