Covariance & GLAST

Agenda
• Review of Covariance
• Application to GLAST
• Kalman Covariance
• Present Status

Bill Atwood, August, 2003 1 GLAST

 Review of Covariance
Ellipse Take a circle – scale the x & y axis:
x2 y2
b 2
 2 1
a a b
x  x cos( )  y sin( )
Rotate by :
y  y cos( )  x sin( )
Results:
cos 2 ( ) sin 2 ( ) 1 1 2 sin ( )
2
cos 2 ( )
x ( 2
2
 2
)  2 xy cos( ) sin( )( 2  2 )  y ( 2
 2
) 1
a b a b a b
Rotations mix x & y. Major & minor axis plus rotation angle  complete description.
Error Ellipse described by Covariance Matrix:

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 xx1 C xy1 
1
C  Inverse(C )   1 1 
1
and C xy  C yx1 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!

Bill Atwood, August, 2003 3 GLAST

 Review 3
Let the fun begin! To disentangle the two descriptions consider
C xy1  C xy cos( ) sin( )(b 2  a 2 )
A 1 1
 
C C
xx yy C xx  C yy a2  b2
sin( 2 )  1  r 2  where r = a/b
thus A   
2  1  r 2 

= 0 = /4 = /2 = 3/4

Also det(C) yields (with a little algebra & trig.):

a  b  det(C )
Now we’re ready to look at results from GLAST!

Bill Atwood, August, 2003 4 GLAST

 Covariance Matrix from Kalman Filter
Results shown for
Tkr1SXY
AxisAsym  
Tkr1SXX  Tkr1SYY

Binned in
cos() and log10(EMC)

Recall however that KF

gives us C in terms of the
track slopes Sx and Sy.

AxisAsym grows like

1/cos2()

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

Bill Atwood, August, 2003 5 GLAST

 Relationship between Slopes and Angles
For functions of the estimated variables the usual prescriptions is:
f 2
 2f    i2 (
variables x
) when the errors are uncorrelated. For correlated
i
f f f f
errors this becomes  f    i ( ) j  ( )Ci , j (
2
)( )
i, j xi x j x i x j

and reduces to the uncorrelated case when Ci , j   i2 i , j

The functions of interest here are:

1 Sy
cos( )  and tan( )  
1 S  S
2
x
2
y
Sx
A bit of math then shows that:

 2  cos 4 ( )cos 2 ( )C xx  2 sin( ) cos( )C xy  sin 2 ( )C yy 

and
 2 
1
tan ( )
2

sin 2
( )C xx  2 sin( ) cos( )C xy  cos 2
( )C yy 
Bill Atwood, August, 2003 6 GLAST
 Angle Errors from GLAST
 has a divergence at  . However sin() cures this.
 decreases as cos2() - while sin() decreases as cos( )
We also expect the components of the covariance matrix to increase as
1
due to the dominance of multiple scattering.
cos( )
 
Plot measured residuals in terms of Fit 's (e.g. meas MC )
 FIT
cos()

log10(EMC)

Bill Atwood, August, 2003 7 GLAST

 Angle Errors 2
What's RIGHT:
1) cos() dependence
2) Energy dependence in Multiple Scattering dominated range

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?

Bill Atwood, August, 2003 8 GLAST

 Angle Errors 3
Comparison of

Event-by-Event PSF
vs
FIT Parameter PSF

(Both Energy Compensated)

Difficult to assess
level of correlation

- probably not zero

- approximately same
factor of 2.3

Bill Atwood, August, 2003 9 GLAST

 Angle Errors: Conclusions
1) Analysis of covariance matrix gives format for modeling instrument response

2) Predictive power of Kalman Fit?

- Factor of 2.3

- May prove a good handle for CT tree determination of "Best PSF"

Bill Atwood, August, 2003 10 GLAST

 Present Analysis Status

Bill Atwood, August, 2003 11 GLAST

 Present Status 2: BGE Rejection
Events left after
Good-Energy Selection: 1904
Events left in VTX Classes: 26
Events left in 1Tkr Classes: 1878

CT BGE Rejection factors obtained:

20:1 (1Tkr)
2:1 (VTX)

NEED FACTOR OF 10X EVENTS BEFORE PROGRESS CAN BE MADE!

Bill Atwood, August, 2003 12 GLAST