how can I extract a fit parameter, that depends on the independent variable, as a vector of that variable? Thanks a lot!

 y = ((coeff(1)./x)+(coeff(2)^2)./(x.^2))*(fitParameter); % but fitParameter also depends on vector x

[darova]

What is fitParameter? Are you using fit() function?

[dpb]

Seems peculiar to square coeff(2), also.

We need the code that generated all the variables and rationale behind what you're trying to do. Makes no sense by itself as is.

[Matteo Masto]

yes I'm using the fit() function and fitParameter is the free parameter that the fit has to find.

coeff(1) and coeff(2) are just numbers, they are not important.. could be any numbers. The function there is the one that I have to fit with some experimental data that also depend on "x". The problem is that the fit function finds me a really bad parameter and so I thought that, since in principle also that fitParameter depends on x, a way in which not only a number is returned but a vector of "fitParameter" with different values for each x value could have improved the fit..

what do you think? Could be? Thanks!!

[darova]

Please attach the whole code. Maybe you are using bad fit function

[Matteo Masto]

% data value
X = [-7.2460   -7.3355   -7.4320   -7.5752   -7.6996   -7.8780   -8.1560   -8.3517   -8.6170   -8.8883 -9.1808   -9.6629  -10.5808  -13.6249  -21.4783] ;
Y = [ -14.5516  -12.5218  -15.2762  -13.5768  -11.7304  -13.1678  -13.5329  -13.7761  -12.5162  -10.0737 -12.3137  -12.0351   -9.6663   -9.0393   -4.9973];
% function I have to fit those data to
y = log((-0.0143*(10^-4)*4*(X.^-1)+((0.6749*10^-4)^2)*(X.^-2))*(fitParameter));
% for the fit I used the curve fitting app 

[darova]

What about tihs?

p = polyfit(X,Y,1);
Y1 = polyval(p,X);
plot(X,Y,'.r',X,Y1)
legend('data','fit')

[Matteo Masto]

no, sorry I have to use that precise function. But sorry again I found that the problem was another one and now I solved. Thanks again!!

Answer 1

[Image Analyst]

Try this:

fitParameter = SomeFunctionOfX(x); % Returns a vector the same length as x.
term1 = coeff(1) ./ x;
term2 = coeff(2) ^ 2 ./ (x. ^ 2);
% Compute formula.  Note: term1, term2, and fitParameter all depend on vector x.
y = (term1 + term2) .* fitParameter; 

You, of course, have to define SomeFunctionOfX() since we don't know what it is.

[Matteo Masto]

thanks, but the problem is that the parameter depends on x in a way that I don't know a priori, I don't know which is that SomeFunctionOf(x)

[Image Analyst]

What made fitParameter? Is it a scalar (one value for all x) or vector with one value for every value of x in hte x vector?