Slow Simulator Description
updated Tues. Nov. 9 15:00 CST 2004
H. Spinka (ANL)

  • Goal: Add Poisson statistics to simulated SMD, pre- and post-shower simulated outputs
  • Method: Convert Geant energy deposited to mips, convert to photoelectrons (p.e.), apply Poisson statistics, reconvert to energy, and ultimately to ADC value.
  • Implementation Plan:

    * read in MC event with EEMC data. Obtain the Geant energy in the strip. Use gains from the database.
    * find the number of mips equivalent.
    * convert mips to p.e.
    * apply Poisson statistics.
    * smear the values with an input for the 1 p.e. resolution. Resolutions for N p.e. are sqrt(N) times bigger than for 1 p.e.
    * perform the inverse to get back to ADC values.

  • Testing: plots illustrating the above algo

    * Use 2.0 p.e./mip for SMD strips and 3.9 p.e./mip for pre- and post-shower counters.
    * Assume a mip is 1.33 MeV in both SMD and pre- and post-shower elements.
    * Put the resolution for the 1 p.e. peak as 0.1 p.e. in order to show the individual p.e. peaks.
    * Use MC file rcf1210_164_4219evts.MuDst.root, a set of minbias events.

    1. Minbias data. Blue curve is for generated ADC spectrum in the SMD strips, all of which have the same gain in the MC. Red curve is the result after applying the Poisson statistics. The number of events drops from 2744 (mean = 28.49) to 1926 (mean = 40.73) when it is assumed the light yield is 2 p.e./mip for all strips. The "total light" is (events)*(mean) = 78177 and 78446, respectively. Note the flat distribution in the MC generated events (before Poisson corrections) at low ADC values is presumably from the triangular shape of the SMD strips.

    2. Same results for pre- and post-shower channels. The number of events drops from 1671 to 1565 for 3.9 p.e./mip. Note the peak in the initially generated ADC spectrum is presumably from mips in the pre-shower tiles, and there may be some low energy events from the post-shower tiles included.

  • More Realistic Input Parameters: (9 Nov 2004) Revised parameters are given below:

    * 3.9 p.e./mip for pre- and post-shower tiles (unchanged). This is the product of 2.6 p.e./mip for tower tiles with a factor of 1.5 more light for the special scintillator used for these tiles (from S. Vigdor). The post-shower tiles should have lower light output of 2.6 p.e./mip - not yet done.
    * energy loss for mips in polystyrene scintillator from PDG = (1.936 MeV/gm/cm^2) * (1.032 gm/cm^3) = 1.998 MeV/cm. For the 7 mm thick SMD strips I use 1.40 MeV/mip, and for the 5 mm thick pre- and post-shower tiles I use 1.00 MeV/mip.
    * I will use the SMD response in p.e./mip from cosmic ray testing at ANL initially. Possibly should add a fudge factor to allow for more light attenuation in the clear fibers and connectors in the actual detector compared to the ANL cosmic ray setup.
    * According to S. Vigdor, tests of the MAPMTs gave 1 p.e. spectra whose counts dropped to half the peak height at the pedestal. Assuming a Gaussian distribution, this would give a sigma = 1 p.e./sqrt(2*ln2) ~= 0.85 p.e. Using such a width will lead to the loss of more events whose simulated ADC values would be negative. Is this the correct thing to do?

  • 3. As above for SMD strips with more realistic 1 p.e. peak resolution, but still 2 p.e./mip for all strips. Initial events = 2744, output events with ADC>0 = 1881.
  • 4. As above for pre- and post-shower tiles with more realistic parameters. Initial events = 1671, output events with ADC>0 = 1605.
  • 5. Present status: work ongoing to add more realistic light yield, 1 p.e. resolution, etc. It is probably still a bit too early for testing with pi0 reconstruction algorithms.
    Please ignore!! This is a clickable plot: The plots with ratios _after_ scale correction are here. X-axis [0.5,1.5] , ( PS.gz).