Dear Collaborator, I will like to share with you some thoughts I have had on the multiplicity analysis and the paper. The content of this note has been circulated for a few days to the people directly involved with the analysis to catch obvious mistakes and add comments. I should stress this is a personal point of view and my assessment of the analysis status. Thus I take responsibility and blame for the content. It was mainly written during a major hiccup in FAA air traffic control at Indianapolis while waiting in Cincinnati (6 hours layover) en route between Islip and Dallas. What a Ghouly and Ghostly evening. PS. Since this was written you have been told on the progress report notes written by JHLee and Yury. They cover several of the issues discussed here as well as providing some of the documentation needed. The aim of this note is to spur a discussion among the people doing the work for the analysis. I have most likely forgotten certain items, and may have misunderstood others. I though think it is valuable to write down the concerns and open questions. This note is intended to outline the issues in regard to the analysis of multiplicity and dN/dh. It is not a description of the analysis in any detail. This is left to the other notes written and in preparation, as well as to additional comments. The multiplicity has been derived from three independent detector systems, namely the tiles, the beam-beam counters and the Si-strip. They are all categorizes as seeing multiple hits for central collisions, and only in the case of Beam-Beam counters with any albeit not sufficient capability for discriminating against background hits. Making and understanding corrections thus becomes crucial. Also the errors, statistical as well as systematic must as well worked out. Tiles Energy scale The linearity of the energy scale i.e. conversion of light input to the ADC count has been demonstrated to be linear (see Note by Ramiro http://www.rhic.bnl.gov/export1/brahms/WWW/private/detectors/mult/tiles-line arity.html ) The energy scale is difficult to establish. It has been done by Cosmic ray calibration. First in laboratory, and then later in situ, triggering on two tiles on opposite side being hit. This ensures a roughly perpendicular entrance - though it has a weighting over angles. This suffers of the problem of not knowing the particle/energy composition of the trigger particle(s). Presumably it is higher energy $\muon$'s thus having a energy deposition of approximately 1.1 - 1.2 times a true min bias signal. See Ramiro's note on tile calibration with cosmic rays (http://www.rhic.bnl.gov/export1/brahms/WWW/private/detectors/mult/cosmic-ca lib-2000.html) Selecting very peripheral events from the data sample. This seems to show a peak approximately near min bias, and with a 1/cos(theta) dependence. The advantage is that the particle composition presumably is close to that in more central collisions. The problem is that the peak is not very prominent. In both cases there is a can of worms in regard to the shape of the peak. Ideally the dE i.e energy loss of the particle is of Landau shape, but the deposited energy cannot be (The high energy tail in Landau comes from delta-electrons that in part escape the detector. There is also the question how the multi-particle response is, given the 1 hit response. Does it scale by the mean value, the most probable value or something in between? There have been a large number of internal discussions on this. I believe to have understood that for a bounded distribution the scaling closely follows the mean. This scale can easily be the single biggest factor that contributes to the dn/deta uncertainty (maybe 10-20%). Conversion of energy scale. Having determined the energy is of course not the end of the story. The energy as observed in a given tile for a given vertex must be converted into an appropriate weight of a dN/deta distribution. At least two approaches are available (and has been investigated - JH,RD,HH,SJS) The first method consists in determining the #equivalent MIPs (as function of eta) using the energy scale. This subsequently gives the 'uncorrected' dN/deta. This includes the effect of secondary energy (see more discussion later) that contributes a significant amount to dN/deta. Such processes come from conversion in beam pipe, the tiles, hadronic interaction. This clearly depends on the particle composition of the primary distributions. This is discussed in JHLees note . The number of primary hits (dN/deta) is found from a MC calculation that determines the correction as fct of eta. The second methods does not go via the calculation of #Mips, but evaluate directly from (eta,..) the correction between total energy deposited and the initial dN/deta. It is of course still dependent on the 'model' chosen for the initial distributions. The saving grace is presumably that this is overwhelmingly pions produced with a reasonable well known mean <pt>. This is the basis for the analysis by Hiro Ito. This later method presumably is somewhat more coherent than the first, though the results should be close. This second method automatically includes the change in <p> with eta, and energy deposit in tiles. The change of energy deposited by the primary particles were discussed in note ( #11) by JJG. The corrections are quite large in both methods in the order of a factor 600/900. What charged particles should be included? This refers to what we mean by a dN/dEta distribution. It springs from a discussion with Hiro Ito and the interpretation of the Hijing and Fritiof event data files. In usual terms one what say this should include all primary charged particles. This obviously are the primary pi+-,K+-,p and ,p-bar . The issue really comes about with the weakly decaying particles as K0->pi,pi and L->p,pi-. A good fraction of these particles will show up in tiles and si-detectors as charged hits. For the Ko(short) most probably will be registered, while for the L only a smaller fraction will be. Ideally one would want to exclude them, but since the detector does not have any capabilities to differentiate one will have to rely on MC. Thus the deduced dN/dEta depends on the underlying model Are one then not better off to include decays e.g from the Ko which are observed by close to 100% rather than subtracting from a model? The Ls are more tricky; because of the longer ct some will definitely pass through the tiles as neutrals and not be registered. Estimates of systematic errors. This MUST be looked at carefully. - Vertex determination (uncertainty from BB). What is the contribution to dN/deta if vertex is smeared by resolution? - Finite size of tiles. - Energy scale determination - Normalization of total minimum bias cross section. Some documentation to be provided. - MC correction between Energy in Tiles and dN/dEta. - Energy calibrations of tiles - Evaluation of weakly decaying baryons, mesons. - Finite size of tiles - Systematic difference between 2 MC methods for corrections. - Systematic effects due to choice of chosen event generator (e.g Hijing, nexus/Fritiof) - Effect due to chosen vertex cuts. E.g. one could evaluate |v|<20 and v<|40| as example. High eta data from the beam-beam counters. It was thought the beam-beam counters be less susceptible to background due to the inherent particle selection requiring beta>.67 and at close to normal incidence. It is though clear from both data and simulations that this is not the case. The background in large part comes from the conversion of gammas in the beampipe, secondaries from hadronic interactions also in the beam pipe, and finally from hadronic interactions in the 3-4 cm Lucite itself. The amount of primary to background varies depends on geometric position of tube, size of tube and centrality and probably also with vertex position A new set of calculations being more systematic has just started. Some problems observed was e.g. that the dN/deta deduced from the large tubes in the left ring i.e. with same geometry could give value differing by 20% in raw values before correction. This should be much better. Could indicate that some tubes are noisy? A careful investigation of raw spectra is needed i.e., selected on multiplicity, and vertex, not integrated over all collisions. Work needed to be completed and documented. - MC correction factors as function of centrality, tube and vertex. - Careful comparison of calculated vs. measured spectra at lower multiplicities to see how well these in fact are reproduced. (For central the MC seems to give a lower average #hits, but if this is due to a real higher value or an underestimate of the background is not known. - A description of the algorithm to evaluate dN/dEta, in particular the statistical errors. - The same question on systematical error sources and associated values as for tiles is valid also here. Results from Si-detectors. This work is also progressing, I am adding below some information I got from Steve. These will give us another handle on the multiplicity measurements. The obvious question is should these be included into the draft paper. Will the paper be too long for a letter? (SJS) " Although calibration issues have slowed the development of dn/deta plots using the si data, I believe we now know the origin of our problems (primarily a VERY non-linear PTQ response and low amplitude) and know how to correct for them. We are still working on a consistency check of the Si calibration. The calibration of the Si detectors was nominally done by 241Am source measurements. However, to confirm this calibration procedure works when we are actually measuring near-MIP particles traversing the counters, we are relying on the location of the single MIP peak and how this peak tracks with the vertex location. " Centrality selection (added 11/2). Steve also reminded me about the definition of centrality, and percentages. So far most has been done with the tiles themselves. But as seen in several plots 6% in Tiles is not the 6% most central in Beam Beam, though close (see figs in DNP talk). If the aim is to compare with other experiments a better choice might be to have a cut on ZDC sum energy (on the more central part of the curve) and accept events below such a value. This is a well calibrated detector and for percentages in 0-20% should be well defined. This is already quite a list of items, and most likely is not complete. I do think all of these have to be answered before a publication can proceed. ------------------------------------------------------ Flemming Videbaek Physics Department Brookhaven National Laboratory tlf: 631-344-4106 e-mail: videbaek@bnl.gov ------------------------------------------------------ Dr. Flemming Videbaek Physics Department Brookhaven National Laboratory tlf: 631-344-4106 e-mail: videbaek@bnl.gov
This archive was generated by hypermail 2b29 : Sun Nov 05 2000 - 11:12:55 EST