Hi Micheal et al, On Thu, 06 Sep 2001 10:29:42 -0500 (CDT) Michael Murray <murray@cyclotronmail.tamu.edu> wrote concerning "Thoughts on Centrality": > Dear Brahmins, > I just read Christian's analysis note 37. The Single > Detector Energy method reminds me of what was done in NA44^1 where > we used a small plastic scintilator behind the target to estimate > centrality. We defined 10% central as the top portion of > intergral. In principle this method required knowing what the total > cross section was but in pratice things such as the pbar dN/dy did > not change much with 10% changes in the total cross section^2. > > I beleive that Chrisitan is correct when he says that both a > constant level of secondaries and a componant proportional to the > true multiplicity do not effect the final centrality. However the > variance in these quantities is a problem for both the energy and > multiplicity methods since it tends to wash out any interesting > physics variations with centrality. This is, as you say, a very general problem that neither the SDE or AM approach can solve. Ofcourse, if we had the money, it could be solved with a higher level of segmentation, so that the expected occupiancy would be less than 1, and correspondingly a lower variance in the occupancy. > Ideally we should evaluate our centrality resolutions and > deconvolute for them, or at least quote them in our papers. Note it > does not make sense to quote the top 2% of centrality if our error > from the fluctuation in secondaries and detector reponse is > 4%. Absolutely right. I think that having two, three or even four (I'm thinking of a ZDC thing here) more or less independent methods for determing the centrality would help a lot. > While the energy method can be used for most analysis I think it is > essential that Hiro and Steve pursue their determination of > multiplicity. This is useful both as a physics measurement in in > itself and as a check with other experiments. Yes, obviously. If I gave the impression that I believe we could skip the final step fromk dposited energy to multiplicity, than I'm sorry: nothing could be farther from the truth. I'm infact very intrested in the multiplicity meassurements from the TMA and SMA. But, in for the centrality determination, that final step needn't be done - that was my point. > A good example of this was with the measurment of pion HBT source > sizes for SS, SAg, SPb and PbPb^3. We linked these together with a > measurement of charged particle multiplicity in our silicon > detector. Of course this multiplicity was much harder to extract > than the centrality from our scintilator. Obviously. The same problem as with the BRAHMS TMA and SMA. > I actually prefer this way of linking different systems versus > "measurements" of the number of particpants, which seem to me to be > more model dependent. Again, the point of having many different detector systems do redundant meassurements, is a good idea. > One thing I don't like about Christian's analysis is the > comparison to models. What I want to know from these models is the > answer to questions such as: > "If Fritioff was a complete description of these collisions how well > would BRAHMS measure centrality, number of particpants, number of > collisions etc." > Therefore we should not use the centrality cuts from the real data > but rather scale them to match Fritioff. Althernatively one could > scale the energy or multiplicity from Fritoff to match the data and > and use the same centrality cuts for both data and Monte Carlo. My main motivation for doing applying the regular analysis chain to the EG data, was so that I could compare to a common number, that is the impact parameter. Hence, that gave the baseline for the comparison of the 3 different methods. You are ofcourse right when you say that the EG data (and/or real data) should be scaled, and I believe I pointed out exactly that weakness of the presented plots. However, I was not using the EGs to interpret the result and come up with any physics conclussions; rather the EGs provided the baseline for the comparison; hence also the use of unrealistic (in that energy regime) EGs, like Fritiof and UrQMD. I realise now, that I didn't stipulate this point very much in the note. I will try to revisit the note, and put in these considerations. > Finally I would like to comment on how we think about centrality. > For AuAu collisions at sqrt(S_nn)=200 the initial state is not two > bags of ping pong balls but rather the overlap of two coherent gluon > fields^4. As these fields become decoherent entropy is produced^5 > which eventually shows up as multiplicity in our detectors. > > Thus it seems to me that multiplicity, or energy depostited in a > given detector, is what we want to measure and not the number of > particpants. While the ZDCs may help us measure the number of > spectators (and so the number of particpants) I think it is better > to use them to compare with other experiments rather than trying to > plot variables such as kaon dN/dy versus the number of participants. I agree fully with that. We do not, and can not, meassure # of participants (or equivilant # of spectators) very well (if at all), but what we can meassure is the energy deposited by the produced particles or fragments the collision nuclei or decay products of the two, and this energy allows us to give an estimate of the number of particles produced in the collision. Hence, that is what we're able to compare things to, not the participant number, impact parameter, and similar quantities that we have no access to. The SDE centrality approach is in many ways an attempt to simplify the meassurements an make it more obvious to the reader what the experiment actually meassured. In a sense, the SDE centrality is a definition based on what we observe. > Michael > > > 0) Brahmin is a person who knows 'Brahma' ie the whole universe. Cool! Let's hope we can become Brahmin. What language/culture/...(?) is this from? > 1) NA44 was a fixed target heavy ion experiment in the last > millenium. Are you really that old? Whoops wait, I remember NA44 - guess that makes me really old too. > 2) If one plots dN/dy v centrality then the error on the total cross > section become an error on the scale of the centrality axis. I'm not really sure I understand this. Could you elaborate, please. > 3) EPJ C18 317 (2000) > 4) Eg hep-ph/0104168 Raju Venugopalan > Small x physics and the initial conditions in heavy ion > collisions > 5) Of course more entropy is produced later on in the collisions Thank you very much for your insightfull comments. BTW, I think I forgot to thank Faude for pointing out the error in the BAN number. I did a few things today to convince myself that the |eta| dependent secondary contribution didn't really affect the SDE centrality, and I think I managed to do so. Unfortunately I have not formulated it properly yet, and since I have to leave now, it must wait until tomorrow for me to share those thoughts with you. Stay tuned. Yours, Christian Holm Christensen ------------------------------------------- Address: Sankt Hansgade 23, 1. th. Phone: (+45) 35 35 96 91 DK-2200 Copenhagen N Cell: (+45) 28 82 16 23 Denmark Office: (+45) 353 25 305 Email: cholm@nbi.dk Web: www.nbi.dk/~cholm
This archive was generated by hypermail 2b30 : Thu Sep 06 2001 - 12:18:06 EDT