Apply boundary condition (upper and lower limit) to parameters during the fitting using the function nlinfit?

Does anyone knows how to apply boundary condition to the parameters during the fitting using the function nlinfit?

I know that this could be done using the function fit, but i need to use the function nlinfit. (in that regards see the post in the link attached)

Thank you very much in advance for the help.

http://mathworks.com/matlabcentral/answers/63043-fit-model-with-3-independent-variables-and-many-parameters-to-data#answer_74613

Answer 1

[Parag]

Hi @Manuel,

The ‘nlinfit’ in MATLAB does not support parameter bounds directly . However, if you are required to use ‘nlinfit’, you can use this workaround to enforce parameter bounds:

Transform the parameters so that they always stay within bounds, even though ‘nlinfit’ optimizes over unbounded parameters.

Example: Suppose you want to constrain F(1) between 0 and 1. You can optimize over a new parameter theta, and define:

F(1) = 1 ./ (1 + exp(-theta));  % logistic transformation: bounded between 0 and 1 as exponent cannot be negative 

so denominator will always be greater than 1 and in limiting case theta approaching to negative infinity this expression will approach to zero.

Hope it helps!

Answer 2

[Torsten]

If you restrict parameters by imposing upper and/or lower bounds, a statistical analysis of the results like that offered by "nlinfit" is no longer possible. That's why you cannot set upper and lower bounds when you use this tool.