| T O P I C R E V I E W |
| uzi |
Posted - 06/20/2001 : 7:28:05 PM
Hi all,
I've been trying to do a non-linear curve fit with USER-DEFINED form EQUATIONS as follows:
h = (6.626E-34) / (2*3.141592)
w = x / h
Er1 = E00 - (f0*wp*wp) / (w*w + G*G) - (f1*wp*wp*(w*w - w01*w01)) / ((w*w - w01*w01)*(w*w - w01*w01) + w*w*G1*G1) - (f2*wp*wp*(w*w - w02*w02)) / ((w*w - w02*w02)*(w*w - w02*w02) + w*w*G2*G2)
Er2 = (f0*wp*wp*G) / (w*(w*w + G*G)) + (f1*wp*wp*G1*w) / ((w*w - w01*w01)*(w*w - w01*w01) + w*w*G1*G1) + (f2*wp*wp*G2*w) / ((w*w - w02*w02)*(w*w - w02*w02) +w*w*G2*G2)
n = sqrt((sqrt(Er1*Er1 + Er2*Er2) + Er1)/2)
k = sqrt((sqrt(Er1*Er1 + Er2*Er2) - Er1)/2)
y = ((n - 1)*(n - 1) + k*k) / ((n + 1)*(n + 1) + k*k)
where x is independent and y is dependent var. E00, f0, f1, f2, wp, w01, w02, G, G1, G2 are parameters. I keep getting Error 28036 or 28037. So far I tried following:
1) Reduce dataset (from 300 samples to 150, then 50, then 20)
2) Hardcode the parameter constraints (all parameters)
3) Set parameters significant digits (to 4 digits)
4) I managed to calculate the Chi-sqr, therefore i guess it is not up to formula
Nothing... After first step i get error 28036 or 28037 (try setting parameter f0 to fixed OR poor parameter initialization - Chi-sqr is 2967 for 300 samples, 734 for first 50 samples). Is this tooooo much? What should I do? Anyone?
And yes, somehow preview is not showing all the pluses in equations, but they are there...
Alek |
|
|
