lsqcurvefit vs lsqnonlin - lsqnonlin wouldn't work...

Hello,

I have a nonlinear fitting question. I can use lsqcurvefit or lsqnonlin in the two methods described below, where lsqcurvefit is a LOT slower but works, yet lsqnonlin deviates at small xdata. Could anyone please take a look and see if there's anything wrong with my lsqnonlin method?

Any suggestion on improving the speed of computation is greatly appreciated too! Thanks!

XR

Goal: Fit [xdata,ydata] to function Y = Fun(X,Y) and obtain 5 parameters -> size(params) = 5,1

Method 1-lsqcurvefit (works):

[params,resnorm] = lsqcurvefit(@Fun1,params0,xdata,ydata,lb,ub,options);

where Fun1 is a function that uses fzero to find the root Y of the expression Y-Fun(X,Y) at all given X and params i.e.,

function Y = Fun1(params,xdata)

f = inline('Y-Fun(X,Y)','','','',...);

for i = 1:length(xdata)

Y(i) = fzero(f,y0,params,xdata(i));

end

end

Method 2-lsqnonlin (doesn't work...):

Redefine a new function Fun2 = Y-Fun(X,Y);

Use data = [xdata ydata] together as input

fit_fun = @(params)Fun2(params,data)

[params,resnorm] = lsqnonlin(fit_fun,params0,lb,ub,options);

[bym]

maybe a more detailed example is in order, I don't understand your goal.

Depending on the length of xdata in your first method, you might try preallocating Y to speed things up