The file FEFF.OUT started life as HPGe95.txt. Cambio converted this to data following <Egy_Eff_SE_DE>. In the process it was rounded to 1 decimal place and then divided by 100. FEFF.OUT is the data used by Cambio to generate the spline that is the efficiency used by the code.
Directory \public_html\Fittery\robfit\TemplFit\eff
The best fit is illustrated in ../../data/Welcome.htm#FittedEfficiency
The final fit form contains exp(1/(1-cE2). This is an exponential of the Pade form discussed in ..\..\..\nlfit\Welcome.doc#PadeApproximates .htm. When fitting the initial efficiency form, the error drops rapidly to about 2%, which is close but detectably off. It then converges very slowy requiring ~100000 steps to minimize completely at about 0.1% or less. I stopped iterating at an arbitrary 200 steps in this fit. The actual fit to the data did not appear to have this problem. The solution to this problem in nlfit was to use multiple precision ~ 90 digits rather than 16. The Pascal version of templfit uses extended precision ~24 digits rather than 16 which will help
.
1. checkder.for checkder.wpj Checkders.vfproj checkz the polynomial derivatives
fAEff.for qfaeff.for mfaeff.for approximating function DOUB, QUAD, MULT polysa.wpj
PolySA.vfproj See nlfit\PolySA\Welcome to
POLYSA.htm
2. PlotPoly.for PlotPoly.wpj reads fit.dir file and makes file of x,y from poly.
3. conv.for – converts efficiency file with arbitrary height to one with 0.8 height.
04/10/2009 01:15p 21,959 PolyEff.docx .htm
06/26/2009 02:39p 43,520 Robmfu.htm