News

  • peak fitting code is separated in a new script: cxif5315/fit-peaks-2017-03-08.py
  • code is cleaned, purified, straight-forwarded and brought to "standard-command-line-interface" style
  • formalism for peak coordinate y(x,phi,beta,...) is checked.
    • It looks correct for l=0
    • for l=1 I am still doubt about 1-st equation. Other than that it is correct.
  • Sign conventions: sign of angle phi and beta returned by the fit to n peaks and saved in the peak list is the same as from direct angles evaluation using 2 peaks in equatorial plane from equations:
    tan(phi) = (y2*d1-y1*d2) / (x1*d2-x2*d1)
    tan(beta)= x*sin(phi) + y*cos(phi) / d
    where x=px/R, y=py/R, d=sqrt(1+x*x+y*y) - 1, px,py-peak coordinates and R-sample to detector distance in the same units as px,py.

Significant code changes

  • sign of angle beta is changed for saving in the list pf peaks.
  • fit is considered not-completed if covariance matrix contains inf or fit returns non-changed input parameters for angles.

Fit results

Consumed time (sec) : 29.693
Events processed    : 9735
Events fitted       : 3517

EQU

Pre-selected event peak statistics

Angles evaluated for 2-peak events

Angles from n-peak fit

Fit quality

Event selection based on fit quality

fit probability > 0.1
Error on phi < 2 deg
Error on beta < 10 deg
|dphi| < 10 deg
|dphi| < 20 deg

Difference between n-peak fit and 2-peak evaluation for 2-peak in EQU events

Small difference between two algorithms of angles evaluation confirms that

  • signs are selected correctly,
  • complicated n-peak fit algorithm produces the same results as simple 2-peak formulas.

 

___________________________________________________________________________

ARC (is not used for now)

Angles from the fit for 2-peak

ARC fit quality

 

 

Difference between fit in ARC (2-peak) and EQU (2-4-peak) regions

Again: not quite sure in formalism for ARC region

 

 

 

 

 

  • No labels