| Author |
 Topic  |
|
|
threedaybuspass
Canada
1 Posts |
 Posted - 09/22/2014 : 12:16:22 PM
|
Hello,
I am running OriginPro 8.5 on Windows 8 and I'm having the following problem:
I have created a user defined function for fitting stretched exponential fluorescence lifetimes y = y0 + A*exp(-(x/t)^B) and put in the bounds that 0 < B < 1, but as soon as I use this function to fit my data, the bounds are ignored and I get a B value that is above 1. Is there a fix for this?
Erin |
|
|
greg
USA
1378 Posts |
Posted - 09/22/2014 : 4:22:34 PM
|
I could not verify this behavior in 8.5.0 SR1.
I created your function and simulated some data where y=13.6, A=24, t=0.24 and B=1.025 with 5% noise.
I then fit with your function with the bounds of B set as you indicated ( 0 < B < 1 ) from a starting point of y=10, A=50, t=1 and B=1.
After fitting, my values were y0=13.8, A=24.12939, t=.23128 and B=1
So the fitter refused to push the value of B higher than 1.
Removing the Bounds gave me values of y0=13.84644, A=.2407143, t=.23188 and B=1.10257 which are closer to the simulation values and acceptable given the noise.
|
 |
|
| |
Topic |
|
|
|
Curve Fitting Boundaries