next up previous
Next: A.1.7 Cut Steps Up: A.1 Basic Fitting Algorithm Previous: A.1.5 Error Propagation

A.1.6 Pull Functions

Define the pull functions

These quantities should be normally distributed with means of zero and widths of one.

For the moment, ignore the y's. Then, since = - and = - , it follows that

Thus,

or

But,

Distinguishing again between well-measured and poorly measured variables,

Thus,

so the pull is

 

Similarly, for poorly measured variables (if the off-diagonal elements are small),

so

 



Scott Snyder Fri May 19 19:19:46 CDT 1995