I am using the fitnlm function, but for some data I am receiving this error - Error using nlinfit>checkFunVals The function you provided as the MODELFUN input has returned Inf

Hi,

I am using fitnlm function to find a non-linea fit for sets of data. This is the function handle: F = @(b,mu)((mu.^(2*b(2)-1)).*exp(4*b(1)*(log(mu)).^2 + b(3)));

mu is an array ranging from 0.3 - 0.9, and the value of b changes per iteration (the function is within a loop) with the range of b(1): 0.01 - 0.03, b(2) = 0 - 1, b3 = 0 -100. The function works for some iterations but not for others. If you plug the values into the function manually, it is all fine.

The error I am recieving is "Error using nlinfit>checkFunVals The function you provided as the MODELFUN input has returned Inf" within "Error in nlinfit>LMfit (line 596)"

if funValCheck && ~isfinite(sse), checkFunVals(r); end

if you plug the values into the function manually, it is all fine. Warnings also appear for some iterations like:

"Warning: Some columns of the Jacobian are effectively zero at the solution, indicating that the model is insensitive to some of its parameters. That may be because those parameters are not present in the model, or otherwise do not affect the predicted values. It may also be due to numerical underflow in the model function, which can sometimes be avoided by choosing better initial parameter values, or by rescaling or recentering. Parameter estimates may be unreliable. " & "Warning: Rank deficient, rank = 0, tol = 0.000000e+00."

Please could someone help?

Thanks,

Josh

[the cyclist]

The ideal way to help us debug this would be to provide

• an example dataset that works
• an example dataset that gives the error
• minimal amount of code that demonstrates the error

You can upload a *.mat and *.m file using the paperclip icon in the INSERT section of the toolbar.

[Joshua]

This function is a part of a much-wider script with input data ranging from 1682 files with 2001 data in each file so I can't really upload the script or send the values. However when b = [-0.0151 0.4496 -17.729] it works, but when b = [-0.0117 0.5638 -17.5178] doesn't work. mu is the same for all values of b. I have used this method for different data and it has worked. But this set of data is much more dense and complex

[Joshua]

With the exact same data, the same iteration could work one time and not the other

[Torsten]

Try whether

F = @(b,mu) b(3)*mu.^b(2).*exp(b(1)*(log(mu)).^2)

instead of

F = @(b,mu)((mu.^(2*b(2)-1)).*exp(4*b(1)*(log(mu)).^2 + b(3)));

works better.

It's easy to deduce the parameters of the original equation once you can solve the modified one.