Eval14 Integration


Definitions

Name Size Total Intensity  
left Nleft Sleft background left
peak Npeak Speak within peak
right Nright Sright background right
bg Nbg Sbg total background
exclude Nexclude Sexclude pixels removed from background

bpb

Background-Peak-Background integration. Originally from one dimensional profiles with a peak in the middle and background at the left and right side of the peak.

The bpb method is used in eval14 for determining the quality of the background. The available data are stored in a box with dimensions Nhor, Nver and Nrot. Now one could do a bpb integration along Nrot on all Nhor*Nver rows of pixels. And in a similar way along Nhor and Nver.
Every row can be considered as a one dimensional profile. For each of the directions we determine the Nleft, Nright and Npeak as well as Sleft, Sright and Speak.
Nbg = Nleft + Nright
Sbg = Sleft + Sright
f = Npeak/Nbg
Ipeak = Speak - f*Sbg
SigmaPoisson = Speak + sqrt(f²*Sbg)
At the end we have intensities for all 3 projections: Ih, Iv and Ir.
M1 = maximum of (Ih,Iv,Ir)
M2 = minimum of (Ih,Iv,Ir)
MA = average of (Ih,Iv,Ir)
The background quality is defined as:
Bgquality = (M1 - M2)/MA
If BgQuality exceeds the value of badbg the reflection will be flagged B to signal a suspect background. Sometimes the values of Nleft or Nright are zero. The one-dimensional profile has no head or tail. The net-intensity cannot be determined. The number of such one-dimensional profiles will be printed between brackets. If the number of missing profiles exceeds the value of badbgallowedmis the intensity of the corresponding intensity is not used in the determination of the background quality. If only one projection can be evaluated, the background quality cannot be determined. The reflection will be flagged UnknownBackground.

plane

For all pixels within the peak, the background value can be estimated using the the background plane constants A,B and C.
bg = A*hor + B*ver + C.
The sum of these estimates is subtracted from Speak, resulting in the net intensity of the reflection.

Sigma Poisson

The procedure for the plane intensity is used to determine the net intensity. What about sigma? One way is to use the formulae given in the bpb paragraph (assuming the pixel values are real counts (and this assumes the gain in the boxfile is correct)). The value thus calculated is called SigmaPoisson.

peak quality

If the sigma of the background plane equals to SigBg, the peak quality of a peak is defined as
weight = SigBg*sqrt(Npeak)
PQ = Inet/weight

Sigma Excluded

If pixels are excluded from the background, you probably should remove some pixels from the peak (you would actually see those pixels if the intensity of the peak was zero).
f = Nexclude/Nbg
Iexclude = Sexclude - f*Sbg
The average contribution to each background pixel then is:
Aexclude = Iexclude/Nbg
And the 'unknown' contribution in Inet is:
SigmaExcluded = Npeak*Aexclude.

Eval14
Eval14 commands