FITEXY

       Best straight-line fit to data with errors in both coordinates

       Linear Least-squares approximation in one-dimension (y = a + b*x),
               when both x and y data have errors

       FITEXY, x, y, A, B, X_SIG= , Y_SIG= , [sigma_A_B, chi_sq, q, TOL=]

       x = array of values for independent variable.
       y = array of data values assumed to be linearly dependent on x.

       X_SIGMA = scalar or array specifying the standard deviation of x data.
       Y_SIGMA = scalar or array specifying the standard deviation of y data.

       TOLERANCE = desired accuracy of minimum & zero location, default=1.e-3.

       A_intercept = constant parameter result of linear fit,
       B_slope = slope parameter, so that:
                       ( A_intercept + B_slope * x ) approximates the y data.

       sigma_A_B = two element array giving standard deviation of
                A_intercept and B_slope parameters, respectively.
                The standard deviations are not meaningful if (i) the
                fit is poor (see parameter q), or (ii) b is so large that
                the data are consistent with a vertical (infinite b) line.
                If the data are consistent with *all* values of b, then
                sigma_A_B = [1e33,e33]
       chi_sq = resulting minimum Chi-Square of Linear fit, scalar
       q - chi-sq probability, scalar (0-1) giving the probability that
              a correct model would give a value equal or larger than the
              observed chi squared.   A small value of q indicates a poor
              fit, perhaps because the errors are underestimated.

       common fitexy, communicates the data for computation of chi-square.

       CHISQ_FITEXY()            ;Included in this file
       MINF_BRACKET, MINF_PARABOLIC, ZBRENT    ;In IDL Astronomy Library
       MOMENT(), CHISQR_PDF()     ;In standard IDL distribution

       From "Numerical Recipes" column by Press and Teukolsky:
       in "Computer in Physics",  May, 1992 Vol.6 No.3
       Also see the 2nd edition of the book "Numerical Recipes" by Press et al.

       Written, Frank Varosi NASA/GSFC  September 1992.
       Now returns q rather than 1-q   W. Landsman  December 1992
       Converted to IDL V5.0   W. Landsman   September 1997
       Use CHISQR_PDF, MOMENT instead of STDEV,CHI_SQR1 W. Landsman April 1998
       Fixed typo for initial guess of slope, this error was nearly
             always insignificant          W. Landsman   March 2000

[Home Page] [Software, Documentation] [IDL Documentation] [Quick Reference] [Feedback]