Hej Bjørn, Christian ...
If you considder a weighted average a fit to a constant then you should
take as weight factors 1/sqr(sigma) assuming you have uncertainties
which you trust.
For this to be reasonable the reduced Chi-squared calculated for the set
of Vertexes using these sigmas should be between 0.1 and 2, i.e. none of
the values should be more than a few of its own sigmas away from the
average.
So a ZDC vertex should not destroy resolution if it is less than a few of
its own sigmas off.
Example
If a = 100 +-1 and b = 90 +- 10
then wa = 0.99 and wb = 0.01
Average = 100*0.99 + 90*0.01 = 99.9
SigmaAvg = sqrt(sqr(0.99*1) + sqr(0.01*10))
= sqrt(0.98 + 0.01) = 0.995
Thus a ZDC vertex does not say much unless it is way off.
Anders
=========================================================
Anders Holm email: aholm@nbi.dk
Niels Bohr institute phone: (45) 35 32 52 13
University of Copenhagen fax: (45) 35 32 50 16
Blegdamsvej 17
DK-2100 Copenhagen
Denmark
========================================================
This archive was generated by hypermail 2b29 : Thu Nov 09 2000 - 07:56:01 EST