File Coverage

blib/lib/PDL/NDBin/Actions_PP.pm

Criterion	Covered	Total	%
statement	9	9	100.0
branch			n/a
condition			n/a
subroutine	4	4	100.0
pod			n/a
total	13	13	100.0

#
# GENERATED WITH PDL::PP! Don't modify!
#
package PDL::NDBin::Actions_PP;

@EXPORT_OK  = qw( PDL::PP _flatten_into PDL::PP _flatten_into_grid PDL::PP _setnulltobad PDL::PP _icount_loop PDL::PP _imax_loop PDL::PP _imin_loop PDL::PP _isum_loop PDL::PP _iavg_loop PDL::PP _istddev_loop PDL::PP _istddev_post );
%EXPORT_TAGS = (Func=>[@EXPORT_OK]);

use PDL::Core;
use PDL::Exporter;
use DynaLoader;



   $PDL::NDBin::Actions_PP::VERSION = 0.019;
   @ISA    = ( 'PDL::Exporter','DynaLoader' );
   push @PDL::Core::PP, __PACKAGE__;
   bootstrap PDL::NDBin::Actions_PP $VERSION;


BEGIN {

# ABSTRACT: XS actions for PDL::NDBin

=head1 NAME

PDL::NDBin::Actions_PP - XS actions for PDL::NDBin

=head1 DESCRIPTION

This module contains the implementation of the core loop of the action classes
in the PDL::NDBin::Action namespace. The subroutines are not intended to be
called by the user.

=cut

# silence Test::Pod::Coverage warning
=for Pod::Coverage set_boundscheck set_debugging

=cut



}





=head1 FUNCTIONS



=cut






=head2 _flatten_into

=for sig

  Signature: (in(m); indx b(m); indx [o] idx(m); double step; double min; PDL_Indx n)

=for ref

Bin a series of values and flatten into an existing list of indices.

=for bad

_flatten_into processes bad values.
It will set the bad-value flag of all output piddles if the flag is set for any of the input piddles.


=cut






*_flatten_into = \&PDL::_flatten_into;





=head2 _flatten_into_grid

=for sig

  Signature: (in(); indx b(); double edge(n); indx [o] idx())

=for ref

Bin a series of values on a specified grid and flatten into an existing list of indices.

=for bad

_flatten_into_grid processes bad values.
It will set the bad-value flag of all output piddles if the flag is set for any of the input piddles.


=cut






*_flatten_into_grid = \&PDL::_flatten_into_grid;





=head2 _setnulltobad

=for sig

  Signature: (in(); indx count(); [o] out())

=for ref

Set empty bins to the bad value.

=for bad

_setnulltobad processes bad values.
The state of the bad-value flag of the output piddles is unknown.


=cut






*_setnulltobad = \&PDL::_setnulltobad;





=head2 _icount_loop

=for sig

  Signature: (in(n); indx idx(n); indx [o] out(m); PDL_Indx msize => m)

Count the number of elements in each bin.

This function returns a piddle of type I if you have a 64-bit-capable
PDL, otherwise it returns a piddle of type I.

=for bad

_icount_loop processes bad values.
It will set the bad-value flag of all output piddles if the flag is set for any of the input piddles.


=cut






*_icount_loop = \&PDL::_icount_loop;





=head2 _imax_loop

=for sig

  Signature: (in(n); indx idx(n); [o] out(m); PDL_Indx msize => m)

=for ref

Find the maximum in each bin.

=for bad

_imax_loop processes bad values.
It will set the bad-value flag of all output piddles if the flag is set for any of the input piddles.


=cut






*_imax_loop = \&PDL::_imax_loop;





=head2 _imin_loop

=for sig

  Signature: (in(n); indx idx(n); [o] out(m); PDL_Indx msize => m)

=for ref

Find the minimum in each bin.

=for bad

_imin_loop processes bad values.
It will set the bad-value flag of all output piddles if the flag is set for any of the input piddles.


=cut






*_imin_loop = \&PDL::_imin_loop;





=head2 _isum_loop

=for sig

  Signature: (in(n); indx idx(n); int+ [o] out(m); indx [t] count(m); PDL_Indx msize => m)

Sum the elements in each bin.

This function returns a piddle of type I or higher, to reduce the risk of
overflow when collecting sums.

=for bad

_isum_loop processes bad values.
It will set the bad-value flag of all output piddles if the flag is set for any of the input piddles.


=cut






*_isum_loop = \&PDL::_isum_loop;





=head2 _iavg_loop

=for sig

  Signature: (in(n); indx idx(n); double [o] out(m); indx [t] count(m); PDL_Indx msize => m)

Compute the average of the elements in each bin.

Credit for the algorithm goes to
L on commandlinefu.com|http://www.commandlinefu.com/commands/view/3437/compute-running-average-for-a-column-of-numbers>:

	awk '{ avg += ($1 - avg) / NR } END { print avg }'

This is a wonderful solution solving many of the problems with more naive
implementations:

=over 4

=item 1.

It's numerically well-behaved: out() is always of the order of magnitude of the
values themselves, unlike the sum of the values, which grows very large as the
number of elements grows large.

=item 2.

The subtraction in() - out() guarantees that the computation will be done in
the correct type (i.e., I instead of the type of the input).

=item 3.

Requires only one temporary.

=item 4.

Requires only one pass over the data.

=back

I used to give the output array type I, but that creates more problems
than it solves. So now, averages are always computed in type I. Besides
being the default type in PDL and the 'natural' floating-point type in C, it
also makes the implementation easier.

=for bad

_iavg_loop processes bad values.
It will set the bad-value flag of all output piddles if the flag is set for any of the input piddles.


=cut






*_iavg_loop = \&PDL::_iavg_loop;





=head2 _istddev_loop

=for sig

  Signature: (in(n); indx idx(n); double [o] out(m); indx [t] count(m); double [t] avg(m); PDL_Indx msize => m)

Compute the standard deviation of the elements in each bin.

Note that we compute the sample standard deviation, I an estimate of the
population standard deviation (which differs by a factor).

Credit for the algorithm goes to
L on commandlinefu.com|http://www.commandlinefu.com/commands/view/3442/display-the-standard-deviation-of-a-column-of-numbers-with-awk>;

	awk '{ delta = $1 - avg; avg += delta / NR; mean2 += delta * ($1 - avg) } END { print sqrt(mean2 / NR) }'

This is a wonderful solution solving many of the problems with more naive
implementations:

=over 4

=item 1.

It's numerically well-behaved.

=item 2.

The subtractions guarantee that the computations will be done in the correct
type (i.e., I instead of the type of the input).

=item 3.

Requires only two temporaries (!).

=item 4.

Requires only one pass over the data.

=back

I used to give the output array type I, but that creates more problems
than it solves. So now, standard deviations are always computed in type
I. Besides being the default type in PDL and the 'natural'
floating-point type in C, it also makes the implementation easier.

=for bad

_istddev_loop processes bad values.
It will set the bad-value flag of all output piddles if the flag is set for any of the input piddles.


=cut






*_istddev_loop = \&PDL::_istddev_loop;





=head2 _istddev_post

=for sig

  Signature: (in(); indx count(); [o] out())

=for ref

Finalization for _istddev_loop().

=for bad

_istddev_post processes bad values.
The state of the bad-value flag of the output piddles is unknown.


=cut






*_istddev_post = \&PDL::_istddev_post;




=head1 AUTHOR

Edward Baudrez

=head1 COPYRIGHT AND LICENSE

This software is copyright (c) 2014 by Edward Baudrez.

This is free software; you can redistribute it and/or modify it under
the same terms as the Perl 5 programming language system itself.

=cut



;



# Exit with OK status

1;