How to take care of matrix dimensions issue when dividing in a function

I'm trying to fit a function that has a division in it. I'm having problems because when you divide a parameter by a data set, I get matrix dimension problems. This is my simple code:

function [y]=apple(x,par)

a=par(1);

b=par(2);

y=(x+b)/a;

end;

Separate editor:

x = [1 2 3 4 5]; y = (x+3)/2;

%set initial guess

par = [1 1];

lowerlim = [0 0]; upperlim = [10 10];

apar=lsqcurvefit(@apple,a,x,y,lowerlim,upperlim);

I'm trying to solve for the parameters, a & b, by fitting the equation y=(x+b)/a to the data set. I'm supposed to get a=2 and b=3. I'm having problems, because the matix dimension for the parameter is only 1x1 dimension whereas the data set is 1x5. I think this is my problem. How do I go about fixing this?

Answer 1

[Walter Roberson]

lsqcurvefit() varies the first parameter to the given function, not the second parameter.

function y = apple(x, xdata)
  a = x(1); b = x(2);
  y = (xdata+b)./ a;
end

and in your other file,

xdata = [1 2 3 4 5]; ydata = (xdata+3) ./ 2;
ab0 = [1 1];  %initial guess
lowerlim = [0 0]; upperlim = [10 10];
ab = lsqcurvefit(@apple, ab0, xdata, ydata, lowerlim, upperlim);
a = ab(1); b = ab(2);