home *** CD-ROM | disk | FTP | other *** search
- The FITLS utility allows an equation with $n$ unknown constants to
- be fitted to experimental data.
-
- For example to fit a simple least squares regression line to a set of
- points you would give FITLS the equation:
-
- a*x+b
-
- FITLS would then solve the equation to find the {\em best} values for
- the constants $a$ and $b$.
-
- FITLS can work with non linear equations, it will ask for initial
- values for the parameters so that a solution around those initial
- guesses will be found.
-
- FITLS writes out a GLE file containing commands to draw the data
- points and the equation it has fitted to them.
-
- Here is a sample FITLS session:
-
- $ fitls
- Input data file (x and y columns optional) [test.dat,1,2] ? fitls.dat
- Loading data from file, fitls.dat, xcolumn=1, ycolumn=2
- Valid operators: +, -, *, /, ^ (power)
- Valid functions:
- abs(), atn(), cos(), exp(), fix(), int()
- log(), log10(), not(), rnd(), sgn(), sin()
- sqr(), sqrt(), tan()
-
- Enter a function of 'x' using constants 'a'...'z'
- e.g. a + b*x (standard linear least squares fit)
- sin(x)*a+b
- a + b*x + c*x^2 + d*x^3
- log(a*x)+(b+x)*c+a
-
- Equation ? sin(a*x)*b+c*x^2+d
- Output file name to write gle file [fitls.gle] ?
- Precision of fit required, [1e-4] ?
- Initial value for constant a [1.0] ?
- Initial value for constant b [1.0] ?
- Initial value for constant c [1.0] ?
- Initial value for constant d [1.0] ?
- 0 evaluations, 1 1 1 1 , fit = 1355.36
- 20 evaluations, 1.97005 1 1 1 , fit = 1281.95
- 40 evaluations, 1.97005 10.228 0.151285 1 , fit = 54.7694
- 60 evaluations, 2.01053 10.228 0.151285 1.06365 , fit = 54.1771
- .
- .
- .
- 440 evaluations, -0.640525 -2.81525 0.13997 1.13871 , fit = 0.940192
- 460 evaluations, -0.638055 -2.82934 0.140971 1.10502 , fit = 0.93842
- 480 evaluations, -0.63808 -2.82357 0.140993 1.10452 , fit = 0.938389
- a = -0.638262 b = -2.81719 c = 0.140722 d = 1.11256
-
- 10 Iterations, sum of squares devided by n = 0.938389
- y = sin(-0.638262*x)*-2.81719+0.140722*x^2+1.11256
-
-