From pinkenburg@bnl.gov Mon Jun 3 15:52:33 2002 Date: Wed, 03 Apr 2002 17:15:14 -0500 From: Chris Pinkenburg Reply-To: e895-l@bnl.gov To: e895-l@bnl.gov Subject: comments for pion paper Hi Jenn, it took a bit longer and I am not entirely through (20 pages just take longer than 4). It's nice to see that the pion stuff gets wrapped up in a longer paper. My main concern are the errors which I don't think are estimated correctly. If I take e.g. the yields face value the error is 0.5%-2% which just doesn't sound right after the elaborate and complex procedure with which they are extracted. ***I LOOKED INTO THIS MORE CAREFULLY AND HAVE TRIED TO ELABORATE ON WHERE THE PID METHOD MAY CAUSE PROBLEMS AND WHERE WE HAVE ASSUMED THINGS ARE OKAY. THE TOTAL YIELDS HAVE ~5-7% ERRORS ADDED Lets get to the finer details: in the abstract, 4 pi momentum spectra sounds strange. Maybe I am not up to date on the lingo but over nearly full rapidity range is probably better. The same for the 4 pi yields, total yield may be more apropriate. ***CHANGED Introduction: Somehow the word copious got stuck with the pions since the arguments about the elliptic flow. From what I learned from a dictionary it means: abundant, too many. While at RHIC that's certainly true but at Bevalac/SIS? I still hear Dan complaining about the few pions with which the EOS pion flow was done at 1 AGeV (same for FOPI but we had more events). In general you really need some references here, C. Hartnack/S. Bass did a lot of IQMD stuff with the delta, B. Hong PLB 407 (1997) 115 did delta abundance in Ni+Ni at 1-2 AGeV, D. Pelte Z.Phys A357 (1997) 215 pion multiplicity in Au+Au@1AGeV. I am not sure what to make out of the sentences about the models, there are also a lot of BUU type models around which do well and the fluid dynamic part is probably more important for flow effects which are not part of this paper - RQMD seems to do fine on the longitudinal flow. Somehow the introduction doesn't introduce the baseline about what you are going to present and why we went through the trouble measuring and analysing this. ***THE INTRO IS COMPLETELY RE-WRITTEN Data collection and analysis: It was hard for me to understand the description of the procedure and I guess someone less familiar with E895 will just be lost here, but I don't have any suggestions how to do better. Is it correct that you made ~7000 projections for each single point in the mt spectra (resulting in a stunning total of nearly a million histograms)? What kind of quality control did you do - experience says paw/root fits everything without complaining too much or failing. Instead of the Kaons the e+/e- are your real problem when it comes to numbers if you look at fig 2. Where did you get their multiplicity from? ***NOT MUCH TO SAY - ITS HARD TO UNDERSTAND IF YOU HAVEN"T SAT DOWN AND DONE IT. ADDED INFO ABOUT ELECTRON CONTAMINATION, THOUGH Results: maybe mentioning the 5% centrality more often reminds the reader what event class is shown instead of just "central" event. I know it's mentioned in the technical section but I don't think repeating that hurts. ***ADDED BASICALLY EVERYWHERE >From page 3 to page 5 there is a long explanation about what formalism to use and it ends in saying we don't know what it means in physics terms. I also don't see what the point of these paragraphs are. Again why did we do this, did we learn something? ***I RESTRICTED THIS DISCUSSION TO THE PARAMETERIZATION WE ACTUALLY USED. THESE DAYS THE VOGUE MODELS ARE BLAST-WAVE, ETC., WHICH AT LEAST TRY TO INCLUDE FLOW IN SOME MEANINGFUL WAY. BUT I DROPPED THE THREAD HERE AND WE WON:T WORRY ABOUT IT. THE GOAL IS TO GET A REASONABLE FIT TO THE SPECTRA FOR EXTRACTING YIELDS AFTER ALL For the yields, as I mentioned at the beginning, the stated errors are way too small to be true. ***SEE ABOVE page 7: I stumbled over the we have shown longitudinal flow - I guess that's a leftover from a reorganization, should say we are going to show... Why did you use a gaussian fit when you expect for physics reasons a 1/cosh(y) behaviour? If you know it doesn't make a difference you have done that exercise, so why not use the apropriate fitting function? A reference for the identical high mt slope of pi+/pi- would be nice to add ***REMIND ME AGAIN IF YOU THINK IT IS CRITICAL The criterion to put an x into Fig 11 looks a bit arbitrary to me (at least the x's don't seem to follow any systematic trend) Putting the pion yields in some context with other measurements at different energies or systems would be nice. Otherwise they are kind of left hanging in midair here. ***I TRIED TO CLARIFY THE "x"'s A BIT AND THE PION MULTIPLICITIES ARE DISCUSSED WITH OTHER ENERGIES LATER ON. page 9: I don't think the direct pion pair production holds for pi0's (can be more than 2 - though I wouldn't look for centauro events in E895), only for pi+/pi-. ***TOOK OUT THE pi0-pi0 STUFF RQMD comparison: For consitency the RQMD results should be shown only for a single impact parameter range which conforms to the data, not b<5 and then b<3. I wonder if the strangely varying temperatures are not an artefact of the fitting. Since the transition value between low mt and high mt is up to the fit you basically fit a concave distribution with two straight lines. For a comparison with data, wouldn't it be more meaningful to make 2 separate single temperature fits in identical low/high mt ranges? I like the plot where the thermal and the resonance contributions are separeted. Maybe you can use this as a guide where the non resonance high mt should kick in. ***THE RQMD IS ALL b<3fm NOW. THE TWO-SLOPE MODEL FOR THE DATA IS JUST TO GET THE YIELDS. WITH THE RQMD WE TRY TO MOTIVATE WHY ONE MIGHT EXPECT THIS SORT OF BEHAVIOR. The figures with the lastcl are interesting but if your point is resonance/himass/particle decay versus thermal you should restrict the figure to these. That would also get rid of the himass sentence where the impression is that we don't know what we are showing. In the last sentence of this paragraph, you mention the mfilt distribution which were used to nail down the impact parameter. Since this is a sore point, could you show the ones you use? I wonder if yours and SUNY's are identical or if they look like the ones Bill used to match the impact parameter. Entropie: That's where the errors for E895 are needed (how come S/Np has an error when neither of it's input values has an error assigned). Is there any other E866/917 reference for Kaons than a phd thesis? If someone would use lambda/k0 multiplicities from a thesis as E895 values we wouldn't be too happy either. ***WE CHANGED THE ENTROPY ANALYSIS SIGNIFICANTLY. SEE WHAT YOU THINK NOW... Summary: You claim that RQMD does not reproduce the low mt differences in pi+/pi- spectra (which is probably true) but there is no plot/table with evidence for this. The claim that the high mt temperature rises with Ebeam doesn't really come out of the data (except 2AGeV) and that high mt pi+/pi- spectra have the same shape is one of your (well founded from other experiments but nevertheless) assumptions ***IT DOESNT REPRODUCE THIS AND I REFERENCE A PAPER BY Nu Xu WHICH SHOWS THAT IF HE ADDS AN AFTERBURNER IT DOES. ALSO, LOOK AT THE SLOPE PARAMETER TABLE FOR RQMD. YOU CAN SEE BY COMPARING PI+ and PI- THAT RQMD CAN'T GET THIS RIGHT. Figures: Fig1: The formula in the insert is fairly useless, it's not explained in the text and it doesn't seem to serve a purpose. If you want to show how well the bands are reproduced, why don't you overlay them with the fit? Also if you use a fine binned 2d histogram and instead of a scatter plot the bands will be more visible since the single points don't show up as pronounced. On the grayscale the Bethe Bloch curve should be visible. dca<2.5 doesn't mean anything outside of E895. ***I AM KEEPING THE SCATTER PLOT BUT I REMOVED THE INSET AND THE DCA Fig 2: maybe use dotted/dashed lines for different particle species to distinguish them better. *** I THINK THEY ARE CLEARLY ENOUGH DIFFERENTIATED BY THE LABELS Fig 3: How many bins? It's probably more honest to plot the 2 d histogram with grayscale z than making a contour plot where the plotting software tends to introduce artificial signatures. ***CONTOUR PLOTS ARE FAIRLY STANDARD WAY TO DO THIS. AN ALTERNATIVE IS A BOX PLOT IF YOU'RE DEAD-SET AGAINST THE CONTOURS. Fig5: Do the fits of 2-4 points have any significance? As far as I see the fit values are shown in Fig 11. ***I ADDED TEXT TO EXPLAIN THAT SOME OF THE LINES ARE NOT FITS BUT SUPERPOSED FUNCTIONS WHOSE PARAMETERS ARE EXTRAPOLATED FROM THE BETTER MEASURED REGIONS. Fig13/14: y axis not labeled ***IT's NUMBER PER EVENT. ANY SUGGESTION FOR A LABEL? Fig 19: what is this strange bracket next to the E895 label? ***NOT A STRANGE BRACKET - RATHER AN INDICATOR OF WHICH DATA POINTS ARE E895 Best Regards Chris P.S. Could you run a speel cheker on the paper, I found two occurances "checmical" on bottom of page 6 and "behviour" in the figure caption of fig 18 whihc would certenly have ben identifed by evn a bad sell checcer. -- ************************************************************* Christopher H. Pinkenburg ; pinkenburg@bnl.gov Brookhaven National Laboratory ; phone: (631) 344-5692 Physics Department Bldg 510 C ; fax: (631) 344-3253 Upton, NY 11973-5000 *************************************************************