Boost-Commit :

Date view	Thread view	Subject view	Author view

Subject: [Boost-commit] svn:boost r74032 - in sandbox/maps/libs: . maps maps/doc maps/doc/html maps/doc/html/maps
From: bjs3141_at_[hidden]
Date: 2011-08-23 16:21:45

Next message: joel_at_[hidden]: "[Boost-commit] svn:boost r74033 - trunk/libs/spirit/test/qi"
Previous message: bjs3141_at_[hidden]: "[Boost-commit] svn:boost r74031 - in sandbox/maps/boost: . maps maps/generic maps/math maps/support"

Author: brian.smith
Date: 2011-08-23 16:21:41 EDT (Tue, 23 Aug 2011)
New Revision: 74032
URL: http://svn.boost.org/trac/boost/changeset/74032

Log:
import

Added:
   sandbox/maps/libs/
   sandbox/maps/libs/maps/
   sandbox/maps/libs/maps/doc/
   sandbox/maps/libs/maps/doc/Jamfile.v2 (contents, props changed)
   sandbox/maps/libs/maps/doc/html/
   sandbox/maps/libs/maps/doc/html/index.html (contents, props changed)
   sandbox/maps/libs/maps/doc/html/maps/
   sandbox/maps/libs/maps/doc/html/maps/array.html (contents, props changed)
   sandbox/maps/libs/maps/doc/html/maps/array_.html (contents, props changed)
   sandbox/maps/libs/maps/doc/html/maps/concepts.html (contents, props changed)
   sandbox/maps/libs/maps/doc/html/maps/pointer.html (contents, props changed)
   sandbox/maps/libs/maps/doc/html/maps/reference.html (contents, props changed)
   sandbox/maps/libs/maps/doc/html/maps/tutorial.html (contents, props changed)
   sandbox/maps/libs/maps/doc/html/maps/view.html (contents, props changed)
   sandbox/maps/libs/maps/doc/html/maps/view_.html (contents, props changed)
   sandbox/maps/libs/maps/doc/maps.qbk (contents, props changed)

Added: sandbox/maps/libs/maps/doc/Jamfile.v2
==============================================================================
--- (empty file)
+++ sandbox/maps/libs/maps/doc/Jamfile.v2 2011-08-23 16:21:41 EDT (Tue, 23 Aug 2011)
@@ -0,0 +1,32 @@
+
+# Copyright John Maddock 2005. Use, modification, and distribution are
+# subject to the Boost Software License, Version 1.0. (See accompanying
+# file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
+
+using quickbook ;
+
+xml maps : maps.qbk ;
+boostbook standalone
+ :
+ maps
+ :
+ # Path for links to Boost:
+ <xsl:param>boost.root=../../../..
+
+ # Some general style settings:
+ <xsl:param>table.footnote.number.format=1
+ <xsl:param>footnote.number.format=1
+
+ # HTML options first:
+ # Use graphics not text for navigation:
+ <xsl:param>navig.graphics=1
+ # How far down we chunk nested sections, basically all of them:
+ <xsl:param>chunk.section.depth=1
+ # Don't put the first section on the same page as the TOC:
+ <xsl:param>chunk.first.sections=0
+ # How far down sections get TOC's
+ <xsl:param>toc.section.depth=1
+ ;
+
+
+

Added: sandbox/maps/libs/maps/doc/html/index.html
==============================================================================
--- (empty file)
+++ sandbox/maps/libs/maps/doc/html/index.html 2011-08-23 16:21:41 EDT (Tue, 23 Aug 2011)
@@ -0,0 +1,77 @@
+<html>
+<head>
+<meta http-equiv="Content-Type" content="text/html; charset=US-ASCII">
+<title>Boost.Maps</title>
+<link rel="stylesheet" href="../../../../doc/src/boostbook.css" type="text/css">
+<meta name="generator" content="DocBook XSL Stylesheets V1.76.1">
+<link rel="home" href="index.html" title="Boost.Maps">
+<link rel="next" href="maps/concepts.html" title="Concepts">
+</head>
+<body bgcolor="white" text="black" link="#0000FF" vlink="#840084" alink="#0000FF">
+<table cellpadding="2" width="100%"><tr>
+<td valign="top"><img alt="Boost C++ Libraries" width="277" height="86" src="../../../../boost.png"></td>
+<td align="center">Home</td>
+<td align="center">Libraries</td>
+<td align="center">People</td>
+<td align="center">FAQ</td>
+<td align="center">More</td>
+</tr></table>
+<hr>
+<div class="spirit-nav"><a accesskey="n" href="maps/concepts.html"><img src="../../../../doc/src/images/next.png" alt="Next"></a></div>
+<div class="book">
+<div class="titlepage">
+<div>
+<div><h1 class="title">
+<a name="maps"></a>Boost.Maps</h1></div>
+<div><div class="authorgroup"><div class="author"><h3 class="author">
+Brian Smith
+</h3></div></div></div>
+<div>Copyright © 2010 -2011 Brian Smith</div>
+<div><div class="legalnotice">
+<a name="id752289"></a>
+ Distributed under the Boost Software License, Version 1.0. (See accompanying
+ file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
+ 
+</div></div>
+</div>
+<hr>
+</div>
+<div class="section">
+<div class="titlepage"><div><div><h2 class="title" style="clear: both">
+<a name="maps.introduction"></a><a class="link" href="index.html#maps.introduction" title="Introduction">Introduction</a>
+</h2></div></div></div>
+
+ The Boost maps library provides sequence containers for statically and dynamically
+ allocated multi-dimensional arrays. The implementations allow standard data
+ element types, data element initialization, subject to conditions imposed by
+ the data element type, and the choice of allocator used for memory allocation,
+ element construction, etc., where appropriate. The library also includes scalar
+ and fixed-size vector and matrix class templates with a DSEL implementation
+ of operators via expression templates.
+ 
+
+ The <a class="link" href="maps/concepts.html" title="Concepts">Concepts</a> section provides an overview
+ of the library as well as a discussion of the supplementary structures that
+ interoperate with the main classes in the library.
+ 
+<div class="itemizedlist"><ul class="itemizedlist" type="disc">
+<li class="listitem">
+ <a class="link" href="maps/concepts.html" title="Concepts">Concepts</a>
+ </li>
+<li class="listitem">
+ <a class="link" href="maps/tutorial.html" title="Tutorial">Tutorial</a>
+ </li>
+<li class="listitem">
+ <a class="link" href="maps/reference.html" title="Reference">Reference</a>
+ </li>
+</ul></div>
+</div>
+</div>
+<table xmlns:rev="http://www.cs.rpi.edu/~gregod/boost/tools/doc/revision" width="100%"><tr>
+<td align="left">Last revised: June 11, 2011 at 13:25:38 GMT</td>
+<td align="right"><div class="copyright-footer"></div></td>
+</tr></table>
+<hr>
+<div class="spirit-nav"><a accesskey="n" href="maps/concepts.html"><img src="../../../../doc/src/images/next.png" alt="Next"></a></div>
+</body>
+</html>

Added: sandbox/maps/libs/maps/doc/html/maps/array.html
==============================================================================
--- (empty file)
+++ sandbox/maps/libs/maps/doc/html/maps/array.html 2011-08-23 16:21:41 EDT (Tue, 23 Aug 2011)
@@ -0,0 +1,41 @@
+<html>
+<head>
+<meta http-equiv="Content-Type" content="text/html; charset=US-ASCII">
+<title>array</title>
+<link rel="stylesheet" href="../../../../../doc/src/boostbook.css" type="text/css">
+<meta name="generator" content="DocBook XSL Stylesheets V1.76.1">
+<link rel="home" href="../index.html" title="Boost.Maps">
+<link rel="up" href="../index.html" title="Boost.Maps">
+<link rel="prev" href="reference.html" title="Reference">
+<link rel="next" href="array_.html" title="array_">
+</head>
+<body bgcolor="white" text="black" link="#0000FF" vlink="#840084" alink="#0000FF">
+<table cellpadding="2" width="100%"><tr>
+<td valign="top"><img alt="Boost C++ Libraries" width="277" height="86" src="../../../../../boost.png"></td>
+<td align="center">Home</td>
+<td align="center">Libraries</td>
+<td align="center">People</td>
+<td align="center">FAQ</td>
+<td align="center">More</td>
+</tr></table>
+<hr>
+<div class="spirit-nav">
+<a accesskey="p" href="reference.html"><img src="../../../../../doc/src/images/prev.png" alt="Prev"></a><a accesskey="u" href="../index.html"><img src="../../../../../doc/src/images/up.png" alt="Up"></a><a accesskey="h" href="../index.html"><img src="../../../../../doc/src/images/home.png" alt="Home"></a><a accesskey="n" href="array_.html"><img src="../../../../../doc/src/images/next.png" alt="Next"></a>
+</div>
+<div class="section"><div class="titlepage"><div><div><h2 class="title" style="clear: both">
+<a name="maps.array"></a><a class="link" href="array.html" title="array">array</a>
+</h2></div></div></div></div>
+<table xmlns:rev="http://www.cs.rpi.edu/~gregod/boost/tools/doc/revision" width="100%"><tr>
+<td align="left"></td>
+<td align="right"><div class="copyright-footer">Copyright © 2010 -2011 Brian Smith
+ Distributed under the Boost Software License, Version 1.0. (See accompanying
+ file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
+ 
+</div></td>
+</tr></table>
+<hr>
+<div class="spirit-nav">
+<a accesskey="p" href="reference.html"><img src="../../../../../doc/src/images/prev.png" alt="Prev"></a><a accesskey="u" href="../index.html"><img src="../../../../../doc/src/images/up.png" alt="Up"></a><a accesskey="h" href="../index.html"><img src="../../../../../doc/src/images/home.png" alt="Home"></a><a accesskey="n" href="array_.html"><img src="../../../../../doc/src/images/next.png" alt="Next"></a>
+</div>
+</body>
+</html>

Added: sandbox/maps/libs/maps/doc/html/maps/array_.html
==============================================================================
--- (empty file)
+++ sandbox/maps/libs/maps/doc/html/maps/array_.html 2011-08-23 16:21:41 EDT (Tue, 23 Aug 2011)
@@ -0,0 +1,41 @@
+<html>
+<head>
+<meta http-equiv="Content-Type" content="text/html; charset=US-ASCII">
+<title>array_</title>
+<link rel="stylesheet" href="../../../../../doc/src/boostbook.css" type="text/css">
+<meta name="generator" content="DocBook XSL Stylesheets V1.76.1">
+<link rel="home" href="../index.html" title="Boost.Maps">
+<link rel="up" href="../index.html" title="Boost.Maps">
+<link rel="prev" href="array.html" title="array">
+<link rel="next" href="pointer.html" title="pointer">
+</head>
+<body bgcolor="white" text="black" link="#0000FF" vlink="#840084" alink="#0000FF">
+<table cellpadding="2" width="100%"><tr>
+<td valign="top"><img alt="Boost C++ Libraries" width="277" height="86" src="../../../../../boost.png"></td>
+<td align="center">Home</td>
+<td align="center">Libraries</td>
+<td align="center">People</td>
+<td align="center">FAQ</td>
+<td align="center">More</td>
+</tr></table>
+<hr>
+<div class="spirit-nav">
+<a accesskey="p" href="array.html"><img src="../../../../../doc/src/images/prev.png" alt="Prev"></a><a accesskey="u" href="../index.html"><img src="../../../../../doc/src/images/up.png" alt="Up"></a><a accesskey="h" href="../index.html"><img src="../../../../../doc/src/images/home.png" alt="Home"></a><a accesskey="n" href="pointer.html"><img src="../../../../../doc/src/images/next.png" alt="Next"></a>
+</div>
+<div class="section"><div class="titlepage"><div><div><h2 class="title" style="clear: both">
+<a name="maps.array_"></a><a class="link" href="array_.html" title="array_">array_</a>
+</h2></div></div></div></div>
+<table xmlns:rev="http://www.cs.rpi.edu/~gregod/boost/tools/doc/revision" width="100%"><tr>
+<td align="left"></td>
+<td align="right"><div class="copyright-footer">Copyright © 2010 -2011 Brian Smith
+ Distributed under the Boost Software License, Version 1.0. (See accompanying
+ file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
+ 
+</div></td>
+</tr></table>
+<hr>
+<div class="spirit-nav">
+<a accesskey="p" href="array.html"><img src="../../../../../doc/src/images/prev.png" alt="Prev"></a><a accesskey="u" href="../index.html"><img src="../../../../../doc/src/images/up.png" alt="Up"></a><a accesskey="h" href="../index.html"><img src="../../../../../doc/src/images/home.png" alt="Home"></a><a accesskey="n" href="pointer.html"><img src="../../../../../doc/src/images/next.png" alt="Next"></a>
+</div>
+</body>
+</html>

Added: sandbox/maps/libs/maps/doc/html/maps/concepts.html
==============================================================================
--- (empty file)
+++ sandbox/maps/libs/maps/doc/html/maps/concepts.html 2011-08-23 16:21:41 EDT (Tue, 23 Aug 2011)
@@ -0,0 +1,526 @@
+<html>
+<head>
+<meta http-equiv="Content-Type" content="text/html; charset=US-ASCII">
+<title>Concepts</title>
+<link rel="stylesheet" href="../../../../../doc/src/boostbook.css" type="text/css">
+<meta name="generator" content="DocBook XSL Stylesheets V1.76.1">
+<link rel="home" href="../index.html" title="Boost.Maps">
+<link rel="up" href="../index.html" title="Boost.Maps">
+<link rel="prev" href="../index.html" title="Boost.Maps">
+<link rel="next" href="tutorial.html" title="Tutorial">
+</head>
+<body bgcolor="white" text="black" link="#0000FF" vlink="#840084" alink="#0000FF">
+<table cellpadding="2" width="100%"><tr>
+<td valign="top"><img alt="Boost C++ Libraries" width="277" height="86" src="../../../../../boost.png"></td>
+<td align="center">Home</td>
+<td align="center">Libraries</td>
+<td align="center">People</td>
+<td align="center">FAQ</td>
+<td align="center">More</td>
+</tr></table>
+<hr>
+<div class="spirit-nav">
+<a accesskey="p" href="../index.html"><img src="../../../../../doc/src/images/prev.png" alt="Prev"></a><a accesskey="u" href="../index.html"><img src="../../../../../doc/src/images/up.png" alt="Up"></a><a accesskey="h" href="../index.html"><img src="../../../../../doc/src/images/home.png" alt="Home"></a><a accesskey="n" href="tutorial.html"><img src="../../../../../doc/src/images/next.png" alt="Next"></a>
+</div>
+<div class="section">
+<div class="titlepage"><div><div><h2 class="title" style="clear: both">
+<a name="maps.concepts"></a><a class="link" href="concepts.html" title="Concepts">Concepts</a>
+</h2></div></div></div>
+<div class="toc"><dl>
+<dt>Introduction</dt>
+<dt>Fixed-Size Arrays</dt>
+<dt>Resizeable Arrays</dt>
+<dt>Array Views</dt>
+<dt><a href="concepts.html#maps.concepts.scalars_vectors_matrices">Scalars, Vectors
+ and Matrices</a></dt>
+</dl></div>
+<div class="section">
+<div class="titlepage"><div><div><h3 class="title">
+<a name="maps.concepts.introduction"></a><a class="link" href="concepts.html#maps.concepts.introduction" title="Introduction">Introduction</a>
+</h3></div></div></div>
+
+ The array of array specification for statically allocated intrinsic multi-dimensional
+ arrays is conceptually well defined in C++. The fixed-size array class templates
+ discussed here include versions whose data member is an intrinsic multi-dimensional
+ array. The arrays' shape is determined from a compile time supplied set of
+ bounds. An alternative approach to constructing multi-dimensional arrays
+ results from using pointers. The structural components in this case are allocated
+ dynamically at runtime allowing for both fixed-size and resizeable varieties.
+ The approach taken here is to allocate all the elements at each dimension
+ in a continuous block of memory from a given set of bounds. For fixed-size
+ arrays this again means the bounds are set at compile time. A resizeable
+ array must be passed a set of bounds in order to construct it in addition
+ to the purpose of resizing.
+ 
+</div>
+<div class="section">
+<div class="titlepage"><div><div><h3 class="title">
+<a name="maps.concepts.fixed_size_arrays"></a><a class="link" href="concepts.html#maps.concepts.fixed_size_arrays" title="Fixed-Size Arrays">Fixed-Size Arrays</a>
+</h3></div></div></div>
+
+ The array types <code class="computeroutput">array</code> and
+ <code class="computeroutput">array_</code> define objects that
+ satisfy the notion of a multi-dimensional array. The data member type generated
+ for the array depends on the template arguments to the class. The header
+ <code class="computeroutput"><array.hpp></code>
+ defines the class template,
+ 
+<pre class="programlisting">template< class T, class B, bool I, class A > class array;
+</pre>
+
+ Here,
+ 
+<div class="itemizedlist"><ul class="itemizedlist" type="disc">
+<li class="listitem">
+ <code class="computeroutput">T</code> is the data type the
+ array holds.
+ </li>
+<li class="listitem">
+ <code class="computeroutput">B</code> is a <code class="computeroutput">boundsN</code> sequence of <code class="computeroutput">N</code>
+ non-zero positive integers.
+ </li>
+<li class="listitem">
+ <code class="computeroutput">I</code> is the zero/default
+ initialization parameter.
+ </li>
+<li class="listitem">
+ <code class="computeroutput">A</code> is an allocator type.
+ </li>
+</ul></div>
+
+ The sequence argument, <code class="computeroutput">B</code>,
+ determines the dimensionality, <code class="computeroutput">N</code>,
+ of the array, as well as the size of each dimension. The library defines
+ a <code class="computeroutput">null::allocator</code> type which is the default allocator
+ type for the class. If a <code class="computeroutput">null::allocator</code>
+ is used for the template argument <code class="computeroutput">A</code>
+ then the resulting data member is a statically allocated array of <code class="computeroutput">N</code> dimensions with the size of each dimension
+ given by successive values in the argument sequence <code class="computeroutput">B</code>,
+ and with data elements of type <code class="computeroutput">T</code>,
+ where the usual restrictions on type <code class="computeroutput">T</code>
+ apply. Otherwise the array data member is the type <code class="computeroutput">T</code>
+ or its non-reference equivalent, which from here on its assumed to be, with
+ <code class="computeroutput">N</code> pointer symbols appended
+ and is constructed at runtime in accordance with template argument's <code class="computeroutput">B</code>, <code class="computeroutput">I</code>
+ and <code class="computeroutput">A</code>. In either case if
+ the argument <code class="computeroutput">T</code> has a non-trivial
+ constructor the data elements will be constructed during the construction
+ of the array, standard case for intrinsic arrays, otherwise data element
+ construction will depend on the argument <code class="computeroutput">I</code>,
+ whose default value is <code class="computeroutput">true</code>.
+ Therefore, arguments are always required for both <code class="computeroutput">T</code>
+ and <code class="computeroutput">B</code>, while <code class="computeroutput">I</code> and <code class="computeroutput">A</code>
+ are optional.
+ 
+
+ As an illustration the following declaration produces an <code class="computeroutput">array</code>
+ object with data member type <code class="computeroutput">int[3][4][5][4][3]</code> and zero
+ initialized data elements.
+ 
+<pre class="programlisting">array< int, bounds5< 3,4,5,4,3 > > a;
+</pre>
+
+ If we now consider the declaration,
+ 
+<pre class="programlisting">array< int, bounds5< 3,4,5,4,3 >, false, std::allocator< int > > a;
+</pre>
+
+ The <code class="computeroutput">array</code> object's data member
+ in this case is an <code class="computeroutput">int*****</code>
+ whose runtime construction requires five memory allocations with the data
+ elements left uninitialized. A point of note concerns the allocator type
+ used in the previous declaration. Strictly, it should have been <code class="computeroutput">std::allocator< int****
+ ></code>, however the nested <code class="computeroutput">rebind</code> struct in the original declaration's
+ allocator is used to bind the template parameter to the correct type. Since
+ a non-static N-dimensional array requires N allocations for N different types,
+ N different allocators are necessary to fully construct the resulting array.
+ 
+
+ The header <code class="computeroutput"><array_.hpp></code>
+ defines the class template,
+ 
+<pre class="programlisting">template< class T, std::size_t N, std::size_t D, bool I, class A > class array_;
+</pre>
+
+ Here,
+ 
+<div class="itemizedlist"><ul class="itemizedlist" type="disc">
+<li class="listitem">
+ <code class="computeroutput">T</code> is the data type the
+ array holds.
+ </li>
+<li class="listitem">
+ <code class="computeroutput">N</code> is the dimensionality
+ of the array.
+ </li>
+<li class="listitem">
+ <code class="computeroutput">D</code> is the size of each
+ of the <code class="computeroutput">N</code> dimensions of
+ the array.
+ </li>
+<li class="listitem">
+ <code class="computeroutput">I</code> is the zero/default
+ initialization parameter.
+ </li>
+<li class="listitem">
+ <code class="computeroutput">A</code> is an allocator type.
+ </li>
+</ul></div>
+
+ The template parameters <code class="computeroutput">T</code>,
+ <code class="computeroutput">I</code> and <code class="computeroutput">A</code>
+ have the same meaning, requirements and default behaviour as the <code class="computeroutput">array</code> class template discussed previously.
+ An <code class="computeroutput">array_</code> object has dimensionality
+ <code class="computeroutput">N</code> with each dimension taking
+ the fixed value <code class="computeroutput">D</code>, both of
+ which are required parameters. Thus, in the following declaration the data
+ member type is <code class="computeroutput">int[5][5][5][5][5]</code> with zero initialized data elements.
+ 
+<pre class="programlisting">array_< int, 5,5 > a;
+</pre>
+
+ The dynamically allocated version with uninitialized data elements results
+ from,
+ 
+<pre class="programlisting">array_< int, 5,5, false, std::allocator< int > > a;
+</pre>
+</div>
+<div class="section">
+<div class="titlepage"><div><div><h3 class="title">
+<a name="maps.concepts.resizeable_arrays"></a><a class="link" href="concepts.html#maps.concepts.resizeable_arrays" title="Resizeable Arrays">Resizeable Arrays</a>
+</h3></div></div></div>
+
+ The arrays discussed in this section satisfy the notion of a dynamically
+ allocated multi-dimensional array. The header <code class="computeroutput"><pointer.hpp></code>
+ defines the class template,
+ 
+<pre class="programlisting">template< class T, std::size_t N, bool I, class A > class pointer;
+</pre>
+
+ Here,
+ 
+<div class="itemizedlist"><ul class="itemizedlist" type="disc">
+<li class="listitem">
+ <code class="computeroutput">T</code> is the data type the
+ array holds.
+ </li>
+<li class="listitem">
+ <code class="computeroutput">N</code> is the dimensionality
+ of the array.
+ </li>
+<li class="listitem">
+ <code class="computeroutput">I</code> is the zero/default
+ initialization parameter.
+ </li>
+<li class="listitem">
+ <code class="computeroutput">A</code> is an allocator type.
+ </li>
+</ul></div>
+
+ Parameters <code class="computeroutput">T</code> and <code class="computeroutput">N</code> are both required, while <code class="computeroutput">I</code> and <code class="computeroutput">A</code>
+ are optional, with respective default's <code class="computeroutput">false</code>
+ and <code class="computeroutput">std::allocator<
+ T ></code>.
+ In order to construct a <code class="computeroutput">pointer</code>
+ type with specific dimensions the auxiliary type <code class="computeroutput">bounds</code>,
+ defined in the header <code class="computeroutput"><bounds.hpp></code>,
+ is required. A <code class="computeroutput">bounds</code> type
+ is a template array type with a single <code class="computeroutput">std::size_t</code>
+ parameter that determines the number of <code class="computeroutput">std::size_t</code>
+ type elements the array should hold. Notwithstanding Scott Meyers item[<a name="id753190" href="#ftn.id753190" class="footnote">1</a>] and similar to Blitz++[<a name="id753199" href="#ftn.id753199" class="footnote">2</a>] arrays the comma operator has been overloaded for convenience.
+ The following declaration and assignment results in the <code class="computeroutput">bounds</code>
+ array <code class="computeroutput">b</code> having values 3,
+ 4, 5, 4 and 3 respectively.
+ 
+<pre class="programlisting">bounds<5> b;
+b = 3,4,5,4,3;
+</pre>
+
+ Creating a <code class="computeroutput">pointer</code> object
+ with uninitialized data elements then follows as,
+ 
+<pre class="programlisting">pointer< double, 5 > a( b );
+</pre>
+
+ To continue, we can further write,
+ 
+<pre class="programlisting">b = 5,4,3,4,5;
+a.resize< false >( b );
+</pre>
+
+ This results in the array being resized to the updated values in <code class="computeroutput">b</code> without preserving the values already
+ present in <code class="computeroutput">a</code>. For a <code class="computeroutput">pointer</code> type with dimensionality <code class="computeroutput">N</code>, <code class="computeroutput">N</code>
+ memory allocations are again required. However, for an already constructed
+ array such as <code class="computeroutput">a</code> above it
+ is only necessary to make a request for memory during resizing if the total
+ number of elements required for each dimension is different from that already
+ present. At least this is the case for all but the final request where a
+ memory allocation is required if the template argument to <code class="computeroutput">resize</code>
+ is <code class="computeroutput">false</code> or if the template
+ argument to <code class="computeroutput">resize</code> is <code class="computeroutput">true</code> and the total number of elements differs
+ from the requested new total, thus maintaining the exception guarantee specified
+ for the class. For example, in the resize operation above the third request
+ for memory is to accommodate 5 x 4 x 3 = 60 elements, while 3 x 4 x 5 = 60
+ elements of the required type are already present, hence, no request for
+ memory is needed. Pedantically, the types <code class="computeroutput">dimensions</code>
+ and <code class="computeroutput">indexes</code> may be used in
+ place of a <code class="computeroutput">bounds</code> type for
+ construction and element access of a resizeable array.
+ 
+</div>
+<div class="section">
+<div class="titlepage"><div><div><h3 class="title">
+<a name="maps.concepts.array_views"></a><a class="link" href="concepts.html#maps.concepts.array_views" title="Array Views">Array Views</a>
+</h3></div></div></div>
+
+ Accessing the elements of an array can be considered from a number of perspectives
+ that inevitably depends on the particular application and patterns in the
+ data represented by the elements. The indexing operator for example is used
+ to access individual elements of an array. However access to specific non-/continuous
+ sequences of elements in the form of slices or subarrays may be required.
+ For an N-dimensional array a slice can be represented in 1 to N-1 dimensions
+ whereas subarrays of N-dimensional arrays are themselves N-dimensional arrays.
+ A <code class="computeroutput">view</code> is a resizeable N-dimensional
+ array of the same dimesionality as the array used to define it. The data
+ elements can be either pointers to or copies of some sequence of data elements
+ from an array whose type parameterizes the <code class="computeroutput">view</code>.
+ 
+
+ The header <code class="computeroutput"><view.hpp></code>
+ defines the class template,
+ 
+<pre class="programlisting">template< class A, bool B > class view;
+</pre>
+
+ Here,
+ 
+<div class="itemizedlist"><ul class="itemizedlist" type="disc">
+<li class="listitem">
+ <code class="computeroutput">A</code> is the array type the
+ <code class="computeroutput">view</code> will be constructed
+ from.
+ </li>
+<li class="listitem">
+ <code class="computeroutput">B</code> signifies our intention
+ to store pointers on an argument of type <code class="computeroutput">A</code>
+ or otherwise.
+ </li>
+</ul></div>
+
+ <code class="computeroutput">A</code> is a required parameter
+ while <code class="computeroutput">B</code> is optional with
+ default value <code class="computeroutput">true</code>, i.e. stores
+ pointers to some sequence of data elements from an array of type <code class="computeroutput">A</code>. The <code class="computeroutput">view</code>
+ type contains the nested class template <code class="computeroutput">range</code>,
+ used to set three values, first, last and step, over one dimension of an
+ array of type <code class="computeroutput">A</code>. The relationship
+ between first, last and step can be considered analogous to the following
+ <code class="computeroutput">for</code> statement.
+ 
+<pre class="programlisting">for( std::size_t i = first; i != last; i += step ) { ... }
+</pre>
+
+ The single template parameter of a <code class="computeroutput">range</code>
+ is the type <code class="computeroutput">std::size_t</code> that determines which dimension
+ the <code class="computeroutput">range</code> is associated with.
+ For an N-dimensional array, N <code class="computeroutput">range</code>'s
+ should be set which together form the sequence of data elements from an argument
+ of type <code class="computeroutput">A</code> that the <code class="computeroutput">view</code> will store. The <code class="computeroutput">range</code>
+ constructor expects an argument of type <code class="computeroutput">view</code>.
+ Implicitly a <code class="computeroutput">view</code> must be
+ declared before setting the <code class="computeroutput">range</code>'s
+ and subsequently an argument of type <code class="computeroutput">A</code>
+ is passed to the <code class="computeroutput">view</code>'s
+ <code class="computeroutput">reset</code> method in order to
+ construct it. To clarify, consider the following,
+ 
+<pre class="programlisting">typedef array_< double, 3,7, true > A;
+typedef view< A > V;
+
+A a;
+V v;
+
+V::range<0> r0( v ); r0 = 1,7,2;
+V::range<1> r1( v ); r1 = 1,7,2;
+V::range<2> r2( v ); r2 = 1,7,2;
+
+v.reset( a );
+</pre>
+
+ The call to <code class="computeroutput">reset</code> in this
+ example constructs a three-dimensional <code class="computeroutput">view</code>
+ with shape 3 x 3 x 3 whose data elements store the addresses of every second
+ data element in the third dimension of every second array in the second dimension
+ of every second array of array in the first dimension of the 7 x 7 x 7 <code class="computeroutput">array_ a</code>.
+ 
+
+ The header <code class="computeroutput"><view_.hpp></code>
+ defines the class template,
+ 
+<pre class="programlisting">template< class A, class R, bool B > class view_;
+</pre>
+
+ Here,
+ 
+<div class="itemizedlist"><ul class="itemizedlist" type="disc">
+<li class="listitem">
+ <code class="computeroutput">A</code> is the array type the
+ <code class="computeroutput">view_</code> will be constructed
+ from.
+ </li>
+<li class="listitem">
+ <code class="computeroutput">R</code> is a sequence of ranges.
+ </li>
+<li class="listitem">
+ <code class="computeroutput">B</code> signifies our intention
+ to store pointers on an argument of type <code class="computeroutput">A</code>
+ or otherwise.
+ </li>
+</ul></div>
+
+ A <code class="computeroutput">view_</code> is used to construct
+ a view of a fixed-size array. The data member of a <code class="computeroutput">view_</code>
+ will be statically or dynamically allocated dependant on the type of the
+ data member of the array type <code class="computeroutput">A</code>.
+ <code class="computeroutput">A</code> and <code class="computeroutput">R</code>
+ are both required parameters with <code class="computeroutput">B</code>
+ optional defaulting to <code class="computeroutput">true</code>.
+ A <code class="computeroutput">view_</code> type is more restricted
+ than a <code class="computeroutput">view</code> in the sense
+ that the parameter <code class="computeroutput">R</code> defines
+ at compile time the ranges over an array of type <code class="computeroutput">A</code>
+ associated with the <code class="computeroutput">view_</code>.
+ Template types <code class="computeroutput">rangesN</code>,
+ <code class="computeroutput">N</code> the dimensionality of an
+ array of type <code class="computeroutput">A</code>, and <code class="computeroutput">range</code> are defined specifically for the
+ template argument <code class="computeroutput">R</code>. Since
+ the range's are predetermined the <code class="computeroutput">view_</code>
+ type defines a constructor accepting an argument of type <code class="computeroutput">A</code>.
+ Thus, we can write,
+ 
+<pre class="programlisting">typedef array_< double, 3,7, true > A;
+typedef ranges3< range< 1,7,2 >, range< 1,7,2 >, range< 1,7,2 > > R;
+typedef view_< A, R > V;
+
+A a;
+V v( a );
+</pre>
+</div>
+<div class="section">
+<div class="titlepage"><div><div><h3 class="title">
+<a name="maps.concepts.scalars_vectors_matrices"></a><a class="link" href="concepts.html#maps.concepts.scalars_vectors_matrices" title="Scalars, Vectors and Matrices">Scalars, Vectors
+ and Matrices</a>
+</h3></div></div></div>
+
+ In certain problem domains we can specify a set of integral dimensions which
+ fully characterize the problem space. Fixed-size matrices for example are
+ common in many algebraic problems. Knowing the initial dimensions in advance
+ allows us to exploit this knowledge in all future computations. For example,
+ given a matrix, M1, with M rows and P columns and a matrix, M2, with P rows
+ and N columns, the product M1 x M2 is a matrix, M3 say, with M rows and N
+ columns. Computational requirements such as the operand and result type dimensions
+ in this example can be encoded in the template parameters of the types involved
+ and also in the operators that act on them. The Boost maps library provides
+ implementations for <code class="computeroutput">scalar</code>
+ and fixed-size <code class="computeroutput">vector</code> and
+ <code class="computeroutput">matrix</code> operators that conform
+ to such requirements, enabling compile time determination of consistent operations
+ on the operands involved. The expression template paradigm has been invoked
+ to enhance performance for mathematically intensive applications. Note that
+ <code class="computeroutput">scalar</code> operands are a convenient
+ way to specify operations involving scalar types and are not intended to
+ be used as the data elements of the <code class="computeroutput">vector</code>
+ or <code class="computeroutput">matrix</code> types. The headers
+ <code class="computeroutput"><scalar.hpp></code>,
+ <code class="computeroutput"><vector.hpp></code>
+ and <code class="computeroutput"><matrix.hpp></code>
+ define the following respective types.
+ 
+<pre class="programlisting">template< class T > class scalar;
+</pre>
+
+ Here,
+ 
+<div class="itemizedlist"><ul class="itemizedlist" type="disc"><li class="listitem">
+ <code class="computeroutput">T</code> is the data type of
+ the stored member.
+ </li></ul></div>
+<pre class="programlisting">template< class T, std::size_t N, bool I, class A > class vector;
+</pre>
+
+ Here,
+ 
+<div class="itemizedlist"><ul class="itemizedlist" type="disc">
+<li class="listitem">
+ <code class="computeroutput">T</code> is the element type.
+ </li>
+<li class="listitem">
+ <code class="computeroutput">N</code> is the number of components.
+ </li>
+<li class="listitem">
+ <code class="computeroutput">I</code> is the zero/default
+ initialization parameter.
+ </li>
+<li class="listitem">
+ <code class="computeroutput">A</code> is an allocator type.
+ </li>
+</ul></div>
+<pre class="programlisting">template< class T, std::size_t M, std::size_t N, bool I, class A > class matrix;
+</pre>
+
+ Here,
+ 
+<div class="itemizedlist"><ul class="itemizedlist" type="disc">
+<li class="listitem">
+ <code class="computeroutput">T</code> is the element type.
+ </li>
+<li class="listitem">
+ <code class="computeroutput">M</code> is the number of rows.
+ </li>
+<li class="listitem">
+ <code class="computeroutput">N</code> is the number of columns.
+ </li>
+<li class="listitem">
+ <code class="computeroutput">I</code> is the zero/default
+ initialization parameter.
+ </li>
+<li class="listitem">
+ <code class="computeroutput">A</code> is an allocator type.
+ </li>
+</ul></div>
+
+ For all three class templates <code class="computeroutput">T</code>
+ is a required parameter. The parameters <code class="computeroutput">N</code>
+ for a <code class="computeroutput">vector</code> and, <code class="computeroutput">M</code> and <code class="computeroutput">N</code>
+ for a <code class="computeroutput">matrix</code> are also required.
+ The remaining parameters <code class="computeroutput">I</code>
+ and <code class="computeroutput">A</code> for the <code class="computeroutput">vector</code> and <code class="computeroutput">matrix</code>
+ class templates are optional with respective default's <code class="computeroutput">true</code>
+ and <code class="computeroutput">null::allocator</code> for both classes.
+ 
+</div>
+<div class="footnotes">
+ <hr width="100" align="left">
+<div class="footnote">[<a id="ftn.id753190" href="#id753190" class="para">1</a>] 
+ Scott Meyers, More Effective C++, Addison Wesley Longman, Inc., 1996, item
+ 7.
+ </div>
+<div class="footnote">[<a id="ftn.id753199" href="#id753199" class="para">2</a>] 
+ Todd Veldhuizen, Blitz++, http://www.oonumerics.org/blitz/.
+ </div>
+</div>
+</div>
+<table xmlns:rev="http://www.cs.rpi.edu/~gregod/boost/tools/doc/revision" width="100%"><tr>
+<td align="left"></td>
+<td align="right"><div class="copyright-footer">Copyright © 2010 -2011 Brian Smith
+ Distributed under the Boost Software License, Version 1.0. (See accompanying
+ file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
+ 
+</div></td>
+</tr></table>
+<hr>
+<div class="spirit-nav">
+<a accesskey="p" href="../index.html"><img src="../../../../../doc/src/images/prev.png" alt="Prev"></a><a accesskey="u" href="../index.html"><img src="../../../../../doc/src/images/up.png" alt="Up"></a><a accesskey="h" href="../index.html"><img src="../../../../../doc/src/images/home.png" alt="Home"></a><a accesskey="n" href="tutorial.html"><img src="../../../../../doc/src/images/next.png" alt="Next"></a>
+</div>
+</body>
+</html>

Added: sandbox/maps/libs/maps/doc/html/maps/pointer.html
==============================================================================
--- (empty file)
+++ sandbox/maps/libs/maps/doc/html/maps/pointer.html 2011-08-23 16:21:41 EDT (Tue, 23 Aug 2011)
@@ -0,0 +1,41 @@
+<html>
+<head>
+<meta http-equiv="Content-Type" content="text/html; charset=US-ASCII">
+<title>pointer</title>
+<link rel="stylesheet" href="../../../../../doc/src/boostbook.css" type="text/css">
+<meta name="generator" content="DocBook XSL Stylesheets V1.76.1">
+<link rel="home" href="../index.html" title="Boost.Maps">
+<link rel="up" href="../index.html" title="Boost.Maps">
+<link rel="prev" href="array_.html" title="array_">
+<link rel="next" href="view.html" title="view">
+</head>
+<body bgcolor="white" text="black" link="#0000FF" vlink="#840084" alink="#0000FF">
+<table cellpadding="2" width="100%"><tr>
+<td valign="top"><img alt="Boost C++ Libraries" width="277" height="86" src="../../../../../boost.png"></td>
+<td align="center">Home</td>
+<td align="center">Libraries</td>
+<td align="center">People</td>
+<td align="center">FAQ</td>
+<td align="center">More</td>
+</tr></table>
+<hr>
+<div class="spirit-nav">
+<a accesskey="p" href="array_.html"><img src="../../../../../doc/src/images/prev.png" alt="Prev"></a><a accesskey="u" href="../index.html"><img src="../../../../../doc/src/images/up.png" alt="Up"></a><a accesskey="h" href="../index.html"><img src="../../../../../doc/src/images/home.png" alt="Home"></a><a accesskey="n" href="view.html"><img src="../../../../../doc/src/images/next.png" alt="Next"></a>
+</div>
+<div class="section"><div class="titlepage"><div><div><h2 class="title" style="clear: both">
+<a name="maps.pointer"></a><a class="link" href="pointer.html" title="pointer">pointer</a>
+</h2></div></div></div></div>
+<table xmlns:rev="http://www.cs.rpi.edu/~gregod/boost/tools/doc/revision" width="100%"><tr>
+<td align="left"></td>
+<td align="right"><div class="copyright-footer">Copyright © 2010 -2011 Brian Smith
+ Distributed under the Boost Software License, Version 1.0. (See accompanying
+ file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
+ 
+</div></td>
+</tr></table>
+<hr>
+<div class="spirit-nav">
+<a accesskey="p" href="array_.html"><img src="../../../../../doc/src/images/prev.png" alt="Prev"></a><a accesskey="u" href="../index.html"><img src="../../../../../doc/src/images/up.png" alt="Up"></a><a accesskey="h" href="../index.html"><img src="../../../../../doc/src/images/home.png" alt="Home"></a><a accesskey="n" href="view.html"><img src="../../../../../doc/src/images/next.png" alt="Next"></a>
+</div>
+</body>
+</html>

Added: sandbox/maps/libs/maps/doc/html/maps/reference.html
==============================================================================
--- (empty file)
+++ sandbox/maps/libs/maps/doc/html/maps/reference.html 2011-08-23 16:21:41 EDT (Tue, 23 Aug 2011)
@@ -0,0 +1,60 @@
+<html>
+<head>
+<meta http-equiv="Content-Type" content="text/html; charset=US-ASCII">
+<title>Reference</title>
+<link rel="stylesheet" href="../../../../../doc/src/boostbook.css" type="text/css">
+<meta name="generator" content="DocBook XSL Stylesheets V1.76.1">
+<link rel="home" href="../index.html" title="Boost.Maps">
+<link rel="up" href="../index.html" title="Boost.Maps">
+<link rel="prev" href="tutorial.html" title="Tutorial">
+<link rel="next" href="array.html" title="array">
+</head>
+<body bgcolor="white" text="black" link="#0000FF" vlink="#840084" alink="#0000FF">
+<table cellpadding="2" width="100%"><tr>
+<td valign="top"><img alt="Boost C++ Libraries" width="277" height="86" src="../../../../../boost.png"></td>
+<td align="center">Home</td>
+<td align="center">Libraries</td>
+<td align="center">People</td>
+<td align="center">FAQ</td>
+<td align="center">More</td>
+</tr></table>
+<hr>
+<div class="spirit-nav">
+<a accesskey="p" href="tutorial.html"><img src="../../../../../doc/src/images/prev.png" alt="Prev"></a><a accesskey="u" href="../index.html"><img src="../../../../../doc/src/images/up.png" alt="Up"></a><a accesskey="h" href="../index.html"><img src="../../../../../doc/src/images/home.png" alt="Home"></a><a accesskey="n" href="array.html"><img src="../../../../../doc/src/images/next.png" alt="Next"></a>
+</div>
+<div class="section">
+<div class="titlepage"><div><div><h2 class="title" style="clear: both">
+<a name="maps.reference"></a><a class="link" href="reference.html" title="Reference">Reference</a>
+</h2></div></div></div>
+<div class="itemizedlist"><ul class="itemizedlist" type="disc">
+<li class="listitem">
+ <a class="link" href="array.html" title="array">array</a>
+ </li>
+<li class="listitem">
+ <a class="link" href="array_.html" title="array_">array_</a>
+ </li>
+<li class="listitem">
+ <a class="link" href="pointer.html" title="pointer">pointer</a>
+ </li>
+<li class="listitem">
+ <a class="link" href="view.html" title="view">view</a>
+ </li>
+<li class="listitem">
+ <a class="link" href="view_.html" title="view_">view_</a>
+ </li>
+</ul></div>
+</div>
+<table xmlns:rev="http://www.cs.rpi.edu/~gregod/boost/tools/doc/revision" width="100%"><tr>
+<td align="left"></td>
+<td align="right"><div class="copyright-footer">Copyright © 2010 -2011 Brian Smith
+ Distributed under the Boost Software License, Version 1.0. (See accompanying
+ file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
+ 
+</div></td>
+</tr></table>
+<hr>
+<div class="spirit-nav">
+<a accesskey="p" href="tutorial.html"><img src="../../../../../doc/src/images/prev.png" alt="Prev"></a><a accesskey="u" href="../index.html"><img src="../../../../../doc/src/images/up.png" alt="Up"></a><a accesskey="h" href="../index.html"><img src="../../../../../doc/src/images/home.png" alt="Home"></a><a accesskey="n" href="array.html"><img src="../../../../../doc/src/images/next.png" alt="Next"></a>
+</div>
+</body>
+</html>

Added: sandbox/maps/libs/maps/doc/html/maps/tutorial.html
==============================================================================
--- (empty file)
+++ sandbox/maps/libs/maps/doc/html/maps/tutorial.html 2011-08-23 16:21:41 EDT (Tue, 23 Aug 2011)
@@ -0,0 +1,292 @@
+<html>
+<head>
+<meta http-equiv="Content-Type" content="text/html; charset=US-ASCII">
+<title>Tutorial</title>
+<link rel="stylesheet" href="../../../../../doc/src/boostbook.css" type="text/css">
+<meta name="generator" content="DocBook XSL Stylesheets V1.76.1">
+<link rel="home" href="../index.html" title="Boost.Maps">
+<link rel="up" href="../index.html" title="Boost.Maps">
+<link rel="prev" href="concepts.html" title="Concepts">
+<link rel="next" href="reference.html" title="Reference">
+</head>
+<body bgcolor="white" text="black" link="#0000FF" vlink="#840084" alink="#0000FF">
+<table cellpadding="2" width="100%"><tr>
+<td valign="top"><img alt="Boost C++ Libraries" width="277" height="86" src="../../../../../boost.png"></td>
+<td align="center">Home</td>
+<td align="center">Libraries</td>
+<td align="center">People</td>
+<td align="center">FAQ</td>
+<td align="center">More</td>
+</tr></table>
+<hr>
+<div class="spirit-nav">
+<a accesskey="p" href="concepts.html"><img src="../../../../../doc/src/images/prev.png" alt="Prev"></a><a accesskey="u" href="../index.html"><img src="../../../../../doc/src/images/up.png" alt="Up"></a><a accesskey="h" href="../index.html"><img src="../../../../../doc/src/images/home.png" alt="Home"></a><a accesskey="n" href="reference.html"><img src="../../../../../doc/src/images/next.png" alt="Next"></a>
+</div>
+<div class="section">
+<div class="titlepage"><div><div><h2 class="title" style="clear: both">
+<a name="maps.tutorial"></a><a class="link" href="tutorial.html" title="Tutorial">Tutorial</a>
+</h2></div></div></div>
+<div class="toc"><dl>
+<dt>Arrays</dt>
+<dt><a href="tutorial.html#maps.tutorial.scalar_vector_matrix_tutorial">Scalars,
+ Vectors and Matrices</a></dt>
+</dl></div>
+<div class="section">
+<div class="titlepage"><div><div><h3 class="title">
+<a name="maps.tutorial.array_tutorial"></a><a class="link" href="tutorial.html#maps.tutorial.array_tutorial" title="Arrays">Arrays</a>
+</h3></div></div></div>
+
+ The intention in this section is to simply reveal the functionality of the
+ multi-dimensional array classes. The dominant feature is element access so
+ we'll consider the various methods available. The following example declares
+ a statically allocated, uninitialized <code class="computeroutput">array</code>,
+ <code class="computeroutput">a</code>, then loops over each dimension
+ setting the value of each data element to the sum of its indexes.
+ 
+<pre class="programlisting">typedef array< int, bounds5< 5,4,3,4,5 > > array_t;
+
+array_t a;
+
+for( std::size_t i = 0; i != a.bound<0>(); ++i )
+for( std::size_t j = 0; j != a.bound<1>(); ++j )
+for( std::size_t k = 0; k != a.bound<2>(); ++k )
+for( std::size_t l = 0; l != a.bound<3>(); ++l )
+for( std::size_t m = 0; m != a.bound<4>(); ++m )
+ a[i][j][k][l][m] = i + j + k + l + m;
+</pre>
+
+ Alternatively, in place of <code class="computeroutput">a[i][j][k][l][m]</code> we can
+ use indirection as follows, <code class="computeroutput">*(*(*(*(*(a+i)+j)+k)+l)+m)</code>.
+ No check with either of these methods is made to ensure the indexes are within
+ the bounds of the array. The <code class="computeroutput">bound</code>
+ method however returns the bound at each dimension and includes a compile
+ time assertion on its argument. Note that this method employs the same zero-based
+ indexing semantics the language provides for indexing array or pointer types.
+ If a debug assertion on the indexes is required we can write the following,
+ 
+<pre class="programlisting">bounds<5> b;
+
+for( std::size_t i = 0; i != a.bound<0>(); ++i )
+for( std::size_t j = 0; j != a.bound<1>(); ++j )
+for( std::size_t k = 0; k != a.bound<2>(); ++k )
+for( std::size_t l = 0; l != a.bound<3>(); ++l )
+for( std::size_t m = 0; m != a.bound<4>(); ++m )
+ a[b = i,j,k,l,m] += 1;
+</pre>
+
+ If the indexes should always be checked then replacing <code class="computeroutput">a[b =
+ i,j,k,l,m]</code> with
+ <code class="computeroutput">a.at((b
+ = i,j,k,l,m))</code>
+ will throw a <code class="computeroutput">std::out_of_range</code> exception on an invalid index.
+ 
+
+ When a function or operator is to be applied on all data elements in the
+ array we can do something along the following lines,
+ 
+<pre class="programlisting">std::for_each( a.begin(), a.end(), boost::lambda::_1 -= 1 );
+</pre>
+
+ The <code class="computeroutput">view</code> class template can
+ used to access non-continuous sequences of data elements as follows,
+ 
+<pre class="programlisting">typedef view< array_t > view_t;
+
+view_t v;
+
+view_t::range<0> r0( v ); r0 = 1,4,1;
+view_t::range<1> r1( v ); r1 = 0,2,1;
+view_t::range<2> r2( v ); r2 = 0,3,1;
+view_t::range<3> r3( v ); r3 = 2,4,1;
+view_t::range<4> r4( v ); r4 = 1,4,1;
+
+v.reset( a );
+
+for( std::size_t i = 0; i != v.bound<0>(); ++i )
+for( std::size_t j = 0; j != v.bound<1>(); ++j )
+for( std::size_t k = 0; k != v.bound<2>(); ++k )
+for( std::size_t l = 0; l != v.bound<3>(); ++l )
+for( std::size_t m = 0; m != v.bound<4>(); ++m )
+ *v[i][j][k][l][m] *= 1;
+
+r1 = 2,4,1; r3 = 0,2,1;
+v.reset( a );
+
+std::for_each( v.begin(), v.end(), *boost::lambda::_1 *= 1 );
+</pre>
+
+ Given that <code class="computeroutput">v</code> represents a
+ 3 x 2 x 3 x 2 x 3 array and that its values are the addresses of elements
+ in the array <code class="computeroutput">a</code>, we can set
+ the values of any array whose dimensions are the same as those of <code class="computeroutput">v</code> to the corresponding values in <code class="computeroutput">a</code>.
+ 
+<pre class="programlisting">b = 3,2,3,2,3;
+pointer< int, 5 > p( b );
+v.to( p );
+</pre>
+
+ Similarly, again provided we store the addresses of a sequence of values
+ from the representative array, we can set those values to the values in an
+ array with the same dimensions as the <code class="computeroutput">view</code>,
+ so that given an array <code class="computeroutput">q</code>
+ with shape 3 x 2 x 3 x 2 x 3 we can write, <code class="computeroutput">v.from(
+ q )</code>.
+ 
+</div>
+<div class="section">
+<div class="titlepage"><div><div><h3 class="title">
+<a name="maps.tutorial.scalar_vector_matrix_tutorial"></a><a class="link" href="tutorial.html#maps.tutorial.scalar_vector_matrix_tutorial" title="Scalars, Vectors and Matrices">Scalars,
+ Vectors and Matrices</a>
+</h3></div></div></div>
+
+ At present, only dense vectors and matrices with associated operations are
+ available. The following gives some examples.
+ 
+<pre class="programlisting">matrix< int, 2,2 > m1, m2;
+vector< int, 2 > r, v, v1, v2;
+scalar< int > s;
+
+s = 3;
+
+for( int i = 0; i != 2; ++i )
+{
+ v[i] = i + 3;
+ v1[i] = i + 1;
+ v2[i] = i + 2;
+
+ for( int j = 0; j != 2; ++j )
+ {
+ m1[i][j] = i + j + 1;
+ m2[i][j] = i + j + 2;
+ }
+}
+
+r = outer_product( m1 * v1, v2 * m2 ) * ( s * v );
+</pre>
+
+ No distinction is made between row and column vectors hence pre and post
+ multiplication of a matrix by the same vector is possible provided the matrix
+ is square or otherwise whenever the number of vector components equals the
+ number of matrix rows for pre-multiplication or the number of matrix columns
+ for post-multiplication. For n matrices, n > 2, it is more efficient to
+ produce temporary intermediate results when they are multiplied in succession
+ using the <code class="computeroutput">*</code> operator. Determining
+ when such a chain of multiplications are being applied has proven to be inefficient
+ thus a temporary matrix is produced for the case n = 2 also. In order to
+ avoid this the <code class="computeroutput">product</code> operator
+ can be applied as shown below.
+ 
+<pre class="programlisting">const unsigned M = 1000;
+
+matrix< int, M,M, true, std::allocator< int > > m, m1, m2, m3;
+
+for( int i = 0; i != M; ++i )
+for( int j = 0; j != M; ++j )
+{
+ m1( i,j ) = 1;
+ m2( i,j ) = 2;
+ m3( i,j ) = 3;
+}
+
+m = product( m1, m2 ) + m3;
+</pre>
+
+ For some system dependant value of M the number of cache misses will dramatically
+ reduce the performance of the above calculation. In such cases we can reorganize
+ the computation as below, although setting up the matrices is not always
+ so easily accomplished.
+ 
+<pre class="programlisting">const unsigned N = 10;
+
+matrix< matrix< matrix< int,
+ N,N, true, std::allocator< int > >,
+ N,N, true, std::allocator< int > >,
+ N,N, true, std::allocator< int > > m, m1, m2, m3;
+
+for( int i = 0; i != N; ++i )
+for( int j = 0; j != N; ++j )
+{
+ for( int k = 0; k != N; ++k )
+ for( int l = 0; l != N; ++l )
+ {
+ for( int m = 0; m != N; ++m )
+ for( int n = 0; n != N; ++n )
+ {
+ m1( i,j )( k,l )( m,n ) = 1;
+ m2( i,j )( k,l )( m,n ) = 2;
+ m3( i,j )( k,l )( m,n ) = 3;
+ }
+ }
+}
+
+m = product( m1, m2 ) + m3;
+</pre>
+
+ Likewise, we can set up blocked computations with non-square matrices.
+ 
+<pre class="programlisting">matrix< matrix< int, 3,2 >, 2,3 > m1;
+matrix< matrix< int, 3,4 >, 2,3 > m2;
+matrix< matrix< int, 2,4 >, 3,3 > m, m3;
+
+for( std::size_t i = 0; i != 2; ++i )
+for( std::size_t j = 0; j != 3; ++j )
+for( std::size_t k = 0; k != 3; ++k )
+for( std::size_t l = 0; l != 2; ++l )
+ m1[i][j][k][l] = i + j + k + l;
+
+for( std::size_t i = 0; i != 2; ++i )
+for( std::size_t j = 0; j != 3; ++j )
+for( std::size_t k = 0; k != 3; ++k )
+for( std::size_t l = 0; l != 4; ++l )
+ m2[i][j][k][l] = i + j + k + l;
+
+for( std::size_t i = 0; i != 3; ++i )
+for( std::size_t j = 0; j != 3; ++j )
+for( std::size_t k = 0; k != 2; ++k )
+for( std::size_t l = 0; l != 4; ++l )
+ m3[i][j][k][l] = i + j + k + l;
+
+m = transpose( m1 ) * m2 + m3;
+</pre>
+
+ Or similarly with a combination of vectors and matrices.
+ 
+<pre class="programlisting">vector< vector< int, 3 >, 2 > v1, v2;
+matrix< matrix< int, 3,4 >, 2,3 > m, m1, m2;
+
+v1[0][0] = 1; v1[0][1] = 2; v1[0][2] = 3;
+v1[1][0] = 4; v1[1][1] = 5; v1[1][2] = 6;
+
+v2[0][0] = 6; v2[0][1] = 5; v2[0][2] = 4;
+v2[1][0] = 3; v2[1][1] = 2; v2[1][2] = 1;
+
+for( std::size_t i = 0; i != 2; ++i )
+for( std::size_t j = 0; j != 3; ++j )
+for( std::size_t k = 0; k != 2; ++k )
+for( std::size_t l = 0; l != 4; ++l )
+ m1[i][j][k][l] = i + j + k + l;
+
+for( std::size_t i = 0; i != 2; ++i )
+for( std::size_t j = 0; j != 3; ++j )
+for( std::size_t k = 0; k != 2; ++k )
+for( std::size_t l = 0; l != 4; ++l )
+ m2[i][j][k][l] = i + j + k + l;
+
+m = transpose( outer_product( v1, v2 ) ) * m1 + m2;
+</pre>
+</div>
+</div>
+<table xmlns:rev="http://www.cs.rpi.edu/~gregod/boost/tools/doc/revision" width="100%"><tr>
+<td align="left"></td>
+<td align="right"><div class="copyright-footer">Copyright © 2010 -2011 Brian Smith
+ Distributed under the Boost Software License, Version 1.0. (See accompanying
+ file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
+ 
+</div></td>
+</tr></table>
+<hr>
+<div class="spirit-nav">
+<a accesskey="p" href="concepts.html"><img src="../../../../../doc/src/images/prev.png" alt="Prev"></a><a accesskey="u" href="../index.html"><img src="../../../../../doc/src/images/up.png" alt="Up"></a><a accesskey="h" href="../index.html"><img src="../../../../../doc/src/images/home.png" alt="Home"></a><a accesskey="n" href="reference.html"><img src="../../../../../doc/src/images/next.png" alt="Next"></a>
+</div>
+</body>
+</html>

Added: sandbox/maps/libs/maps/doc/html/maps/view.html
==============================================================================
--- (empty file)
+++ sandbox/maps/libs/maps/doc/html/maps/view.html 2011-08-23 16:21:41 EDT (Tue, 23 Aug 2011)
@@ -0,0 +1,41 @@
+<html>
+<head>
+<meta http-equiv="Content-Type" content="text/html; charset=US-ASCII">
+<title>view</title>
+<link rel="stylesheet" href="../../../../../doc/src/boostbook.css" type="text/css">
+<meta name="generator" content="DocBook XSL Stylesheets V1.76.1">
+<link rel="home" href="../index.html" title="Boost.Maps">
+<link rel="up" href="../index.html" title="Boost.Maps">
+<link rel="prev" href="pointer.html" title="pointer">
+<link rel="next" href="view_.html" title="view_">
+</head>
+<body bgcolor="white" text="black" link="#0000FF" vlink="#840084" alink="#0000FF">
+<table cellpadding="2" width="100%"><tr>
+<td valign="top"><img alt="Boost C++ Libraries" width="277" height="86" src="../../../../../boost.png"></td>
+<td align="center">Home</td>
+<td align="center">Libraries</td>
+<td align="center">People</td>
+<td align="center">FAQ</td>
+<td align="center">More</td>
+</tr></table>
+<hr>
+<div class="spirit-nav">
+<a accesskey="p" href="pointer.html"><img src="../../../../../doc/src/images/prev.png" alt="Prev"></a><a accesskey="u" href="../index.html"><img src="../../../../../doc/src/images/up.png" alt="Up"></a><a accesskey="h" href="../index.html"><img src="../../../../../doc/src/images/home.png" alt="Home"></a><a accesskey="n" href="view_.html"><img src="../../../../../doc/src/images/next.png" alt="Next"></a>
+</div>
+<div class="section"><div class="titlepage"><div><div><h2 class="title" style="clear: both">
+<a name="maps.view"></a><a class="link" href="view.html" title="view">view</a>
+</h2></div></div></div></div>
+<table xmlns:rev="http://www.cs.rpi.edu/~gregod/boost/tools/doc/revision" width="100%"><tr>
+<td align="left"></td>
+<td align="right"><div class="copyright-footer">Copyright © 2010 -2011 Brian Smith
+ Distributed under the Boost Software License, Version 1.0. (See accompanying
+ file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
+ 
+</div></td>
+</tr></table>
+<hr>
+<div class="spirit-nav">
+<a accesskey="p" href="pointer.html"><img src="../../../../../doc/src/images/prev.png" alt="Prev"></a><a accesskey="u" href="../index.html"><img src="../../../../../doc/src/images/up.png" alt="Up"></a><a accesskey="h" href="../index.html"><img src="../../../../../doc/src/images/home.png" alt="Home"></a><a accesskey="n" href="view_.html"><img src="../../../../../doc/src/images/next.png" alt="Next"></a>
+</div>
+</body>
+</html>

Added: sandbox/maps/libs/maps/doc/html/maps/view_.html
==============================================================================
--- (empty file)
+++ sandbox/maps/libs/maps/doc/html/maps/view_.html 2011-08-23 16:21:41 EDT (Tue, 23 Aug 2011)
@@ -0,0 +1,40 @@
+<html>
+<head>
+<meta http-equiv="Content-Type" content="text/html; charset=US-ASCII">
+<title>view_</title>
+<link rel="stylesheet" href="../../../../../doc/src/boostbook.css" type="text/css">
+<meta name="generator" content="DocBook XSL Stylesheets V1.76.1">
+<link rel="home" href="../index.html" title="Boost.Maps">
+<link rel="up" href="../index.html" title="Boost.Maps">
+<link rel="prev" href="view.html" title="view">
+</head>
+<body bgcolor="white" text="black" link="#0000FF" vlink="#840084" alink="#0000FF">
+<table cellpadding="2" width="100%"><tr>
+<td valign="top"><img alt="Boost C++ Libraries" width="277" height="86" src="../../../../../boost.png"></td>
+<td align="center">Home</td>
+<td align="center">Libraries</td>
+<td align="center">People</td>
+<td align="center">FAQ</td>
+<td align="center">More</td>
+</tr></table>
+<hr>
+<div class="spirit-nav">
+<a accesskey="p" href="view.html"><img src="../../../../../doc/src/images/prev.png" alt="Prev"></a><a accesskey="u" href="../index.html"><img src="../../../../../doc/src/images/up.png" alt="Up"></a><a accesskey="h" href="../index.html"><img src="../../../../../doc/src/images/home.png" alt="Home"></a>
+</div>
+<div class="section"><div class="titlepage"><div><div><h2 class="title" style="clear: both">
+<a name="maps.view_"></a><a class="link" href="view_.html" title="view_">view_</a>
+</h2></div></div></div></div>
+<table xmlns:rev="http://www.cs.rpi.edu/~gregod/boost/tools/doc/revision" width="100%"><tr>
+<td align="left"></td>
+<td align="right"><div class="copyright-footer">Copyright © 2010 -2011 Brian Smith
+ Distributed under the Boost Software License, Version 1.0. (See accompanying
+ file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
+ 
+</div></td>
+</tr></table>
+<hr>
+<div class="spirit-nav">
+<a accesskey="p" href="view.html"><img src="../../../../../doc/src/images/prev.png" alt="Prev"></a><a accesskey="u" href="../index.html"><img src="../../../../../doc/src/images/up.png" alt="Up"></a><a accesskey="h" href="../index.html"><img src="../../../../../doc/src/images/home.png" alt="Home"></a>
+</div>
+</body>
+</html>

Added: sandbox/maps/libs/maps/doc/maps.qbk
==============================================================================
--- (empty file)
+++ sandbox/maps/libs/maps/doc/maps.qbk 2011-08-23 16:21:41 EDT (Tue, 23 Aug 2011)
@@ -0,0 +1,558 @@
+[book Boost.Maps
+ [quickbook 1.5]
+ [authors [Smith, Brian]]
+ [copyright 2010-2011 Brian Smith]
+ [id maps]
+ [license
+ Distributed under the Boost Software License, Version 1.0.
+ (See accompanying file LICENSE_1_0.txt or copy at
+ [@http://www.boost.org/LICENSE_1_0.txt http://www.boost.org/LICENSE_1_0.txt])
+ ]
+]
+
+[section:introduction Introduction]
+
+The Boost maps library provides sequence containers for statically and dynamically
+allocated multi-dimensional arrays. The implementations allow standard data element types,
+data element initialization, subject to conditions imposed by the data element type, and the
+choice of allocator used for memory allocation, element construction, etc., where appropriate.
+The library also includes scalar and fixed-size vector and matrix class templates with a DSEL
+implementation of operators via expression templates.
+
+The [link maps.concepts Concepts] section provides an overview of the library as well as a
+discussion of the supplementary structures that interoperate with the main classes in the
+library.
+
+*[link maps.concepts Concepts]
+*[link maps.tutorial Tutorial]
+*[link maps.reference Reference]
+
+[endsect]
+
+[section:concepts Concepts]
+
+[section:introduction Introduction]
+
+The array of array specification for statically allocated intrinsic multi-dimensional arrays is
+conceptually well defined in C++. The fixed-size array class templates discussed here include
+versions whose data member is an intrinsic multi-dimensional array. The arrays' shape is
+determined from a compile time supplied set of bounds. An alternative approach to constructing
+multi-dimensional arrays results from using pointers. The structural components in this case
+are allocated dynamically at runtime allowing for both fixed-size and resizeable varieties. The
+approach taken here is to allocate all the elements at each dimension in a continuous block of
+memory from a given set of bounds. For fixed-size arrays this again means the bounds are set at
+compile time. A resizeable array must be passed a set of bounds in order to construct it in
+addition to the purpose of resizing.
+
+[endsect] [/ Introduction]
+
+[section:fixed_size_arrays Fixed-Size Arrays]
+
+The array types `array` and `array_` define objects that satisfy the notion of a
+multi-dimensional array. The data member type generated for the array depends on the
+template arguments to the class. The header `<array.hpp>` defines the class template,
+
+ template< class T, class B, bool I, class A > class array;
+
+Here,
+
+* `T` is the data type the array holds.
+* `B` is a `boundsN` sequence of `N` non-zero positive integers.
+* `I` is the zero/default initialization parameter.
+* `A` is an allocator type.
+
+The sequence argument, `B`, determines the dimensionality, `N`, of the array,
+as well as the size of each dimension. The library defines a `null::allocator` type which
+is the default allocator type for the class. If a `null::allocator` is used for the
+template argument `A` then the resulting data member is a statically allocated array of
+`N` dimensions with the size of each dimension given by successive values in the argument
+sequence `B`, and with data elements of type `T`, where the usual restrictions on type `T`
+apply. Otherwise the array data member is the type `T` or its non-reference
+equivalent, which from here on its assumed to be, with `N` pointer symbols appended and is
+constructed at runtime in accordance with template argument's `B`, `I` and `A`. In either
+case if the argument `T` has a non-trivial constructor the data elements will be
+constructed during the construction of the array, standard case for intrinsic
+arrays, otherwise data element construction will depend on the argument `I`, whose default
+value is `true`. Therefore, arguments are always required for both `T` and `B`, while
+`I` and `A` are optional.
+
+As an illustration the following declaration produces an `array` object with data member
+type `int[3][4][5][4][3]` and zero initialized data elements.
+
+ array< int, bounds5< 3,4,5,4,3 > > a;
+
+If we now consider the declaration,
+
+ array< int, bounds5< 3,4,5,4,3 >, false, std::allocator< int > > a;
+
+The `array` object's data member in this case is an `int*****` whose runtime construction
+requires five memory allocations with the data elements left uninitialized. A point of note
+concerns the allocator type used in the previous declaration. Strictly, it should have been
+`std::allocator< int**** >`, however the nested `rebind` struct in the original
+declaration's allocator is used to bind the template parameter to the correct type. Since
+a non-static N-dimensional array requires N allocations for N different types, N different
+allocators are necessary to fully construct the resulting array.
+
+The header `<array_.hpp>` defines the class template,
+
+ template< class T, std::size_t N, std::size_t D, bool I, class A > class array_;
+
+Here,
+
+* `T` is the data type the array holds.
+* `N` is the dimensionality of the array.
+* `D` is the size of each of the `N` dimensions of the array.
+* `I` is the zero/default initialization parameter.
+* `A` is an allocator type.
+
+The template parameters `T`, `I` and `A` have the same meaning, requirements and default
+behaviour as the `array` class template discussed previously. An `array_` object has
+dimensionality `N` with each dimension taking the fixed value `D`, both of which are
+required parameters. Thus, in the following declaration the data member type is
+`int[5][5][5][5][5]` with zero initialized data elements.
+
+ array_< int, 5,5 > a;
+
+The dynamically allocated version with uninitialized data elements results from,
+
+ array_< int, 5,5, false, std::allocator< int > > a;
+
+[endsect] [/ Fixed-Size Arrays]
+
+[section:resizeable_arrays Resizeable Arrays]
+
+The arrays discussed in this section satisfy the notion of a dynamically allocated
+multi-dimensional array. The header `<pointer.hpp>` defines the class template,
+
+ template< class T, std::size_t N, bool I, class A > class pointer;
+
+Here,
+
+* `T` is the data type the array holds.
+* `N` is the dimensionality of the array.
+* `I` is the zero/default initialization parameter.
+* `A` is an allocator type.
+
+Parameters `T` and `N` are both required, while `I` and `A` are optional, with respective
+default's `false` and `std::allocator< T >`. In order to construct a `pointer` type with
+specific dimensions the auxiliary type `bounds`, defined in the header `<bounds.hpp>`, is
+required. A `bounds` type is a template array type with a single `std::size_t` parameter
+that determines the number of `std::size_t` type elements the array should hold.
+Notwithstanding Scott Meyers item[footnote Scott Meyers, More Effective C++, Addison Wesley
+Longman, Inc., 1996, item 7.] and similar to Blitz++[footnote Todd Veldhuizen, Blitz++,
+[@http://www.oonumerics.org/blitz/].] arrays the comma operator has been overloaded for
+convenience. The following declaration and assignment results in the `bounds` array `b`
+having values 3, 4, 5, 4 and 3 respectively.
+
+ bounds<5> b;
+ b = 3,4,5,4,3;
+
+Creating a `pointer` object with uninitialized data elements then follows as,
+
+ pointer< double, 5 > a( b );
+
+To continue, we can further write,
+
+ b = 5,4,3,4,5;
+ a.resize< false >( b );
+
+This results in the array being resized to the updated values in `b` without preserving the
+values already present in `a`. For a `pointer` type with dimensionality `N`, `N` memory
+allocations are again required. However, for an already constructed array such as `a` above
+it is only necessary to make a request for memory during resizing if the total number of
+elements required for each dimension is different from that already present. At least this
+is the case for all but the final request where a memory allocation is required if the
+template argument to `resize` is `false` or if the template argument to `resize` is `true`
+and the total number of elements differs from the requested new total, thus maintaining the
+exception guarantee specified for the class. For example, in the resize operation above the
+third request for memory is to accommodate 5 x 4 x 3 = 60 elements, while 3 x 4 x 5 = 60
+elements of the required type are already present, hence, no request for memory is needed.
+Pedantically, the types `dimensions` and `indexes` may be used in place of a `bounds` type
+for construction and element access of a resizeable array.
+
+[endsect] [/ Resizeable Arrays]
+
+[section:array_views Array Views]
+
+Accessing the elements of an array can be considered from a number of perspectives that
+inevitably depends on the particular application and patterns in the data represented
+by the elements. The indexing operator for example is used to access individual elements of
+an array. However access to specific non-/continuous sequences of elements in the form of
+slices or subarrays may be required. For an N-dimensional array a slice can be represented
+in 1 to N-1 dimensions whereas subarrays of N-dimensional arrays are themselves N-dimensional
+arrays. A `view` is a resizeable N-dimensional array of the same dimensionality as the array used
+to define it. The data elements can be either pointers to or copies of some sequence of data
+elements from an array whose type parameterizes the `view`.
+
+The header `<view.hpp>` defines the class template,
+
+ template< class A, bool B > class view;
+
+Here,
+
+* `A` is the array type the `view` will be constructed from.
+* `B` signifies our intention to store pointers on an argument of type `A` or otherwise.
+
+`A` is a required parameter while `B` is optional with default value `true`, i.e. stores
+pointers to some sequence of data elements from an array of type `A`. The `view` type
+contains the nested class template `range`, used to set three values, first, last and
+step, over one dimension of an array of type `A`. The relationship between first, last and
+step can be considered analogous to the following `for` statement.
+
+ for( std::size_t i = first; i != last; i += step ) { ... }
+
+The single template parameter of a `range` is the type `std::size_t` that determines which
+dimension the `range` is associated with. For an N-dimensional array, N `range`'s should be
+set which together form the sequence of data elements from an argument of type `A` that the
+`view` will store. The `range` constructor expects an argument of type `view`. Implicitly a
+`view` must be declared before setting the `range`'s and subsequently an argument of type
+`A` is passed to the `view`'s `reset` method in order to construct it. To clarify, consider
+the following,
+
+ typedef array_< double, 3,7, true > A;
+ typedef view< A > V;
+
+ A a;
+ V v;
+
+ V::range<0> r0( v ); r0 = 1,7,2;
+ V::range<1> r1( v ); r1 = 1,7,2;
+ V::range<2> r2( v ); r2 = 1,7,2;
+
+ v.reset( a );
+
+The call to `reset` in this example constructs a three-dimensional `view` with shape
+3 x 3 x 3 whose data elements store the addresses of every second data element in the third
+dimension of every second array in the second dimension of every second array of array in
+the first dimension of the 7 x 7 x 7 `array_ a`.
+
+The header `<view_.hpp>` defines the class template,
+
+ template< class A, class R, bool B > class view_;
+
+Here,
+
+* `A` is the array type the `view_` will be constructed from.
+* `R` is a sequence of ranges.
+* `B` signifies our intention to store pointers on an argument of type `A` or otherwise.
+
+A `view_` is used to construct a view of a fixed-size array. The data member of a `view_`
+will be statically or dynamically allocated dependant on the type of the data member of
+the array type `A`. `A` and `R` are both required parameters with `B` optional defaulting
+to `true`. A `view_` type is more restricted than a `view` in the sense that the parameter
+`R` defines at compile time the ranges over an array of type `A` associated with the
+`view_`. Template types `rangesN`, `N` the dimensionality of an array of type `A`, and
+`range` are defined specifically for the template argument `R`. Since the range's are
+predetermined the `view_` type defines a constructor accepting an argument of type `A`.
+Thus, we can write,
+
+ typedef array_< double, 3,7, true > A;
+ typedef ranges3< range< 1,7,2 >, range< 1,7,2 >, range< 1,7,2 > > R;
+ typedef view_< A, R > V;
+
+ A a;
+ V v( a );
+
+
+[endsect] [/ Array Views]
+
+[section:scalars_vectors_matrices Scalars, Vectors and Matrices]
+
+In certain problem domains we can specify a set of integral dimensions which fully
+characterize the problem space. Fixed-size matrices for example are common in many
+algebraic problems. Knowing the initial dimensions in advance allows us to exploit this
+knowledge in all future computations. For example, given a matrix, M1, with M rows and P
+columns and a matrix, M2, with P rows and N columns, the product M1 x M2 is a matrix, M3 say,
+with M rows and N columns. Computational requirements such as the operand and result type
+dimensions in this example can be encoded in the template parameters of the types involved
+and also in the operators that act on them. The Boost maps library provides implementations
+for `scalar` and fixed-size `vector` and `matrix` operators that conform to such
+requirements, enabling compile time determination of consistent operations on the operands
+involved. The expression template paradigm has been invoked to enhance performance for
+mathematically intensive applications. Note that `scalar` operands are a convenient way to
+specify operations involving scalar types and are not intended to be used as the data
+elements of the `vector` or `matrix` types. The headers `<scalar.hpp>`, `<vector.hpp>` and
+`<matrix.hpp>` define the following respective types.
+
+ template< class T > class scalar;
+
+Here,
+
+* `T` is the data type of the stored member.
+
+ template< class T, std::size_t N, bool I, class A > class vector;
+
+Here,
+
+* `T` is the element type.
+* `N` is the number of components.
+* `I` is the zero/default initialization parameter.
+* `A` is an allocator type.
+
+ template< class T, std::size_t M, std::size_t N, bool I, class A > class matrix;
+
+Here,
+
+* `T` is the element type.
+* `M` is the number of rows.
+* `N` is the number of columns.
+* `I` is the zero/default initialization parameter.
+* `A` is an allocator type.
+
+For all three class templates `T` is a required parameter. The parameters `N` for a `vector`
+and, `M` and `N` for a `matrix` are also required. The remaining parameters `I` and `A` for
+the `vector` and `matrix` class templates are optional with respective default's `true` and
+`null::allocator` for both classes.
+
+[endsect] [/ Scalars, Vectors and Matrices]
+
+[endsect] [/ Concepts]
+
+[section:tutorial Tutorial]
+
+[section:array_tutorial Arrays]
+
+The intention in this section is to simply reveal the functionality of the multi-dimensional
+array classes. The dominant feature is element access so we'll consider the various methods
+available. The following example declares a statically allocated, uninitialized `array`, `a`,
+then loops over each dimension setting the value of each data element to the sum of its indexes.
+
+ typedef array< int, bounds5< 5,4,3,4,5 > > array_t;
+
+ array_t a;
+
+ for( std::size_t i = 0; i != a.bound<0>(); ++i )
+ for( std::size_t j = 0; j != a.bound<1>(); ++j )
+ for( std::size_t k = 0; k != a.bound<2>(); ++k )
+ for( std::size_t l = 0; l != a.bound<3>(); ++l )
+ for( std::size_t m = 0; m != a.bound<4>(); ++m )
+ a[i][j][k][l][m] = i + j + k + l + m;
+
+Alternatively, in place of `a[i][j][k][l][m]` we can use indirection as follows,
+`*(*(*(*(*(a+i)+j)+k)+l)+m)`. No check with either of these methods is made to ensure the
+indexes are within the bounds of the array. The `bound` method however returns the bound at each
+dimension and includes a compile time assertion on its argument. Note that this method employs
+the same zero-based indexing semantics the language provides for indexing array or pointer types.
+If a debug assertion on the indexes is required we can write the following,
+
+ bounds<5> b;
+
+ for( std::size_t i = 0; i != a.bound<0>(); ++i )
+ for( std::size_t j = 0; j != a.bound<1>(); ++j )
+ for( std::size_t k = 0; k != a.bound<2>(); ++k )
+ for( std::size_t l = 0; l != a.bound<3>(); ++l )
+ for( std::size_t m = 0; m != a.bound<4>(); ++m )
+ a[b = i,j,k,l,m] += 1;
+
+If the indexes should always be checked then replacing `a[b = i,j,k,l,m]` with
+`a.at((b = i,j,k,l,m))` will throw a `std::out_of_range` exception on an invalid index.
+
+When a function or operator is to be applied on all data elements in the array we can do
+something along the following lines,
+
+ std::for_each( a.begin(), a.end(), boost::lambda::_1 -= 1 );
+
+The `view` class template can used to access non-continuous sequences of data elements as
+follows,
+
+ typedef view< array_t > view_t;
+
+ view_t v;
+
+ view_t::range<0> r0( v ); r0 = 1,4,1;
+ view_t::range<1> r1( v ); r1 = 0,2,1;
+ view_t::range<2> r2( v ); r2 = 0,3,1;
+ view_t::range<3> r3( v ); r3 = 2,4,1;
+ view_t::range<4> r4( v ); r4 = 1,4,1;
+
+ v.reset( a );
+
+ for( std::size_t i = 0; i != v.bound<0>(); ++i )
+ for( std::size_t j = 0; j != v.bound<1>(); ++j )
+ for( std::size_t k = 0; k != v.bound<2>(); ++k )
+ for( std::size_t l = 0; l != v.bound<3>(); ++l )
+ for( std::size_t m = 0; m != v.bound<4>(); ++m )
+ *v[i][j][k][l][m] *= 1;
+
+ r1 = 2,4,1; r3 = 0,2,1;
+ v.reset( a );
+
+ std::for_each( v.begin(), v.end(), *boost::lambda::_1 *= 1 );
+
+Given that `v` represents a 3 x 2 x 3 x 2 x 3 array and that its values are the addresses of
+elements in the array `a`, we can set the values of any array whose dimensions are the same as
+those of `v` to the corresponding values in `a`.
+
+ b = 3,2,3,2,3;
+ pointer< int, 5 > p( b );
+ v.to( p );
+
+Similarly, again provided we store the addresses of a sequence of values from the representative
+array, we can set those values to the values in an array with the same dimensions as the `view`,
+so that given an array `q` with shape 3 x 2 x 3 x 2 x 3 we can write, `v.from( q )`.
+
+[endsect] [/ Arrays]
+
+[section:scalar_vector_matrix_tutorial Scalars, Vectors and Matrices]
+
+At present, only dense vectors and matrices with associated operations are available. The
+following gives some examples.
+
+ matrix< int, 2,2 > m1, m2;
+ vector< int, 2 > r, v, v1, v2;
+ scalar< int > s;
+
+ s = 3;
+
+ for( int i = 0; i != 2; ++i )
+ {
+ v[i] = i + 3;
+ v1[i] = i + 1;
+ v2[i] = i + 2;
+
+ for( int j = 0; j != 2; ++j )
+ {
+ m1[i][j] = i + j + 1;
+ m2[i][j] = i + j + 2;
+ }
+ }
+
+ r = outer_product( m1 * v1, v2 * m2 ) * ( s * v );
+
+No distinction is made between row and column vectors hence pre and post multiplication of a
+matrix by the same vector is possible provided the matrix is square or otherwise whenever the
+number of vector components equals the number of matrix rows for pre-multiplication or the
+number of matrix columns for post-multiplication. For n matrices, n > 2, it is more efficient
+to produce temporary intermediate results when they are multiplied in succession using the `*`
+operator. Determining when such a chain of multiplications are being applied has proven to be
+inefficient thus a temporary matrix is produced for the case n = 2 also. In order to avoid this
+the `product` operator can be applied as shown below.
+
+ const unsigned M = 1000;
+
+ matrix< int, M,M, true, std::allocator< int > > m, m1, m2, m3;
+
+ for( int i = 0; i != M; ++i )
+ for( int j = 0; j != M; ++j )
+ {
+ m1( i,j ) = 1;
+ m2( i,j ) = 2;
+ m3( i,j ) = 3;
+ }
+
+ m = product( m1, m2 ) + m3;
+
+For some system dependant value of M the number of cache misses will dramatically reduce the
+performance of the above calculation. In such cases we can reorganize the computation as below,
+although setting up the matrices is not always so easily accomplished.
+
+ const unsigned N = 10;
+
+ matrix< matrix< matrix< int,
+ N,N, true, std::allocator< int > >,
+ N,N, true, std::allocator< int > >,
+ N,N, true, std::allocator< int > > m, m1, m2, m3;
+
+ for( int i = 0; i != N; ++i )
+ for( int j = 0; j != N; ++j )
+ {
+ for( int k = 0; k != N; ++k )
+ for( int l = 0; l != N; ++l )
+ {
+ for( int m = 0; m != N; ++m )
+ for( int n = 0; n != N; ++n )
+ {
+ m1( i,j )( k,l )( m,n ) = 1;
+ m2( i,j )( k,l )( m,n ) = 2;
+ m3( i,j )( k,l )( m,n ) = 3;
+ }
+ }
+ }
+
+ m = product( m1, m2 ) + m3;
+
+Likewise, we can set up blocked computations with non-square matrices.
+
+ matrix< matrix< int, 3,2 >, 2,3 > m1;
+ matrix< matrix< int, 3,4 >, 2,3 > m2;
+ matrix< matrix< int, 2,4 >, 3,3 > m, m3;
+
+ for( std::size_t i = 0; i != 2; ++i )
+ for( std::size_t j = 0; j != 3; ++j )
+ for( std::size_t k = 0; k != 3; ++k )
+ for( std::size_t l = 0; l != 2; ++l )
+ m1[i][j][k][l] = i + j + k + l;
+
+ for( std::size_t i = 0; i != 2; ++i )
+ for( std::size_t j = 0; j != 3; ++j )
+ for( std::size_t k = 0; k != 3; ++k )
+ for( std::size_t l = 0; l != 4; ++l )
+ m2[i][j][k][l] = i + j + k + l;
+
+ for( std::size_t i = 0; i != 3; ++i )
+ for( std::size_t j = 0; j != 3; ++j )
+ for( std::size_t k = 0; k != 2; ++k )
+ for( std::size_t l = 0; l != 4; ++l )
+ m3[i][j][k][l] = i + j + k + l;
+
+ m = transpose( m1 ) * m2 + m3;
+
+Or similarly with a combination of vectors and matrices.
+
+ vector< vector< int, 3 >, 2 > v1, v2;
+ matrix< matrix< int, 3,4 >, 2,3 > m, m1, m2;
+
+ v1[0][0] = 1; v1[0][1] = 2; v1[0][2] = 3;
+ v1[1][0] = 4; v1[1][1] = 5; v1[1][2] = 6;
+
+ v2[0][0] = 6; v2[0][1] = 5; v2[0][2] = 4;
+ v2[1][0] = 3; v2[1][1] = 2; v2[1][2] = 1;
+
+ for( std::size_t i = 0; i != 2; ++i )
+ for( std::size_t j = 0; j != 3; ++j )
+ for( std::size_t k = 0; k != 2; ++k )
+ for( std::size_t l = 0; l != 4; ++l )
+ m1[i][j][k][l] = i + j + k + l;
+
+ for( std::size_t i = 0; i != 2; ++i )
+ for( std::size_t j = 0; j != 3; ++j )
+ for( std::size_t k = 0; k != 2; ++k )
+ for( std::size_t l = 0; l != 4; ++l )
+ m2[i][j][k][l] = i + j + k + l;
+
+ m = transpose( outer_product( v1, v2 ) ) * m1 + m2;
+
+[endsect] [/ Scalars, Vectors and Matrices]
+
+[endsect] [/ Tutorial]
+
+[section:reference Reference]
+
+*[link maps.array array]
+*[link maps.array_ array_]
+*[link maps.pointer pointer]
+*[link maps.view view]
+*[link maps.view_ view_]
+
+[endsect] [/ Reference]
+
+
+[section:array array]
+
+[endsect] [/ array]
+
+[section:array_ array_]
+
+[endsect] [/ array_]
+
+[section:pointer pointer]
+
+[endsect] [/ pointer]
+
+[section:view view]
+
+[endsect] [/ view]
+
+[section:view_ view_]
+
+[endsect] [/ view_]

Next message: joel_at_[hidden]: "[Boost-commit] svn:boost r74033 - trunk/libs/spirit/test/qi"
Previous message: bjs3141_at_[hidden]: "[Boost-commit] svn:boost r74031 - in sandbox/maps/boost: . maps maps/generic maps/math maps/support"

Date view	Thread view	Subject view	Author view

Boost-Commit list run by bdawes at acm.org, david.abrahams at rcn.com, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk