Simulation Meeting Minutes
             ==========================
                                            Feb 28,2001
                                            S. Protopopescu

The simulation meeting of Feb.28 was dedicated to a discussion
on how to handle handle calorimeter data to ensure that
Monte Carlo and real data can be treated on an equal footing.
It was agreed that it would be beneficial to have both
types of data look the same as early as possible in the
Monte Carlo generation process.

In order to accomplish this the following decisions were made:

1) The CalWeight package will be called in D0Sim before the MergeMinbias
   package to convert the hard scatter d0g* CalDataChunk to
   a CalDataChunk weighted so as to be the same as
   the CalDataChunk generated by CalUnpToMC package (starting
   from calorimeter UnpackDataChunk). For now it was
   agreed that the weights should be the ones optimized for
   calorimeter data only. This has the following implications:
   a) Once the weights are applied the energy scale for mixture
      or plate d0g* generation should be the same and it becomes possible
      to combine events without checking first that they are plate or mixture. 
   b) In the MergeMinbias package the weights will need to be applied to
      the minbias CalDataChunk *before* being added to the hard scatter
      (but only if the minbias is from d0g*, not if it is from real data).
   c) To first order no additional weights will be needed when
      making L1 cal trigger towers (as they are supposed to be weighted
      to correspond to calorimeter only optimization). In practice there
      may be differences in the actual weights implemented in the
      trigger and a mechanism will need to be introduced in the code
      to handle that.

2) The CalUnpToMC and CalMCToUnp packages need to be modified to separate 
   the conversion from ADC to GeV from the weights so that the same optimized
   weights  can be applied to the real data. The CalDataChunk used as input
   to make a RawDataChunk must be identical to the CalDataChunk that one gets
   back unpacking the RawDataChunk and the energy scale should be the
   same for real data and MC so they can be combined. 

3) Because the CalDataChunk will have data weighted with calorimeter only
   optimized weights any correction to energy using preshowers will need
   first to reweight the calorimeter data to correspond to optimized
   weights using calorimeter + preshower. There is some benefit in this:
   em objects that have no preshower energy do not need reweighting.

What drives items (1) and (2) is to give D0Sim the capability of merging
MC generated events with real data events.


In order to do this we plan:

4) Create a program (D0RawToSim?) that can take RawDataChunk and convert
   it to pseudo SimXXXChunk + CalDataChunk (so it looks like d0g* ouput).

A point that was missed during the discussion was that step 1b) is not
necessary if CalWeight is called in d0g* rather than in d0sim. But we
may want to live with the additional complication in order to gain flexibility.