Weeks of 11102007

1
Weeks of 1/1-10/2007

• Getting back up to speed
• Worked on Poisson fit problem (glitch at 1400
  MeV/c2)
• Rearrange tasks in the way fits are done

Old Way
New Way

• Make fit
• Read either Data or Toy MC sample
• Write results file.
• Make toy MC ntuple

• Make fit
• Read either Data or Toy MC sample
• Write results file.

• Process output file
• Make toy MC ntuple
• Make plots in PAW

• Process output file
• Make plots in PAW

Make tests using the toy MC ntuple
2
More Tests on Toy MC Using Square plot and
Poisson fit method (WMD/VZ)
3
Square Plot/Poisson Fit(Perfect Toy MC!)
Replace binned Toy MC input by bins with expected
populations !
Phewww !!
4
Square Plot/Poisson Fit(Almost Perfect Toy MC!)
Replace binned Toy MC input by bins with expected
populations rounded to nearest integer value !
Phewww !!
5
Square Plot/Poisson Fit(Realistic Toy MC!)
Replace binned Toy MC input by bins with expected
populations randomized with Poisson distribution.
Puzzling !!
6
Square Plot/Poisson Fit(Fit to Actual Toy MC
Sample)
Simply use the Toy MC sample.
Puzzling !!
7
Square Plot/Poisson Fit(Exclude Boundary Region
from Fit)
Simply use the Toy MC sample. Exclude boundary
regions from fit.
Puzzling !!
8
Reminder plots
9
Poisson Fit (Isobar Model)
Slight discrepancies Binning may not be
optimum Reasonably satisfactory!
Starting from very far away, get essentially
correct solution. ? Poisson method per se is
correctly implemented.
10
Poisson Fit (EIPWA)
- Serious Problem
11
Problem with Splining?