Minuit Alignment Update

1
Minuit Alignment Update
Jianchun Wang 04/10/00
Outline Monte Carlo generator MC
alignment Alignment of FPIX0 pstop at 15º Plan
2
Monte Carlo Generator

• Robs simulation code (template for user)
• simulate a track using beam condition
• compute intersection at each plane
• smear the position with a fixed resolution

• Reconstruction
• SSD simulation shown
• SPD has error

3
Monte Carlo Generator

• What I did (use Robs template)
• Add Npix-dependent resolution percentage
• Add inefficiency of the detector
• Output in proper format (skip Robs
  reconstruction)
• Useful files for people who want to try
• mc_main.F, mc_get_sigma.F, mc_sigma.txt
• in directory wangjc/testbeam/minuit_fit/mc_g
  en/

4
Monte Carlo Generator

• Comparison to run 6653 (FPIX0-pstop at 15º)
• after inefficiency and resolution simulation

5
Minuit Fitting

• Add control in tb_monitor.nml
• trk_control input, output, none
• trk_file track.dat
• Track selection criteria before minuit fit
• Only 1 track in the event
• At least 4 planes have hits
• At least 2 good y hits
• Kalman filter OK
• ?2 / ndof lt 4 (?)
• FPIX0 row 15-50, FPIX1 row 1-85
• Code send to CVS

6
MC alignment

• The ?2 of Kalman filter normalized to ndof shown
• The first one uses generator geometry for
  reconstruction
• The parameters are randomly changed as bad
  geometry (?7 0.278 ? 0.200 rad)
• The alignment takes 4-5 CPU hours (10K event)
• Several different geometry configurations tested

7
MC alignment

• Residual of FPIX0-pstop at 15º shown
• The resolutions with generator geometry and
  aligned one are both 5.81 ?m
• The resolution with bad geometry is 10.8 ?m,
  and the center is shifted

8
MC alignment
Resolution for different detector
9
Data Alignment

• FPIX0 pstop at 15o (run 6653), 10K tracks
  selected
• ?2 of Kalman filter before and after alignment
  shown
• The mean value is slightly improved from 1.10 to
  1.08

10
Residual of FPIX0

• The sigma of plane 7 residual changed slightly
  from 6.54 to 6.50 ?m
• It is quite possible that Charles alignment is
  good enough in this case

11
To be done

• get_sigma.F need to be updated
• Resolution simulation for NPIX1
• ????