lsqcurvefit constrain to real fitting parameters

Hi,

I'm trying to fit my data using the 'lsqcurvefit' command. However this returns me imaginairy fitting parameters, which I don't want. Is there any way to constrain the fitting parameters to real numbers?

The function I'm trying to fit is the following:

function F = eWLC(pp,y)

F = pp(2)*(1-1/2*sqrt(((4.1)./(y-pp(4))./pp(1)))+(y-pp(4))./pp(3));

thanks,

Rosalie

Answer 1

[Adam Parry]

I'm sorry that I don't have an answer but I am also suffering from the same problem.

Did you come up with a fix, or was it to do with the equation itself?

[Star Strider]

From a quick look, it would seem that constraining

pp(4) < y

might keep everything real, but without knowing more I can't be sure.

What are the parameter value estimates that 'lsqcurvefit' returns? Are the complex parameter estimates conjugates? How well does the resulting function fit the data?