NMS is a collection of Software from the book "Numerical Methods
and Software" by D. Kahaner, C. Moler, and S. Nash,  Prentice Hall, 
Englewood Cliffs, NJ, 1989.  All are portable Fortran subroutines.
A brief description of the user-callable routines follows.

 BESI0  Computes the hyperbolic Bessel function of the first kind
        of order zero. Double precision version is DBESI0.
  BESJ  Computes an N member sequence of J Bessel functions
        J/SUB(ALPHA+K-1)/(X), K=1,..,N for non-negative ALPHA and X.
        Double precision version is DBESJ.
CFFT2D  Two dimensional complex fast Fourier transform. Double precision 
        version is DCFT2D.
 CFFTB  Unnormalized inverse of CFFTF. Double precision version is 
        DCFFTB.
 CFFTF  Forward transform of a complex, periodic sequence. Double 
        precision version is DCFFTF.
   ERF  Computes the error function. Double precision version is DERF.
  ERFC  Computes the complementary error function. Double precision 
        version is DERFC.
EZFFTB  A simplified real, periodic, backward transform. Double 
        precision version is DEZFTB.
EZFFTF  A simplified real, periodic, forward transform. Double 
        precision version is DEZFTF.
  FMIN  Returns an approximation to the point where F attains a minimum 
        on the interval (AX,BX). Double precision version is DFMIN.
 FZERO  Searches for a zero of a function F(X) in a given
        interval (B,C). It is designed primarily for problems
        where F(B) and F(C) have opposite signs. Double precision 
        version is DFZERO.
 GAMMA  Computes the Gamma function. Double precision version is DGAMMA.
 PCHEV  Evaluates the function and first derivative of a piecewise
        cubic Hermite or spline function at an array of points XVAL,
        easy to use. Double precision version is DPCHEV.
 PCHEZ  Easy to use spline or cubic Hermite interpolation. Double 
        precision version is DPCHEZ.
 PCHQA  Evaluates the definite integral of a piecewise cubic Hermite
        or spline function over an arbitrary interval, easy to use. 
        Double precision version is DPCHQA.
  Q1DA  Approximates one dimensional integrals of user defined
        functions, easy to use. Double precision version is DQ1DA.
  QAGI  Calculates an approximation result to a given
        INTEGRAL I = Integral of F over (BOUND,+INFINITY)
        or I = Integral of F over (-INFINITY,BOUND)
        or I = Integral of F over (-INFINITY,+INFINITY)
        Hopefully satisfying following claim for accuracy
        ABS(I-RESULT).LE.MAX(EPSABS,EPSREL*ABS(I)).
        Double precision version is DQAGI.
  QK15  To compute I = Integral of F over (A,B), with error estimate
        and J = integral of ABS(F) over (A,B). Double precision 
        version is DQK15.
  RNOR  Generates normal random numbers, with mean zero and
        unit standard deviation, often denoted N(0,1).
SDRIV2  Solves N ordinary differential equations of the form 
        dY(I)/dT = F(Y(I),T), given the initial conditions Y(I) = YI. 
        The program has options to allow the solution of both stiff 
        and non-stiff differential equations. Double precision version 
        is DDRIV2.
 SGEFS  Solves a GENERAL real NXN system of linear equations.
        Double precision version is DGEFS.
 SNSQE  The easy-to-use version of SNSQ which finds a zero
        of a system of N nonlinear functions in N variables by a
        modification of Powell hybrid method. This code is the
        combination of the MINPACK codes(Argonne) HYBRD1 and HYBRJ1.
        Double precision version is DNSQE.
 SQRLS  Solves an overdetermined, underdetermined or singular
        system of linear equations in least square sense.  The
        solution is obtained using a QR factorization of the
        M  by  N  coefficient matrix  A. Double precision version is 
        DQRLS.
 SSVDC  Perform the singular value decomposition of a real NXP
        matrix. Double precision version is DSVDC.
UNCMIN  Minimizes a smooth nonlinear function of n variables.
        A subroutine that computes the function value at any point
        must be supplied. Derivative values are not required.
        Double precision version is UNCMND.
   UNI  Generates real uniform random numbers on [0,1).