1. Non-identified Two Particle Correlations from Run I Understanding drift chamber tracking – Tracker and candidatory – Two particle efficiencies/ghosts A first look. Momentum distributions, etc. First correlation function Gamow corrections And from here…

2. The current status of DC tracking As I understand it. – Tracker (old): Assumes x wires parallel to beam. Works in nominal dc position, even with minor changes in geometry Large fraction of tracks with reconstructed UV wires – Candidatory (new): Same pattern recognition philosophy as above. Built to withstand major changes in geometry (I.e. retracted arms) Better interface to the rest of dc software Cleaner separation of pattern recognition from hit association Most tracks are not reconstructed through UV wires (‘simple’ bug) – currently bypasses this problem by boot-strapping with PC1 What we need to use in the long run – but unusable now. (?)

3. DC tracking variables: Use bounded, natural variables for pattern recognition (I.e. not slope and intercept) Two particle effects occur at low  and  with the x wires and  and zed with the uv. – Can see from this that two particle inefficiencies occur at low relative  and  => almost no two track inefficiency in opposite charged tracks. 3-momentum reconstructed from fit to associated hits in PHDchTrackModel

4. Ghosting in tracker: Small (but measurable) fraction of tracks have a ghost pair at low  and . (=> low q) 3 solutions: – 1. Fix tracker to remove ghosts (in progress by DC group) – 2. Quantitatively understand 2- particle in/over-efficiencies and correct for it – 3. Cut it out You don’t have to think to hard to know which choice I made here … Cut in 2-dimensional  space.  Real Mixed

5. Ghosting in zed A look at the raw  (zed) distribution [top] also shows a clear ghosting problem. After cutting on |  |<.005 && |  |<.005, however, the peak in zed disappears I should probably make a full 3d cut, but this will do for this talk…  zed Real Mixed Real Mixed

6. Motivation for studying correlation functions* [A reminder or impetus for further personal study…] Amplitude for detecting two bosons (pions in my case) in above configuration is: Then the probability is the amplitude squared, and integrating over the source: x y * This slide stolen almost completely from W. Zajc, Proc. Of 4th Intersections Conference (1991) (r 2,p 2 ) (r 1,p 1 ) The 2 particle counting rate is related to the Fourier transform of the source. The normalization of  requires C(q=0) -> 2. Then for a simple Gaussian source: Want to know more?

7. Event characterization Runs: – 9979, 9981, 9987, 9988, 9992 – 20 files Events: ~20k – <½ in pairs I’ll show today Some naïve cuts just to get us started – |BBC zvtx| < 30cm (Track model works out to +/- 45 cm – > 50 tracks reconstructed (~central) => Pairs: – 760K + + – 430K - - – 1.1M + -

8. Kinematic distributions DC pattern recognition and track model built to work above 180 MeV. – Cut out low p t part – Unphysical Bug in track model (?) And a k t distribution … – ~700 MeV for all tracks – ~400 MeV for all tracks with q<100MeV k t (GeV/c)

9. q distributions q=|p 1 -p 2 | Before the ghost tracks cut the real distribution shows a strong peak at low q. After the ghosting cut in both the real and mixed the distributions are similar. (though we hope they’re not identical) Now as experimentalists, we determine the probability of detecting pairs at relative momentum q by measuring the ratio: A(q)/B(q) – A(q) => distribution of measured q from same event pairs – B(q) => from different (uncorrelated) events Real Mixed Real Mixed q (GeV/c)

10. Raw correlation function Recall: in ideal case, C(q)  2 But we don’t have ideal bosons since they are charged – leads to a depletion at low q due to the coulomb repulsion of the pair ++ --

11. Gamow Correction First correction: Gamow correction for particles from point source: Dependent on mass of particles in pair … Ratios of particles from Min Bias hijing as a function of q –  = 87.6% –  K = 9.3% –  p = 2.5% – KK =.3%

12. Gamow correcting opposites In red: – raw correlation function for opposite pairs In blue: – corrected for Gamow assuming a point source. It appears to be an over correction – not surprising since we don’t have a point source or if we did we should call a CERN press conference. +-

13. And Gamow Corrected… Now same charge pairs corrected for Gamow. Again, an over-correction at low q. (off scale) Low statistics. But, heck … let’s fit a Gaussian to that and see what we get. -- ++

14. Analysis Note #7 Simple study of PHENIX’s ability to resolve source size: – Throw two tracks into acceptance. – Impose artificial correlation on pair. – Reconstruct R from correlation. Questions: http://www.phenix.bnl.gov/p/info/an/007/

15. Comparisons – Slight beam energy dependence – strong centrality and k t dependence. – What’s shown here is for higher than most previous experiments. In the end we’ll do it as a function of k t for direct comparison The point: – First study – Correlation function is visible with a small fraction of the data set – Results of 1D fit are in the ballpark Root-sR Tside QM99.354.74Soltz et al..258.74.2Ganz.2517.25.0Ganz.41303.5-5.0Us

16. Lots still to do: Understand two particle ghosts/efficiencies/etc p/q resolutions PID! – Requires working DC code, track model and TOF Multi-dimensional anaysis Full coulomb correction Centrality/k t dependence Etc, etc, etc. Another postdoc at LLNL (Mike Heffner) has just begun additional studies on correlations David Brown, source imaging guru, joining theory effort in October Schedule to have publishable results by QM 2001