"Input argument is undefined" when using nlinfit

Despite the variables all being defined in the workspace, implementation of the nlinfit function returns the error "Input argument p is undefined". The code I used follows:

fun = @(b,x,y,p) b(1).*x.*p.*((y/x).^b(2) - 1) + b(3)
b0 = [1 1 1];
mdl = nlinfit([x y p],q,fun, b0)

b0 gives the initial values for the coefficients to be determined - b1,b2,b3. fun is an anonymous function according to which the data is to be fit. x,y,p,q are all column vectors, which are visible in the workspace. Any help as to what is going on here would be greatly appreciated. Thank you very much in advance.

[Mahdi]

It looks like you don't have the variable p defined at all in your work space.

[Mahdi]

If you already have y and p defined, why not simplify the problem and make it

fun=@(b,x) b(1).*x.*p.*((y/x).^b(2) - 1) + b(3)
b0=[1 1 1];
mdl=nlinfit(x,q,fun,b0)

Answer 1

[the cyclist]

Something like this?

x = rand(100,1);
y = rand(100,1);
p = rand(100,1);
q = rand(100,1);

M = [x y p];

fun = @(b,M) b(1).*M(:,1).*M(:,3).*((M(:,2)./M(:,1)).^b(2) - 1) + b(3)
b0 = [1 1 1];
mdl = nlinfit(M,q,fun, b0)

[Gautam]

Thank you, but can you explain why please. It worked, but I see no difference except adding another difficult-to-type step. Thank you very much, in any case.

[the cyclist]

Because nlinfit is only passing one variable [the matrix I defined as M, (called "X" in the documentation)] to fun(). In the syntax you used, MATLAB has no way of knowing that the array

[x,y,p]

should be parsed into the three separate variables of your function.

[Gautam]

Oh, okay. Thank you very much.