Multi-curve fit with different equations and global shared parameters

Hi at all,

I want to realize a multi-curve fit of different sets of data with different equations but global shared parameters. Consider the following situation:

4 different raw data curves (e.g. signal) dependent on the same x-vector (e.g. time) describing different properties of the observed system.

data1 = (x,y1)

data2 = (x,y2)

data3 = (x,y3)

data4 = (x,y4)

Each curve can be fitted separately with 4 different equations sharing some global conserved parameters (equations are exemplary, the real consist of mich more factors and variables):

fit1 = N1 * exp(x/t) * a * exp(x/b)

fit2 = N2 * exp(x/t) * c * exp(x/d)

fit3 = N3 * exp(x/t) * (a+c) * (b-d)

fit4 = N4 * exp(x/t) * (a-c) * (b+d)

As you can see, all fits effect the others, some fit parameters are globally shared and some only appear in a single equation.

How is it possible to fit these curves simultaneously and with specified boundaries for each "subfit" to find the global best solution for all parameters. I tried nlinmultifit but sadly didn't get it run.

Thanks in advance and best regards

Markus

Answer 1

[xu]

lsqmultinonlin