This is a question on non-linear curve fitting, not on model selection and not particularly on how to do regression in matlab.
I have 2 models for the same physical process, each a function of the vector x, and both have two coefficients, for m1, they are x0 and b and for m2 they are x0 and theta, where theta is notationally, b/x0. The models return slightly different values for x0 and theta from m2 does not general equal b/x0 from m1.
I am looking for advice on how to justify the use of one model over the other.
x = linspace(0,1,1e3);
m1 = @(x) (1 - (x./x0)) .* exp(-b .* (x/x0));
m2 = @(x) (1 - (x./x0)) .* exp(-theta .* x);