Validation of Pythia using Run1B J/psi sample

• In Sept we switched from Pythia 6.115 to Pythia 6.129
• what did we learn since then:
  • decay table (user-visible numbering of decay channels) have changed in Pythia 6.129
  • number of final state particles went down by a factor close to 2 in Pythia 6.129. Since this observation was found to be process-independent we believed that it has something to do with the fragmentation
• Stan found which Pythia parameters were responsible for the observed change in the multiplicity. To get back to Pythia 6.115 one needs to set the following parameters:

          parp(81) = 1.4           (was 1.9  in Pythia 6.129)
          parp(82) = 1.55          (was 2.1  in Pythia 6.129)
          parp(90) = 0             (was 0.16 in Pythia 6.129)
  

  after this is done, one can use Pythia 6.129 and get (practically) the same multiplicities as with Pythia 6.115
• nevertheless the question remains which defaults are closer to reality
• we used Run 1B data - inclusive J/psi stream by the B-group
  • BPAD's (by Slawek and Jonathan) can be found at cdfsga:/data17/bottom/psia_7b
  • generated ntuples (in UC format) used for validation are available
    from cdfsga:/data69/upgrade/run1_data/psimimu
  • see MC validation page for more details, the description of ntuple format can also be found there
• Reference distributions for the pt and eta tracks of the reconstructed tracksin .eps and .gif show the CTC acceptance
• The distributions for the number of tracks and the number of tracks plotted versus instantaneous luminosity (in units of 10^30) in .eps and .gif. From these one concludes that at zero luminosity (no extra MB events) there is about 33 tracks in CTC acceptance. For the estimate we assume that CTC effectively covers 3 units of eta.
• To verify that we understand the dependence on the instantaneous luminosity we plotted similar distributions for the number of primary vertices reconstructed in the event - see the distributions in .eps and .gif
• One can see that at Lum=0 offline code reconstructs 2 vertices in average. This means that we are not counting the right vertices. The right one are those with class=12 (good vertices). See the distributions for the vertices of class 12 in .eps and .gif
• both the intercept and the slope make sense:
  • at zero luminosity one would expect to reconstruct 1 primary vertex and this is what we see (1.0 +/- 0.13)
  • given the bunch spacing of 3.5 microseconds and the total ppbar cross-section at sqrt(s)=1800 of 51mb one would expect the slope dNv12/dIlum = 0.178, we see 0.16+/-0.04
• assuming that effectively we cover 3 units of eta with high track reconstruction efficiency(=1), we get dN/deta = 11 in the central region for inclusive J/psi events
• the MC part: we run Pythia 6.129 with 2 sets of the parameters mentioned above: with the defaults of Pythia 6.129 (set 1) and with the defaults of Pythia 6.115(set 2)
• the simulated subprocesses:

   ==============================================================================
   I                                                                            I
   I             PYTHIA will be initialized for a p on pbar collider            I
   I                  at   1800.000 GeV center-of-mass energy                   I
   I                                                                            I
   ==============================================================================
  
   ******** PYMAXI: summary of differential cross-section maximum search ********
  
             ==========================================================
             I                                      I                 I
             I  ISUB  Subprocess name               I  Maximum value  I
             I                                      I                 I
             ==========================================================
             I                                      I                 I
             I   86   g + g -> J/Psi + g            I    8.3168D-03   I
             I   87   g + g -> chi_0c + g           I    6.7199D-02   I
             I   88   g + g -> chi_1c + g           I    2.2754D-02   I
             I   89   g + g -> chi_2c + g           I    8.1032D-02   I
             I                                      I                 I
             ==========================================================
  

• the multiplicities of final state particles (ISTDHEP=1) are shown here on the top plot. The bottom plot contains the distributions for the pseudorapidities for the charged particles with Pt.gt.0.3 The distributions, corresponding the the defaults of Pythia 6.129 are in green, the distributions corresponding to the defaults of Pythia 6.115 are shown in light-brown.
• there are 1000 events in the these distributions, the distribution for pseudorapidity has bin 0.1, so from the bottom plot we derive dN/deta ~= 4.0 for Pythia 6.129 and dN/deta ~= 7.0 for Pythia 6.115.
• one could expect increase in the multiplicities by about 15-20% due to the secondary interactions, so if we were simulating inclusive J/psi events we'd expect dN/deta ~= 8.5 for Pythia 6.115, the same number for Pythia 6.129 (with its defaults) would be dN/deta ~= 5 in the central region (we are assuming track reconstruction efficiency to be 100%).
• given that we see dN/deta ~= 11 in the data, the defaults by Pythia 6.115 seem to be more realistic. We think that we need to use them to generate the data for for MDC1.

Discussion

• it would make sense to contact T.Sjostrand and ask his opinion
• Rick Field has been comparing Pythia to minbias data, we need to talk to him also (CDF-5067)
  
• also of relevance is the CDF note CDF-4097