----------------------------------------

E895 Meeting Minutes at LBNL

----------------------------------------

taken by Dieter Best, Tue Mar 10 1998

Attended by: Gulshan Rai (GR), Dieter Best (DB), Jerry Chance (JC)
Sergei Panitkin (SP), Heng Liu (HL), Jim Draper (JD)

A) NEWS ITEMS:

        Some changes will occur on PDSF in 3-4 weeks which may cause
        minor (hopefully) disruptions to PASS1.

        8 dual PENTIUM  CPUs and one 4 processor SUN workstation has been ordered for
        STAR/Grand Challenge/Mock Data. These machines will form a mini cluster on
        PDSF and will not be available to E895 or other users. These 9 machines will
        replace 9 existing SUN machines and therefore represents a loss of processing
        for E895. The HPs are unaffected but E895 competes for time with ATLAS and 
        NA49.

B) DATA PROCESSING:

        PDSF SUNs are runing 4 AGeV Au+Au. About 300 files have been processed,
        about 120 are pending analysis.

        PDSF HPs are running 2 AGeV Au+Au. 20% processed.

        PENTIUM are running 6 AGeV Au+Au. 10-15% processed

C) SGIs (by JC):
        We have a working eosgx and pass1 executable on the 64 bit SGIs.
        Requested access to additional SGI(s)

        Showed dE/dx for Au+Be at 4 AGeV:
        o  2D hit finder has "bounce" sheet in data ==> 50% more junk tracks.
        o  1D hit finder has "bounce removal" and is 5-6 times faster.
        o  BTF tracking works very well in low multiplicity environment

D) Ntupling (by HL):
        o  Converted 100 2 AGeV AU+AU xdf files to ntuples.
        o  Showed gain variation (dedx_scale) as function of run number.
        o  Conclusion: dedx_scale necessary since variation in gain
        as function of time can occur. Do this for all the data.

E) Handling of auto correlation in flow analysis (by JD). Long discussion centered 
   around one transparency. Bottom line, JD argues weighting not required for the 
   particle being removed from the Q vector determination. Not everyone was convinced.
   Postscript: GR circulates his own version arguing the opposite - the jury is out, 
   no verdict yet.


F) ACTION ITEMS:
        1. set up additional SGI
        2. run benchmark files on SGI