PDL::GSL::INTEG - PDL interface to numerical integration routines in GSL

This is an interface to the numerical integration package present in the GNU Scientific Library, which is an implementation of QUADPACK.

Functions are named gslinteg_{algorithm} where {algorithm} is the QUADPACK naming convention. The available functions are:

•

gslinteg_qng: Non-adaptive Gauss-Kronrod integration

•

gslinteg_qag: Adaptive integration

•

gslinteg_qags: Adaptive integration with singularities

•

gslinteg_qagp: Adaptive integration with known singular points

•

gslinteg_qagi: Adaptive integration on infinite interval of the form (-\infty,\infty)

•

gslinteg_qagiu: Adaptive integration on infinite interval of the form (a,\infty)

•

gslinteg_qagil: Adaptive integration on infinite interval of the form (-\infty,b)

•

gslinteg_qawc: Adaptive integration for Cauchy principal values

•

gslinteg_qaws: Adaptive integration for singular functions

•

gslinteg_qawo: Adaptive integration for oscillatory functions

•

gslinteg_qawf: Adaptive integration for Fourier integrals

Each algorithm computes an approximation to the integral, I, of the function f(x)w(x), where w(x) is a weight function (for general integrands w(x)=1). The user provides absolute and relative error bounds (epsabs,epsrel) which specify the following accuracy requirement:

|RESULT - I| <= max(epsabs, epsrel |I|)

The routines will fail to converge if the error bounds are too stringent, but always return the best approximation obtained up to that stage

All functions return the result, and estimate of the absolute error and an error flag (which is zero if there were no problems). You are responsible for checking for any errors, no warnings are issued unless the option {Warn => 'y'} is specified in which case the reason of failure will be printed.

Please check the GSL documentation for more information.

   use PDL;
   use PDL::GSL::INTEG;

   my $a = 1.2;
   my $b = 3.7;
   my $epsrel = 0;
   my $epsabs = 1e-6;

   # Non adaptive integration
   my ($res,$abserr,$ierr,$neval) = gslinteg_qng(\&myf,$a,$b,$epsrel,$epsabs);
   # Warnings on
   my ($res,$abserr,$ierr,$neval) = gslinteg_qng(\&myf,$a,$b,$epsrel,$epsabs,{Warn=>'y'});

   # Adaptive integration with warnings on
   my $limit = 1000;
   my $key = 5;
   my ($res,$abserr,$ierr) = gslinteg_qag(\&myf,$a,$b,$epsrel,
                                     $epsabs,$limit,$key,{Warn=>'y'});

   sub myf{
     my ($x) = @_;
     return exp(-$x**2);
   }

This function applies the Gauss-Kronrod 10-point, 21-point, 43-point and 87-point integration rules in succession until an estimate of the integral of f over ($a,$b) is achieved within the desired absolute and relative error limits, CW$epsabs and CW$epsrel. It is meant for fast integration of smooth functions. It returns an array with the result, an estimate of the absolute error, an error flag and the number of function evaluations performed.

Usage:

  ($res,$abserr,$ierr,$neval) = gslinteg_qng($function_ref,$a,$b,
                                             $epsrel,$epsabs,[{Warn => $warn}]);

Example:

   my ($res,$abserr,$ierr,$neval) = gslinteg_qng(\&f,0,1,0,1e-9);
   # with warnings on
   my ($res,$abserr,$ierr,$neval) = gslinteg_qng(\&f,0,1,0,1e-9,{Warn => 'y'});

   sub f{
     my ($x) = @_;
     return ($x**2.6)*log(1.0/$x);
   }

This function applies an integration rule adaptively until an estimate of the integral of f over ($a,$b) is achieved within the desired absolute and relative error limits, CW$epsabs and CW$epsrel. On each iteration the adaptive integration strategy bisects the interval with the largest error estimate; the maximum number of allowed subdivisions is given by the parameter CW$limit. The integration rule is determined by the value of CW$key, which has to be one of (1,2,3,4,5,6) and correspond to the 15, 21, 31, 41, 51 and 61 point Gauss-Kronrod rules respectively. It returns an array with the result, an estimate of the absolute error and an error flag.

Please check the GSL documentation for more information.

Usage:

  ($res,$abserr,$ierr) = gslinteg_qag($function_ref,$a,$b,$epsrel,
                                      $epsabs,$limit,$key,[{Warn => $warn}]);

Example:

  my ($res,$abserr,$ierr) = gslinteg_qag(\&f,0,1,0,1e-10,1000,1);
  # with warnings on
  my ($res,$abserr,$ierr) = gslinteg_qag(\&f,0,1,0,1e-10,1000,1,{Warn => 'y'});

  sub f{
     my ($x) = @_;
     return ($x**2.6)*log(1.0/$x);
   }

This function applies the Gauss-Kronrod 21-point integration rule adaptively until an estimate of the integral of f over ($a,$b) is achieved within the desired absolute and relative error limits, CW$epsabs and CW$epsrel. The algorithm is such that it accelerates the convergence of the integral in the presence of discontinuities and integrable singularities. The maximum number of allowed subdivisions done by the adaptive algorithm must be supplied in the parameter CW$limit.

Please check the GSL documentation for more information.

Usage:

  ($res,$abserr,$ierr) = gslinteg_qags($function_ref,$a,$b,$epsrel,
                                       $epsabs,$limit,[{Warn => $warn}]);

Example:

  my ($res,$abserr,$ierr) = gslinteg_qags(\&f,0,1,0,1e-10,1000);
  # with warnings on 
  ($res,$abserr,$ierr) = gslinteg_qags(\&f,0,1,0,1e-10,1000,{Warn => 'y'});

  sub f{
     my ($x) = @_;
     return ($x)*log(1.0/$x);
   }

This function applies the adaptive integration algorithm used by gslinteg_qags taking into account the location of singular points until an estimate of the integral of f over ($a,$b) is achieved within the desired absolute and relative error limits, CW$epsabs and CW$epsrel. Singular points are supplied in the piddle CW$points, whose endpoints determine the integration range. So, for example, if the function has singular points at x_1 and x_2 and the integral is desired from a to b (a < x_1 < x_2 < b), CW$points = pdl(a,x_1,x_2,b). The maximum number of allowed subdivisions done by the adaptive algorithm must be supplied in the parameter CW$limit.

Please check the GSL documentation for more information.

Usage:

  ($res,$abserr,$ierr) = gslinteg_qagp($function_ref,$points,$epsabs,
                                       $epsrel,$limit,[{Warn => $warn}])

Example:

  my $points = pdl(0,1,sqrt(2),3);
  my ($res,$abserr,$ierr) = gslinteg_qagp(\&f,$points,0,1e-3,1000);
  # with warnings on
  ($res,$abserr,$ierr) = gslinteg_qagp(\&f,$points,0,1e-3,1000,{Warn => 'y'});

  sub f{
    my ($x) = @_;
    my $x2 = $x**2;
    my $x3 = $x**3;
    return $x3 * log(abs(($x2-1.0)*($x2-2.0)));
  }

This function estimates the integral of the function f over the infinite interval (-\infty,+\infty) within the desired absolute and relative error limits, CW$epsabs and CW$epsrel. After a transformation, the algorithm of gslinteg_qags with a 15-point Gauss-Kronrod rule is used. The maximum number of allowed subdivisions done by the adaptive algorithm must be supplied in the parameter CW$limit.

Please check the GSL documentation for more information.

Usage:

  ($res,$abserr,$ierr) = gslinteg_qagi($function_ref,$epsabs,
                                       $epsrel,$limit,[{Warn => $warn}]);

Example:

  my ($res,$abserr,$ierr) = gslinteg_qagi(\&myfn,1e-7,0,1000);
  # with warnings on
  ($res,$abserr,$ierr) = gslinteg_qagi(\&myfn,1e-7,0,1000,{Warn => 'y'});

  sub myfn{    
    my ($x) = @_;
    return exp(-$x - $x*$x) ;
  }

This function estimates the integral of the function f over the infinite interval (a,+\infty) within the desired absolute and relative error limits, CW$epsabs and CW$epsrel. After a transformation, the algorithm of gslinteg_qags with a 15-point Gauss-Kronrod rule is used. The maximum number of allowed subdivisions done by the adaptive algorithm must be supplied in the parameter CW$limit.

Please check the GSL documentation for more information.

Usage:

  ($res,$abserr,$ierr) = gslinteg_qagiu($function_ref,$a,$epsabs,
                                        $epsrel,$limit,[{Warn => $warn}]);

Example:

  my $alfa = 1;
  my ($res,$abserr,$ierr) = gslinteg_qagiu(\&f,99.9,1e-7,0,1000);
  # with warnings on
  ($res,$abserr,$ierr) = gslinteg_qagiu(\&f,99.9,1e-7,0,1000,{Warn => 'y'});

  sub f{
    my ($x) = @_;
    if (($x==0) && ($alfa == 1)) {return 1;}
    if (($x==0) && ($alfa > 1)) {return 0;}
    return ($x**($alfa-1))/((1+10*$x)**2);
  }

This function estimates the integral of the function f over the infinite interval (-\infty,b) within the desired absolute and relative error limits, CW$epsabs and CW$epsrel. After a transformation, the algorithm of gslinteg_qags with a 15-point Gauss-Kronrod rule is used. The maximum number of allowed subdivisions done by the adaptive algorithm must be supplied in the parameter CW$limit.

Please check the GSL documentation for more information.

Usage:

  ($res,$abserr,$ierr) = gslinteg_qagl($function_ref,$b,$epsabs,
                                       $epsrel,$limit,[{Warn => $warn}]);

Example:

  my ($res,$abserr,$ierr) = gslinteg_qagil(\&myfn,1.0,1e-7,0,1000);
  # with warnings on
  ($res,$abserr,$ierr) = gslinteg_qagil(\&myfn,1.0,1e-7,0,1000,{Warn => 'y'});

  sub myfn{
    my ($x) = @_;
    return exp($x);
  }

This function computes the Cauchy principal value of the integral of f over (a,b), with a singularity at c, I = \int_a^b dx f(x)/(x - c). The integral is estimated within the desired absolute and relative error limits, CW$epsabs and CW$epsrel. The maximum number of allowed subdivisions done by the adaptive algorithm must be supplied in the parameter CW$limit.

Please check the GSL documentation for more information.

Usage:

  ($res,$abserr,$ierr) = gslinteg_qawc($function_ref,$a,$b,$c,$epsabs,$epsrel,$limit)

Example:

  my ($res,$abserr,$ierr) = gslinteg_qawc(\&f,-1,5,0,0,1e-3,1000);
  # with warnings on
  ($res,$abserr,$ierr) = gslinteg_qawc(\&f,-1,5,0,0,1e-3,1000,{Warn => 'y'});

  sub f{
    my ($x) = @_;
    return 1.0 / (5.0 * $x * $x * $x + 6.0) ;
  }

The algorithm in gslinteg_qaws is designed for integrands with algebraic-logarithmic singularities at the end-points of an integration region. Specifically, this function computes the integral given by I = \int_a^b dx f(x) (x-a)^alpha (b-x)^beta log^mu (x-a) log^nu (b-x). The integral is estimated within the desired absolute and relative error limits, CW$epsabs and CW$epsrel. The maximum number of allowed subdivisions done by the adaptive algorithm must be supplied in the parameter CW$limit.

Please check the GSL documentation for more information.

Usage:

  ($res,$abserr,$ierr) = 
      gslinteg_qawc($function_ref,$alpha,$beta,$mu,$nu,$a,$b,
                    $epsabs,$epsrel,$limit,[{Warn => $warn}]);

Example:

  my ($res,$abserr,$ierr) = gslinteg_qaws(\&f,0,0,1,0,0,1,0,1e-7,1000);
  # with warnings on
  ($res,$abserr,$ierr) = gslinteg_qaws(\&f,0,0,1,0,0,1,0,1e-7,1000,{Warn => 'y'});

  sub f{
    my ($x) = @_;
    if($x==0){return 0;}
    else{
      my $u = log($x);
      my $v = 1 + $u*$u;
      return 1.0/($v*$v);
    }
  }

This function uses an adaptive algorithm to compute the integral of f over (a,b) with the weight function sin(omega*x) or cos(omega*x) which of sine or cosine is used is determined by the parameter CW$opt ('cos' or 'sin'). The integral is estimated within the desired absolute and relative error limits, CW$epsabs and CW$epsrel. The maximum number of allowed subdivisions done by the adaptive algorithm must be supplied in the parameter CW$limit.

Please check the GSL documentation for more information.

Usage:

  ($res,$abserr,$ierr) = gslinteg_qawo($function_ref,$omega,$sin_or_cos,
                                $a,$b,$epsabs,$epsrel,$limit,[opt])

Example:

  my $PI = 3.14159265358979323846264338328;
  my ($res,$abserr,$ierr) = PDL::GSL::INTEG::gslinteg_qawo(\&f,10*$PI,'sin',0,1,0,1e-7,1000);
  # with warnings on
  ($res,$abserr,$ierr) = PDL::GSL::INTEG::gslinteg_qawo(\&f,10*$PI,'sin',0,1,0,1e-7,1000,{Warn => 'y'});

  sub f{
    my ($x) = @_;
    if($x==0){return 0;}
    else{ return log($x);} 
  }

This function attempts to compute a Fourier integral of the function f over the semi-infinite interval [a,+\infty). Specifically, it attempts tp compute I = \int_a^{+\infty} dx f(x)w(x), where w(x) is sin(omega*x) or cos(omega*x) which of sine or cosine is used is determined by the parameter CW$opt ('cos' or 'sin'). The integral is estimated within the desired absolute error limit CW$epsabs. The maximum number of allowed subdivisions done by the adaptive algorithm must be supplied in the parameter CW$limit.

Please check the GSL documentation for more information.

Usage:

  gslinteg_qawf($function_ref,$omega,$sin_or_cos,$a,$epsabs,$limit,[opt])

Example:

  my ($res,$abserr,$ierr) = gslinteg_qawf(\&f,$PI/2.0,'cos',0,1e-7,1000);
  # with warnings on
  ($res,$abserr,$ierr) = gslinteg_qawf(\&f,$PI/2.0,'cos',0,1e-7,1000,{Warn => 'y'});

  sub f{
    my ($x) = @_;
    if ($x == 0){return 0;}
    return 1.0/sqrt($x)    
  }

Feedback is welcome. Log bugs in the PDL bug database (the database is always linked from http://pdl.perl.org).

PDL

The GSL documentation is online at

  http://sources.redhat.com/gsl/ref/gsl-ref_toc.html

This file copyright (C) 2003 Andres Jordan <andresj@physics.rutgers.edu> All rights reserved. There is no warranty. You are allowed to redistribute this software documentation under certain conditions. For details, see the file COPYING in the PDL distribution. If this file is separated from the PDL distribution, the copyright notice should be included in the file.

The GSL integration routines were written by Brian Gough. QUADPACK was written by Piessens, Doncker-Kapenga, Uberhuber and Kahaner.

  Signature: (double a(); double b(); double epsabs();
                   double epsrel(); double [o] result(); double [o] abserr(); 
                   int [o] neval(); int [o] ierr(); int warn(); SV* funcion)

info not available

qng_meat does not process bad values. It will set the bad-value flag of all output piddles if the flag is set for any of the input piddles.

  Signature: (double a(); double b(); double epsabs();double epsrel(); int limit();
                   int key(); double [o] result(); double [o] abserr();int n();int [o] ierr();int warn();; SV* funcion)

info not available

qag_meat does not process bad values. It will set the bad-value flag of all output piddles if the flag is set for any of the input piddles.

  Signature: (double a(); double b(); double epsabs();double epsrel(); int limit();
                   double [o] result(); double [o] abserr();int n();int [o] ierr();int warn();; SV* funcion)

info not available

qags_meat does not process bad values. It will set the bad-value flag of all output piddles if the flag is set for any of the input piddles.

  Signature: (double pts(l); double epsabs();double epsrel();int limit();
                   double [o] result(); double [o] abserr();int n();int [o] ierr();int warn();; SV* funcion)

info not available

qagp_meat does not process bad values. It will set the bad-value flag of all output piddles if the flag is set for any of the input piddles.

  Signature: (double epsabs();double epsrel(); int limit();
                   double [o] result(); double [o] abserr(); int n(); int [o] ierr();int warn();; SV* funcion)

info not available

qagi_meat does not process bad values. It will set the bad-value flag of all output piddles if the flag is set for any of the input piddles.

  Signature: (double a(); double epsabs();double epsrel();int limit();
                   double [o] result(); double [o] abserr();int n();int [o] ierr();int warn();; SV* funcion)

info not available

qagiu_meat does not process bad values. It will set the bad-value flag of all output piddles if the flag is set for any of the input piddles.

  Signature: (double b(); double epsabs();double epsrel();int limit();
                   double [o] result(); double [o] abserr();int n();int [o] ierr();int warn();; SV* funcion)

info not available

qagil_meat does not process bad values. It will set the bad-value flag of all output piddles if the flag is set for any of the input piddles.

  Signature: (double a(); double b(); double c(); double epsabs();double epsrel();int limit();
                   double [o] result(); double [o] abserr();int n();int [o] ierr();int warn();; SV* funcion)

info not available

qawc_meat does not process bad values. It will set the bad-value flag of all output piddles if the flag is set for any of the input piddles.

  Signature: (double a(); double b();double epsabs();double epsrel();int limit();
                 double [o] result(); double [o] abserr();int n();
                 double alpha(); double beta(); int mu(); int nu();int [o] ierr();int warn();; SV* funcion)

info not available

qaws_meat does not process bad values. It will set the bad-value flag of all output piddles if the flag is set for any of the input piddles.

  Signature: (double a(); double b();double epsabs();double epsrel();int limit();
                 double [o] result(); double [o] abserr();int n();
                 int sincosopt(); double omega(); double L(); int nlevels();int [o] ierr();int warn();; SV* funcion)

info not available

qawo_meat does not process bad values. It will set the bad-value flag of all output piddles if the flag is set for any of the input piddles.

  Signature: (double a(); double epsabs();int limit();
                 double [o] result(); double [o] abserr();int n();
                 int sincosopt(); double omega(); int nlevels();int [o] ierr();int warn();; SV* funcion)

info not available

qawf_meat does not process bad values. It will set the bad-value flag of all output piddles if the flag is set for any of the input piddles.