How can I use TolFun to stop iteration when resnorm less than it?

Hi,

I am using lsqcurvefit to do parameter estimation. I do not need to have a very small resnorm at the end. So if the resnorm is already less than my tolerance, it is more than enough for me.

So, e.g., I set TolFun to 1e-2:

optionsLSQ = optimset('Display','Iter','TolFun',1e-2);

[ki,resnorm] = lsqcurvefit(@funObj,ki,tdata,xdata,lb,[],optionsLSQ);

And this is what I get:

                                         Norm of      First-order 
 Iteration  Func-count     f(x)          step          optimality   CG-iterations
     0          3        0.113218                         0.311
     1          6      0.00709841       0.237501         0.0365            0
     2          9      0.00463864      0.0462264        0.00193            0
Optimization terminated: first-order optimality less than OPTIONS.TolFun,
 and no negative/zero curvature detected in trust region model.

As you can see, at iteration 1, f(x)<0.01, and yet the iteration does not stop. What should I do to make it stop right after f(x)<TolFun?

Thank you in advance.

Answer 1

[Arnaud Miege]

Have a look at Tolerances and Stopping Criteria in the documentation.

HTH,

Arnaud

[Amad]

Thank you Arnaud,

It's clear now why it is like that. My further question is that, is there some other way to stop iteration based on the value of f(x)?

Best Regards,

Apri

[Arnaud Miege]

Not that I know of, sorry.