
Criterian for selecting data points to be fitted


aperature bias = 0

must have at least 10 or more non-zero points in rpa curve

if less that 1.3 Re high voltage monitor must be turned on

maximum countr rate value must at least 5.0

must have at least 4 points

starting from end of rpa curve find 3 consecutive point of increasing value, reset end of rpa curve to here

make certain last point is 1 sigma above noise level (the points excluded in previous step), if not drop point and check new last point, continue until criteria is met, must have at least 3 points left

starting at new end of selected rpa curve stop first point greater that 80% of maximum of spin curve, if non found stop at last point less that maximum, must have at least 3 points left





change curve from count rate curve to l**2 curve


if number of points are 5 or less:

do a linear least squares fit (linfit) to the data, if the linear correlation coefficient (lcc) greater than 0.800 then points will be used, if not, data set is discarded


if number of points are greater then 5:

do a linfit to the bottom 5 and a linfit to the top 5 points, in 6 or more points do linfit to the middle 5, saving the lcc and slope for each case. if all three lccs are less than 0.800 discard data set

through a series of tests find the set of 5 with the best lcc slope combination


once set of 5 has been selected add rest of points one at a time and redo linfit, if lcc gets worse discard point otherwise keep it, do this until all points are checked

we now have the points to be used
