Skip to content
Prev 383796 / 398502 Next

unstable results of nlxb fit

Dear all.

Thank you for your answers. I will try Duncan's approach (if I could manage 
it).

The issue is that first part of my data (actually temperature) up to certain 
time approximately follow one exponential. After that, another process 
prevails and the temperature increase starts to be "explosive". That is why I 
used these two exponentials. As I have many experiments I wanted to perform 
the fit programmatically.

Which leads me to the approach that in each cycle I perform a plot which I 
visually inspect. If I consider the fit satisfactory I keep results. If not, I 
perform the fit with different starting values until it is OK. I am aware that 
it is not optimal but should be easiest.

Thank you again.

Best regards
Petr