How to fit a function with two or more dependant variables (outputs)?

I have a complex matlab function that takes x -> (y1,y2) based on several passed parameters. I would like to use curve fitting to find the optimal parameters that fit the function to data. Are there any such tools in MATLAB to accomplish this? nlinfit and the others seem to only want to work with a single independant variable column vector and a single dependant variable column vector (i.e. x -> y).

[Torsten]

Use X->Y with X=[x,x], Y=[y1,y2].

[Ryan Bisbey]

Thanks for the prompt response @Torsten. With nlinfit when the dependant input is an array not a vector you get the error:

Requires a vector second input argument.

Which results from:

elseif ~isvector(y)

error(message('stats:nlinfit:NonVectorY'));

end

in the nlinfit code. Were you suggesting I use another curve fitting function than nlinfit?

[Torsten]

What is the difference between an array and a vector ?

If you enter X and Y as independent and dependent vectors, all should work out fine.

Answer 1

[Matt J]

lsqcurvefit does not care about the dimensionality of the independent or dependant variables.

[Ryan Bisbey]

@Matt J

Just tried lsqcurve fit and it works for my function with its input/output dimensions. Thanks!

[Ryan Bisbey]

@Star Strider, your message was helpful too to confirm that the dimensionality of nlinfit output must be a vector not a matrix. (I think the message is deleted now.) Thanks!

[Star Strider]

My pleasure.

I almost always delete my un-accepted answers.

Answer 2

[M Mirrashid]

https://ch.mathworks.com/matlabcentral/fileexchange/112600-transit-search-optimization-algorithm-matlab-codes