The “Other” STAR-PHENIX Discrepancy Differences in the f analyses Charles Maguire Vanderbilt University L/H PWG meeting August 25, 2005

PHENIX-STAR discrepancy Outline Numerical statement of Run2 discrepancy Update from QM’05 presentations Systematic comparisons of Run2 data It’s not just a one data point error effect (Rafelski conjecture) Action plan for further studies Plans for eventual contact again with STAR (and Rafelski?) Previous short contact before QM’04, no resolution 8/25/2005 PHENIX-STAR discrepancy

Run2 discrepancy factors for dN/dy 1.69 +/- 0.27 (0 - 10% centrality) Current Status of the Run2 Discrepancy Looking at the integrated yield numbers dN/dy Publications from Run2 Au+Au data at 200 GeV STAR in PL B 612 (April 2005) 181-189 PHENIX in PR C 72 (July 2005) 014903 One common centrality bin and minimum bias data can be directly compared using exponential fitting Centrality STAR dN/dy PHENIX dN/dy STAR T PHENIX T 0 - 10% 6.65 +/- 0.35 +/- 0.73 3.94 +/- 0.60 +/- 0.62 357 +/- 14 +/- 39 376 +/- 24 +/- 20 Mininum Bias 2.40 +/- 0.07 +/- 0.26 1.34 +/- 0.09 +/- 0.21 353 +/- 9 +/- 39 366 +/- 11 +/- 18 Run2 discrepancy factors for dN/dy 1.69 +/- 0.27 (0 - 10% centrality) 1.79 +/- 0.13 (minimum bias) statistical errors only 8/25/2005 PHENIX-STAR discrepancy

Update from QM’05 Presentations Both STAR and PHENIX presented Run4 yield results Both stated that their Run4 results agreed with their Run2 results By inference the integrated yield discrepancy persists for Run4 Questions were asked about the STAR-PHENIX discrepancy Obvious interest in getting this discrepancy resolved STAR had a back-up slide not shown quoting a Rafelski et al. preprint which questioned the lowest PHENIX mT data point in 0-10% centrality (that back-up slide will be shown later in this presentation) STAR and PHENIX presented the first v2 results for the f STAR’s result tended to be higher than PHENIX’s result but the error bars are too large to conclude that there is a discrepancy The yield discrepancy may diminish the credence of the v2 quotes ? PHENIX declined to show RdAu results for the f STAR declined to show Run4 RAA results for the f but PHENIX did show Run4 RAA results claiming f suppression In its Run2 paper STAR had shown an absence of suppression for the f 8/25/2005 PHENIX-STAR discrepancy

Minimum Bias Data Comparison I Two data sets are plotted The two sets of data are fitted by exponential functions. The same mT range is used in both fits,omitting low mT STAR points No significant change in STAR results as compared with using their full minimum bias mT range 8/25/2005 PHENIX-STAR discrepancy

PHENIX-STAR discrepancy Minimum Bias Data Comparison II Same data sets are plotted Same mT range used but omitting lowest mT PHENIX data point Slight changes in both dN/dy values, but the discrepancy ratio is still the same at 1.7 +/- 0.2 (statistical) The fact that STAR and PHENIX disagree at the intermediate pT values means that STAR doesn’t see f suppression but PHENIX does 8/25/2005 PHENIX-STAR discrepancy

PHENIX-STAR discrepancy Most Central Bin Data Comparison I The 0-10% data sets are plotted Same mT range used including lowest mT PHENIX data point but omitting the lowest STAR mT points No significant change in STAR’s result as compared with using all their data. 8/25/2005 PHENIX-STAR discrepancy

Most Central Bin Data Comparison II The 0-10% data sets are plotted. Same mT fit range used but dropping lowest mT PHENIX data point. Significant changes in STAR’s results with just one more data point omitted: T jumps from 344 to 395 MeV dN/dy drops from 6.77 to 4.84 STAR’s lowest 6 mT data points appear to be on a different exponential line compared to the highest 6 mT points in the 0-10% bin. This was not the case for the min bias bin: issue of STAR’s “kink” analysis for higher pT kaons? No significant change from PHENIX’s original result with lowest mT point. Discrepancy ratio becomes 1.3 +/- 0.3 8/25/2005 PHENIX-STAR discrepancy

RAA and RCP From Run2 STAR (PLB figure) STAR sees almost the same RCP ~0.5 for the f as PHENIX in the Run2 data STAR sees no RAA suppression for the f in the 0-5% centrality bin from Run2 for the pT range 1.5 to 3.4 GeV/c STAR sees RAA ~1.5 enhancement for the f in the 60-80% centrality bin for the pT range 1.5 to 3.4 GeV/c 8/25/2005 PHENIX-STAR discrepancy

RAA and RCP from Run4 PHENIX (QM’05 prelim figures) PHENIX claims RAA ~0.4 suppression for the f in the 0-10% centrality bin for the pT range 1.5 to 2.1 GeV/c comparable to effect seen in the p0 Suppression decreases in the 60-90% bin, again comparable to the p0 RCP for the f comparable to what STAR has published for Run2 8/25/2005 PHENIX-STAR discrepancy

Unshown STAR Backup Slide from QM’05 (but it shows what they are thinking) Johann Rafelski etc. nucl-th/0412072 Rafelski et al. conjecture that the lowest PHENIX mT data point is a factor of 1.5 too low and this drives the low dN/dy value for PHENIX. Since STAR has more low mT data, then STAR’s result is more believable 0 - 10% centrality STAR data 0 - 10% centrality PHENIX data My opinion (see next slide): By using a linear plot Rafelski et al. obscure the difference in normalization at the higher mT points. Rafelski et al. also ignore the minimum bias data discrepancy. 0 - 10% centrality STAR and PHENIX 8/25/2005 PHENIX-STAR discrepancy

Checking the Rafelski et al Conjecture Johann Rafelski etc. nucl-th/0412072 Rafelski et al. conjecture that the lowest PHENIX mT data point is a factor of 1.5 too low and this drives the low dN/dy value for PHENIX. Since STAR has more low mT data, then STAR’s result is more believable Actual fact: Only if all of PHENIX’s data points were increased by a factor of 1.5 would there be agreement with STAR’s results. A change of 1.5 in just the lowest PHENIX mT data point gives very little change in the fit result for dN/dy and T. Changes are actually well within the fit uncertainties. 8/25/2005 PHENIX-STAR discrepancy

Question about RdAu for the f in PHENIX Values appear to be systematically low? 8/25/2005 PHENIX-STAR discrepancy

Possible mistakes made by PHENIX analysis Numerical error Simplest explanation is that there is a persistent numerical mistake of order ~1.5 in computing final yield values Difficult to believe since there have been more than one independent analyses of the yield values (UCR, WIS, and VU) for Run2 and Run4 Monte Carlo error fiducial volume Real dead areas not completely accounted for in simulation Difficult to believe since there have been extensive real vs MC acceptance checks in both East and West Arms Such an error would affect all analyses, but pairs worse than singles Monte Carlo error radiation length Assume PISA omits a significant 1-2% radiation length; would have to be after DC/PC1 in view of photon conversion studies Increased absoprtion would attenuate lower momentum kaon Assumption needs to be systematically studied with PISA changes Would also want to check the effect on the proton acceptances 8/25/2005 PHENIX-STAR discrepancy

Initial List of Action Items Embedding for Run4 The embedding effort for Run4 was not completed in time for QM’05 Dedicated embedding effort needs to be resumed for Run4 Who will do it and where? Monte Carlo radiation length checks Will first check current sensitivity, e.g. how much energy is being lost by kaons and protons traversing all the way to the EMCal or TOF? Will add 1 and 2% additions at the PC1 region Checks will be made on proton and kaon acceptances (CFM will oversee this) Monte Carlo fiducial volume re-checks There are gaping holes in the the Dch acceptance which are run dependent. Are these being properly accounted? Any serious fiducial volume change affects all analyses Schedule for external contacts with STAR and Rafelski et al.? Want to avoid the issue being “settled” by a theory preprint Should we send a “work-in-progress” set of plots to each? 8/25/2005 PHENIX-STAR discrepancy