An Independent Look at Parity Violation in the Strong Interaction via v 1 2 * v 2

1. An Independent Look at Parity Violation in the Strong Interaction via v12 * v2 Jim Thomas Lawrence Berkeley National Laboratory

2. The Signal: Separation of Charge wrt the RP • Kharzeev et al. have proposed that angular momentum is (globally) conserved in a heavy ion collision • If a chirally restored bubble is created in the collision, then positively charged quarks will go up … then hadronize … and yield an excess of positive pions above the plane (or the other way around) + + + • Familiar Tools: Perform a Fourier decomposition of the momentum-space particle distributions • For example, v2 is the 2nd harmonic Fourier coefficient of the distribution of particles with respect to the reaction plane - - - isotropic directed elliptic

3. The Observable • The coefficients for the Fourier expansion of the invariant yield are or • where the average is taken over all particles in the event and R is the known reaction plane angle (e.g. from the FTPC if we are using TPC data) • The second method is a true two particle correlation (many details left out) • Note that a ‘normal’ v1 measurement for pions in a Au-Au reaction has an intrinsic symmetry that suggests weighting by sign() • We won’t do this. We are looking for charge flow that goes up/down • We do weight the sum, so ‘normal’ v1 cancels out. Assumes symmetric  acceptance. • The signal is very small … so Voloshin proposed a very clever signal • Measure v12 * v2 because v2 is large and it amplifies the v1 signal. • The observable v12 * v2 is a very clever way to measure charge sensitive flow and, in fact, has been seen in several different collision systems and several different estimates of R • If then

4. Voloshin et al. see a signal • A signal appears to be there independent of how you determine the reaction plane. • Normally would assume that ++, -- have equal and opposite in sign to +- • Kharzeev suggests ‘bubble’ on edge of collision zone and one side absorbed Voloshin, QM08 Poster

5. A profound observation • This would be a very significant discover if it is real • Is it due to parity violation or something else? • Resonances? • Acceptance effects (aka dead sector in the TPC) • Voloshin et al. have submitted a ‘paper proposal’ to STAR whereby they have found a reasonably large signal for the flow of positive particles vs negative particles (++, --, not +- ) • It has been stewing for a while because the +- signal was not understood • In support of their case, they see the signal in Au-Au and Cu-Cu, as well as with several different ways to define the reaction plane angle

6. One potential problem … + - If the reaction plane angle is perfectly random, then this effect should go away … however, the P violating effect only translates to one or two particles per event so the randomization has to be nearly perfect

7. I’ll risk one provocative observation Phi Weights and Phi Acceptance are not symmetric for +,- particles Not really a surprise

8. But I was surprised when I calculated  • Assuming that I am not making a mistake (an assumption) then the observed event plane ‘sticks’ in the lab frame near 30 degrees. It is possible to imagine that when 20% of the events ‘stick’ in the lab frame then detector acceptance effects can affect small signals

9. Philosophy of getting involved • I thought I would try to take a completely independent look at the signal to see if I can find it … and if it exists, can I test it .vs. my own unique view of the systematic errors in the STAR TPC • Learn a little bit about flow analysis … • I’ve never done it before so I’m a complete neophyte • What I’ve learned so far is that V2 at RHIC is so large that any idiot can find it • said with a wink and a smile • The answer is yes. Dhevan (UCLA) can reproduce the signal. Alexei (Yale) can too. Vasily + JT (BNL & LBL) can too. Can JT do it in a solo effort? Can any idiot find a charge sensitive v12 *v2 signal?

10. Vasily & JT Methodology • Use standard STAR microDSTs and the Farm (not using picoDSTs) • Use standard STAR software tools for reading the mDSTs (i.e. not from the flow library) • Analyses appear to be simple enough that this is feasible • Double Blind (maybe even more blind than that :) • Seeking advice and wisdom from Flow Leaders but not Voloshin et al. • Not intended to insult anyone … just trying to be totally independent • Leave no stone unturned … or maybe I should say that because I am a novice, I am confident I will turn over a few stones that the experts think are ordinary & mundane stones • Use the power of the Scheduler and the Farm to do a few things that aren’t obviously efficient • Use different methods, including the cumulant method to analyze v1 * v22 • Make cuts on PID, charge sensitive weights, etc. • With a thorough knowledge of the techniques, ask if SpaceCharge, differential acceptance and other distortions can imitate the signal

11. JT code : cuts on the data and produces a summary output. Output of this step goes to analysis code written by Vasily • Statistics analysed: 14 M AuAu 200 GeV, Run 4 |Zvertex| < 30. dca < 3.0 0.15 < Pt < 2.0 15 < Nhits < 45 nHitPoints/nHitsPossible>0.52

12. Run 4, AuAu 200 GeV, 9.5 M (Vasily & JT) 10 x Vasily’s code produces similar results to Voloshin et al. Note: no acceptance corrections or phi weights

13. Run 4, AuAu 200 GeV, 9.5 M (Ilya & Sergei) The signal as reported by Voloshin et al.

14. What are the effects of PID? (Signal with pions)

15. Run 4, AuAu 200 GeV, ~14M ( ID)

16. Run 4, AuAu 200 GeV, ~14M (no e+/e-)

17. Run 4, AuAu 200 GeV, ~14M nHitDedxMin = 15

18. What does nHitsDedx do? It is a rapidity cut. nHitsDedx rejected tracks nHitsDedxMin = 15 nHitsDedx off

19. nHitsDedx leads to changes in phi acceptance nHitsDedx rejected tracks nHitsDedxMin = 15 nHitsDedx off

20. nHitsDedx: A Mystery … If you try to do any sort of particle ID, you must turn on this cut and the act of employing this cut appears to create false signals. • In order to select an identified particle, you have to request a track with a minimum number of calibrated clusters for dEdx measurements. Not all clusters that lie on a track are good for dE/dx measurements because the gain changes near the ends of pad rows and/or the gain changes near the endcaps and thus cannot be calibrated. • nHitsDedx selects the number of dEdx clusters on a track. The maximum possible is 45, with a mean number of hits of about 30, and so it is common to require that at least 15 of these are good dEdx hits before declaring that a track has a good particle ID. (This is different than > nHitsFit which merely counts the number of hits used in fitting a track.) • We found signals just by requiring nHitsDedx > 15. We did not have to go to step 2 and select a particle type... just turn on the nHitsDeDx cut without selecting a particle type and this will happen. The v12 *v2 signal is not unique to parity violation

21. A different approach: 3 Particle Cumulant results nHitsDedx Off

22. 3 Particle Cumulant results nHitsDedx Off Very similar results to Vasily and Illya’s analyses

23. Similar nHitsDedx behaviour as before … nHitsDedx = 15 nHitsDedx Off Anamolous results is robust across techniques … but now we have an additional tool that is easy to turn on and off

24. Add the cross terms and a very pretty result nHitsDedx = 15 • Cross terms take out lower order corrections including acceptance effects. Seems to work very smoothly. nHitsDedx = 15 • First time that I am aware of (not necessarily a strong statement) that cross terms do any work

25. v2 with and without the cross terms are the same without cross terms   with cross terms Very small changes for v2. Acceptance corrections and phi weights not really needed.

26. Conclusions • There is a signal for v12 * v2 which is present in the Au-Au data • Several different groups and different techniques can find it • The question is: what is it? • It may be the signature of parity violation in the strong interaction • This would be very profound and a wonderful discovery • It may be an artifact of complex acceptance cuts • If the reaction plane is not perfectly randomized with respect to the detector, then simple acceptance effects can lead to a signal • It may be the result of other physics • The v12 *v2 signal is not unique to parity violation • nHitsDedx is a quality cut that creates a false signal • Removing the lower order multi-particle correlations does a very pretty job of cleaning up the nHitsDedx acceptance correction … and didn’t remove the nominal signal. Very nice. • More to come, later …