From: Claus O. E. Jorgensen (ekman@nbi.dk)
Date: Mon Jun 30 2003 - 08:32:16 EDT
Hi JH, > Let me ask a couple of questions on plots in the latest draft. > > 1. The yield at eta~2.2 has been significantly increased > (by a factor of "~2") comparing figures shown at colloquium and the one > in the latest draft. Is it due to a simple mistake in the previous plot making > or "new and improved" analysis? > > 2. I don't understand why R_AA(eta~2) is similar as before. If you look at > HIro's figures (http://pii3.brahms.bnl.gov/~hito/pp/UA1_Correction.html) > at pt = 3-4 GeV/c, there is a factor of ~2-2.5 difference in the correction factor > between h-(2<|eta|<2.2)/h+-/2 (|eta|<2.5) and h+-(2<|eta|<2.2)/h+- (|eta|<2.5). > This should change the R factor significantly at eta~2.2. Why no change? > Are we looking at obsolete figures? I understand your concern. I had a stupid factor-of-two mistake at some point (h-/2 vs h- !) that could have propagated to some of the figures and might be able to explain the two points above. I'm surprized if the figures (yields and Raa) are not in sync, but at this point it's impossible for me to go back and check the different changes in the code and connect them to different version of the figures. The plots now shown on the web-page are the latest and they should be in sync and (I'm sure) without any obvious mistakes. > > 3. You mentioned about the systematic errors in the previous message: > "...................A lot of the systematic errors cancel out > when you make the central to "semi-peripheral" (Rcp) ratios, but not the > errors introduced in the Nbin calculation! Actually this is to 1st > approximation the only systematic error in this ratio. Errors introduced > by correcting for acceptance will cancel out, errors in efficiencies will > cancel out (the dependence on centrality is very little, especially for > high field runs) and of course the reference normalization will cancel > out." > I don't think it's quite true. As Dana pointed out, there is kinky feature in > the 40-60% ratios in Fig. 2, which is much bigger than statistical > uncertainties. This might be due to some unknown centrality dependences > (background?) which don't cancel out. That means there is some > point-by-point uncertainties don't cancel out. There is another overall > uncertainty from different systematics in normalizing centrality on top > of uncertainties from using Nbin calculated from models. > They might be some overlaps between two of them, but certainly > they are different kind of uncertainties. You're right - of course not all the systematic errors cancel out... Cheers, Claus
This archive was generated by hypermail 2.1.5 : Mon Jun 30 2003 - 08:36:25 EDT