Radphi Video Conference 10/28/2003
continued 11/04/2003

October 31, 2003
University of Connecticut
Storrs, CT

Meeting Agenda:

1. Progress on kinematic fit - D. Krop
2. Progress on BGV calibration - D. Steiner
3. Remaining tasks for NIM paper - S. Teige
4. Research plans - M. Kornicer
5. Next meeting or workdays - R. Jones

• http://dustbunny.physics.indiana.edu/~teige/nim
• http://zeus.phys.uconn.edu/radphi/call-notes
• http://zeus.phys.uconn.edu/~kornicer/radphi/LGDresolution
• http://zeus.phys.uconn.edu/radphi/calibration/figs.html
• http://dustbunny.physics.indiana.edu/~dkrop/kinefit
• http://cabibbo.physics.wm.edu/~steiner/radphi

Present: R. Jones, M. Kornicer, D. Steiner, D. Armstrong, S. Teige, D. Krop, C. Steffen, J. Marie

• Dan has made a lot of progress on his kinematic fit. His first work on the subject was to start off with a 2gamma sample and to try to isolate the single pi0 exclusive reaction. This was not successful, but some things were learned. First, it was noted that Monte Carlo showers at high energy (>2GeV) are reconstructed with significantly less energy than generated. This problem had already been noted by M. Kornicer. R. Jones is working on it. Second, it was noted how poorly the fit works when coupled to a mis-tag, that is when the initial gamma energy is taken from the wrong tagging counter. Besides introducing much more error in the reconstructed values, mis-tags also lead to a bias towards lower fit energies than the true (generated) values. There was some puzzlement about the reason for such a bias. No clean reason was found for this effect - something in how the mis-tag generator is coded?

  The ultimate reason for moving on from the pi0 exclusive sample was the realization that our 2-cluster resolution is marginally sufficient to resolve high energy pi0's. For symmetric decays the 2 clusters are only 5-7 cm apart at 5 GeV, which is at the limit of what can be resolved with 4cm blocks.

• Dan also reported his discovery of the decoding error that led to a mistaken value for the phi coordinate of recoils in the barrel. This fix has been worked out by R. Jones. It has now been checked into CVS.
• Dan now moves on to the study of exclusive omega's in the 3gamma sample. Here again he observed the low-side bias of reconstructed showers in MC, and reproduced his earlier observation of difficulties when the incident gamma energy is unknown or wrong. His chi2 is not quite flat from his fits to MC data. R. Jones said this is expected due to non-Gaussian nature of errors. The "pulls" in the 3gamma system are very narrow, but the (phi,theta) recoil are very wide. The Theta recoil is also shifted. Two cuts are introduced to help reduce background in the real data sample:
  • minimum cluster separation cut
  • minimum gamma energy in event
  After this, the tail of the pi0 disappeared and the spectrum is more clean, but nearly all omega's are preserved. Dan decided to use a linearized form of (theta,phi)recoil chi2 contribution to better study the tails of these distributions.

  At the end of the discussion, Dan showed pull distribution for (theta/phi)recoil that peak at zero. The theta distribution still has a large and asymmetric tail, so that its average, as opposed to its peak, is far from zero. This should be understood.

• Dan asked what should be done to deal with accidental tags. D. Armstrong suggested that he separately reconstruct each tag inside a suitably chosen coincidence window, and produce N renditions of each event. R. Jones explained the accidentals subtraction procedure that can be used to eliminate this double counting at the end of the analysis. He also pointed Dan to the code that uses the latest timing calibration results from D. Steiner.