Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Agenda
• Review of Covariance
• Application to GLAST
• Kalman Covariance
• Present Status
Distance between a point with an error and another point measured in ’s:
( n ) 2 r T C 1 r where r ( x x ) and
n Simply weighting the
C xx1 C xy1
1
C Inverse(C ) 1 1
1
and C xy C yx1 distance by 1/2
C yx C yy
Bill Atwood, August, 2003 2 GLAST
Review 2
Multiplying it out gives:
C 1
C 1
xy x
(n ) r C r ( x, y ) 1
2 T 1 xx
1
x 2 1
C xx 2 xyC 1
xy y 2 1
C yy
C xy C yy y
Where I take x 0 without loss of generality.
This is the equation of an ellipse! Specifically for 1 error ellipse (n = 1) we identify:
cos 2 ( ) sin 2 ( ) sin 2 ( ) cos 2 ( ) 1 1
C 1
xx 2
2
C 1
yy 2
2
C 1
xy sin( ) cos( )( 2
2)
a b a b a b
1 C yy C xy
1
and C C where det( C ) (C C
xx yy C 2
xy )
det(C ) xy C xx
C xy2
And the correlation coefficient is defined as: r2
C xxC yy
Summary: The inverse of the Covariance Matrix describes an ellipse where
the major and minor axis and the rotation angle map directly onto
its components!
a b det(C )
Now we’re ready to look at results from GLAST!
Binned in
cos() and log10(EMC)
Peak amplitude ~ .4
1 (1 r 2 ) 1 2 A(max) a 1 2 .4 1 1
A(max) r2 r
2 (1 r 2 ) 1 2 A(max) b 1 2 .4 9 3
log10(EMC)
What's WRONG:
1) Overall normalization of estimated errors (FIT)
- off by a factor of ~ 2.3!!!
2) Energy dependence as measurement errors begin to dominate
- discrepancy goes away(?)
Both of these correlate with with the fact that the fitted 2's are
much larger then 1 at low energy (expected?).
How well does the Kalman Fit PSF model the event to event PSF?
Event-by-Event PSF
vs
FIT Parameter PSF
Difficult to assess
level of correlation
- approximately same
factor of 2.3
- Factor of 2.3