Some thoughts on long-term requirements for acceptance calculations

- Will Brooks

1) We may need to carry more information than presently allowed in MCTK/MCVX - e.g. partial wave information. Need another bank.

Could make as flexible as possible. Different experiments may have quite different requirements. Use reserved bank numbers for particular purposes, e.g. "bank number 3-12 reserved for E1 run group, and for these x_int represents a partial wave id and x_float represents the amplitude."
Possible structure:

!
TABLE MCX
!
!    Extra Monte Carlo Information Bank
!
!COL ATT-name    FMT    Min    Max              ! Comments
     1             x_int           I         0     0xFFFF         ! Generic integer flag
     2            x_float         F   -1.0   +1.0                 ! Generic floating point number
!
END TABLE

Alternative: could design a new bank for each purpose, start with E1 and G1 running periods. Could name the bank after the running period to get a unique name.

At a minimum, we definitely need an event weight. We may also want a flag to indicate the event is accepted by the trigger. Could also have flags for gsimko, tsim, out-of-time tracks, background@luminosity.

2) If we're going to use GSIMKO and TSIM, we need an accounting system.

Steps to producing an acceptance calculation:

Define the set of data files to be analyzed.
Run the data through a particular reconstruction program release -> collection of reconstructed data files. Each file got its calibration constants from the map. Each file represents a certain amount of integrated luminosity.
Run GSIM for some large number of events -> collection of GSIM files. Now run GSIMKO on these files, to produce files with proper missing channels; must preserve the fraction of luminosity acquired with a given dead channel map. Must have a complete map before the GSIMKO run. (Must run electrons through this process as well as hadrons.) We scale the number of simulated particles to which a given dead channel map is applied by the fraction of luminosity of the data taken under those conditions. We cannot mix Monte Carlo input files with, e.g., different Q^2 ranges (unless they are somehow luminosity-matched). GSIMKO presently applies dead channel map to all the events in a given file. We either need to be able to apply a given map to a fraction of the events in a file, or have it write out to files of a different size, or have some other mechanism for handling this problem. Note: the acceptance, in detail, depends on the initial set of data files to be analyzed.
Run TSIM on all the GSIMKO output files. If the threshold or trigger definition was changed for any of the data files, the trigger will have to be applied to the correct fraction of the data, and TSIM should be able to be applied to subsets of the data in a given file.
Run the simulated data through the same reconstruction program release with which the data were analyzed.
In some program (which one?), bin the MC input data, bin the GSIM/GSIMKO/TSIM data in the same way, take the ratio, apply any event weight to the ratio, and then we know the acceptance in that bin.

(Insert diagram of above scheme here)

3) Do we need a background generator for acceptance calculations? One possible scheme is to generate a background library with GSIM and combine the data with the background after the simulation.

4) Do we need out-of-time tracks for acceptance calculations?