ACMALGS.ZIP

     The 14 FORTRAN source algorithms in this subset from the ACM collection
have proven particularly useful for exploratory data analysis by least 
squares parametric and non-parametric methods.  Procedures based on
advanced matrix decompositions and smoothing splines are featured in this
compilation.  They were selected for their numerical stability and robustness
in the presence of ill conditioned data.  They also are amenable to rapid
translation into other high level languages.

     Citations are as follows by ACM algorithm number:

     476   Six subprograms for curve fitting using splines under tension.
     525   ADAPT, Adaptive smooth curve fitting.
     526   Bivariate interpolation and smooth surface fitting for irregularly
              distributed data points.
     573   NL2SOL, an adaptive nonlinear least-squares algorithm.
              (Channel 1 users should note that Phillip H. Sherrod has)
              (implemented this algorithm in his NONLIN series, available)
              (as NONLIN15.ZIP as of 8/28/92).
     581   An improved algorithm for computing the singular value decomposition.
     600   Translation of Algorithm 507.  Procedures for quintic spline
              interpolation.
     615   The best subset of parameters in least absolute value regression.
     633   An algorithm for linear dependency analysis of multivariate data.
     634   CONST and EVAL. Routines for fitting multinomials in a least-
              squares sense.
     642   A fast procedure for calculating minimun cross-validation cubic
              smoothing splines.
     665   MACHAR, a subroutine to dynamically determine machine parameters.
     672   Generation of interpolatory quadrature rules of the highest degree
              of precision with preassigned nodes for general weight functions.
     691   Improving QUADPACK automatic integration routines.
     697   Univariate interpolation that has the accuracy of a third degree
              polynomial.
    