fitgaussian

fit a gaussian like curve to data

FITGAUSS is a function to fit a gaussian like curve "f" to experimental data by Marquardt-Levenberg non-linear least squares minimization. The fitting function has a form of a*exp(-((x-b)/c)^2)+d*x+e. This means the curve is build up a line and a gaussian. INPUTS: "x,y" is input data. "init" is initial guess for parameteres [a b c d e]. If is empty they will be determined automatically from input data and "w" is weight vector (default is ones(size(x))). OUTPUTS: "f" is the values of fitted function. "X" is the estimated parameters. "err" is the normalized error. "it" is number of iterations. "extreme" function from Lic. on Physics Carlos Adrián Vargas Aguilera is used in this function.

Example:
x=1:100;
a=30; b=45; c=10; d=.3; e=20;
f=a*exp(-((x-b)./c).^2)+d*x+e;
fn=f+2*randn(size(f));
[F X]=fitgauss(x,fn);
X=
29.61, 45.20, 10.20, 0.31, 19.18
plot(x,fn,'.',x,F,'r')