Hi:
Note that all the exponential functions are over-parameterized, so there may be multiple initial values for every parameters.
I did not actually fit such curve, maybe you can try this first. To fit a segmental function, you can define a function like:
if (x < x0) y = a+b*x;
else y = y0 + A1*exp((x-x0)/t1);
Then you can use some code before the function definition to calculate the x0 (or manually calculate the value and then fix this parameter in NLFit dialog). The idea is, the inflexion appears in y=0, if you calculate the 2nd derivative value of the curve. To find the inflexion, you need to:
1. Smooth the curve to reduce noise, or fit it by 2nd, 3rd polynomial function to create a smooth curve.
2. Look for the first x where y = 0. Actually, we calculate the 2nd derivative from discrete data, you may see there are no y = 0 points. So if abs(y-0) < "some value", you can treat it as an inflexion.
Hope it helps!
Thanks
Larry
OriginLab Technical Services
