3 <meta http-equiv="Content-Type" content="text/html; charset=US-ASCII">
4 <title>Exponential Integral Ei</title>
5 <link rel="stylesheet" href="../../math.css" type="text/css">
6 <meta name="generator" content="DocBook XSL Stylesheets V1.79.1">
7 <link rel="home" href="../../index.html" title="Math Toolkit 2.11.0">
8 <link rel="up" href="../expint.html" title="Exponential Integrals">
9 <link rel="prev" href="expint_n.html" title="Exponential Integral En">
10 <link rel="next" href="../hypergeometric.html" title="Hypergeometric Functions">
12 <body bgcolor="white" text="black" link="#0000FF" vlink="#840084" alink="#0000FF">
13 <table cellpadding="2" width="100%"><tr>
14 <td valign="top"><img alt="Boost C++ Libraries" width="277" height="86" src="../../../../../../boost.png"></td>
15 <td align="center"><a href="../../../../../../index.html">Home</a></td>
16 <td align="center"><a href="../../../../../../libs/libraries.htm">Libraries</a></td>
17 <td align="center"><a href="http://www.boost.org/users/people.html">People</a></td>
18 <td align="center"><a href="http://www.boost.org/users/faq.html">FAQ</a></td>
19 <td align="center"><a href="../../../../../../more/index.htm">More</a></td>
22 <div class="spirit-nav">
23 <a accesskey="p" href="expint_n.html"><img src="../../../../../../doc/src/images/prev.png" alt="Prev"></a><a accesskey="u" href="../expint.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="../hypergeometric.html"><img src="../../../../../../doc/src/images/next.png" alt="Next"></a>
26 <div class="titlepage"><div><div><h3 class="title">
27 <a name="math_toolkit.expint.expint_i"></a><a class="link" href="expint_i.html" title="Exponential Integral Ei">Exponential Integral Ei</a>
28 </h3></div></div></div>
30 <a name="math_toolkit.expint.expint_i.h0"></a>
31 <span class="phrase"><a name="math_toolkit.expint.expint_i.synopsis"></a></span><a class="link" href="expint_i.html#math_toolkit.expint.expint_i.synopsis">Synopsis</a>
33 <pre class="programlisting"><span class="preprocessor">#include</span> <span class="special"><</span><span class="identifier">boost</span><span class="special">/</span><span class="identifier">math</span><span class="special">/</span><span class="identifier">special_functions</span><span class="special">/</span><span class="identifier">expint</span><span class="special">.</span><span class="identifier">hpp</span><span class="special">></span>
35 <pre class="programlisting"><span class="keyword">namespace</span> <span class="identifier">boost</span><span class="special">{</span> <span class="keyword">namespace</span> <span class="identifier">math</span><span class="special">{</span>
37 <span class="keyword">template</span> <span class="special"><</span><span class="keyword">class</span> <span class="identifier">T</span><span class="special">></span>
38 <a class="link" href="../result_type.html" title="Calculation of the Type of the Result"><span class="emphasis"><em>calculated-result-type</em></span></a> <span class="identifier">expint</span><span class="special">(</span><span class="identifier">T</span> <span class="identifier">z</span><span class="special">);</span>
40 <span class="keyword">template</span> <span class="special"><</span><span class="keyword">class</span> <span class="identifier">T</span><span class="special">,</span> <span class="keyword">class</span> <a class="link" href="../../policy.html" title="Chapter 20. Policies: Controlling Precision, Error Handling etc">Policy</a><span class="special">></span>
41 <a class="link" href="../result_type.html" title="Calculation of the Type of the Result"><span class="emphasis"><em>calculated-result-type</em></span></a> <span class="identifier">expint</span><span class="special">(</span><span class="identifier">T</span> <span class="identifier">z</span><span class="special">,</span> <span class="keyword">const</span> <a class="link" href="../../policy.html" title="Chapter 20. Policies: Controlling Precision, Error Handling etc">Policy</a><span class="special">&);</span>
43 <span class="special">}}</span> <span class="comment">// namespaces</span>
46 The return type of these functions is computed using the <a class="link" href="../result_type.html" title="Calculation of the Type of the Result"><span class="emphasis"><em>result
47 type calculation rules</em></span></a>: the return type is <code class="computeroutput"><span class="keyword">double</span></code> if T is an integer type, and T otherwise.
50 The final <a class="link" href="../../policy.html" title="Chapter 20. Policies: Controlling Precision, Error Handling etc">Policy</a> argument is optional and can
51 be used to control the behaviour of the function: how it handles errors,
52 what level of precision to use etc. Refer to the <a class="link" href="../../policy.html" title="Chapter 20. Policies: Controlling Precision, Error Handling etc">policy
53 documentation for more details</a>.
56 <a name="math_toolkit.expint.expint_i.h1"></a>
57 <span class="phrase"><a name="math_toolkit.expint.expint_i.description"></a></span><a class="link" href="expint_i.html#math_toolkit.expint.expint_i.description">Description</a>
59 <pre class="programlisting"><span class="keyword">template</span> <span class="special"><</span><span class="keyword">class</span> <span class="identifier">T</span><span class="special">></span>
60 <a class="link" href="../result_type.html" title="Calculation of the Type of the Result"><span class="emphasis"><em>calculated-result-type</em></span></a> <span class="identifier">expint</span><span class="special">(</span><span class="identifier">T</span> <span class="identifier">z</span><span class="special">);</span>
62 <span class="keyword">template</span> <span class="special"><</span><span class="keyword">class</span> <span class="identifier">T</span><span class="special">,</span> <span class="keyword">class</span> <a class="link" href="../../policy.html" title="Chapter 20. Policies: Controlling Precision, Error Handling etc">Policy</a><span class="special">></span>
63 <a class="link" href="../result_type.html" title="Calculation of the Type of the Result"><span class="emphasis"><em>calculated-result-type</em></span></a> <span class="identifier">expint</span><span class="special">(</span><span class="identifier">T</span> <span class="identifier">z</span><span class="special">,</span> <span class="keyword">const</span> <a class="link" href="../../policy.html" title="Chapter 20. Policies: Controlling Precision, Error Handling etc">Policy</a><span class="special">&);</span>
66 Returns the <a href="http://mathworld.wolfram.com/ExponentialIntegral.html" target="_top">exponential
69 <div class="blockquote"><blockquote class="blockquote"><p>
70 <span class="inlinemediaobject"><img src="../../../equations/expint_i_1.svg"></span>
72 </p></blockquote></div>
73 <div class="blockquote"><blockquote class="blockquote"><p>
74 <span class="inlinemediaobject"><img src="../../../graphs/expint_i.svg" align="middle"></span>
76 </p></blockquote></div>
78 <a name="math_toolkit.expint.expint_i.h2"></a>
79 <span class="phrase"><a name="math_toolkit.expint.expint_i.accuracy"></a></span><a class="link" href="expint_i.html#math_toolkit.expint.expint_i.accuracy">Accuracy</a>
82 The following table shows the peak errors (in units of epsilon) found on
83 various platforms with various floating point types, along with comparisons
84 to Cody's SPECFUN implementation and the <a href="http://www.gnu.org/software/gsl/" target="_top">GSL-1.9</a>
85 library. Unless otherwise specified any floating point type that is narrower
86 than the one shown will have <a class="link" href="../relative_error.html#math_toolkit.relative_error.zero_error">effectively
90 <a name="math_toolkit.expint.expint_i.table_expint_Ei_"></a><p class="title"><b>Table 8.78. Error rates for expint (Ei)</b></p>
91 <div class="table-contents"><table class="table" summary="Error rates for expint (Ei)">
104 GNU C++ version 7.1.0<br> linux<br> long double
109 GNU C++ version 7.1.0<br> linux<br> double
114 Sun compiler version 0x5150<br> Sun Solaris<br> long double
119 Microsoft Visual C++ version 14.1<br> Win32<br> double
127 Exponential Integral Ei
132 <span class="blue">Max = 5.05ε (Mean = 0.821ε)</span><br> <br>
133 (<span class="emphasis"><em><cmath>:</em></span> Max = 14.1ε (Mean = 2.43ε) <a class="link" href="../logs_and_tables/logs.html#errors_GNU_C_version_7_1_0_linux_long_double_expint_Ei___cmath__Exponential_Integral_Ei">And
139 <span class="blue">Max = 0.994ε (Mean = 0.142ε)</span><br> <br>
140 (<span class="emphasis"><em>GSL 2.1:</em></span> Max = 8.96ε (Mean = 0.703ε))
145 <span class="blue">Max = 5.05ε (Mean = 0.835ε)</span>
150 <span class="blue">Max = 1.43ε (Mean = 0.54ε)</span>
157 Exponential Integral Ei: double exponent range
162 <span class="blue">Max = 1.72ε (Mean = 0.593ε)</span><br> <br>
163 (<span class="emphasis"><em><cmath>:</em></span> Max = 3.11ε (Mean = 1.13ε))
168 <span class="blue">Max = 0.998ε (Mean = 0.156ε)</span><br> <br>
169 (<span class="emphasis"><em>GSL 2.1:</em></span> Max = 1.5ε (Mean = 0.612ε))
174 <span class="blue">Max = 1.72ε (Mean = 0.607ε)</span>
179 <span class="blue">Max = 1.7ε (Mean = 0.66ε)</span>
186 Exponential Integral Ei: long exponent range
191 <span class="blue">Max = 1.98ε (Mean = 0.595ε)</span><br> <br>
192 (<span class="emphasis"><em><cmath>:</em></span> Max = 1.93ε (Mean = 0.855ε))
199 <span class="blue">Max = 1.98ε (Mean = 0.575ε)</span>
208 <br class="table-break"><p>
209 It should be noted that all three libraries tested above offer sub-epsilon
210 precision over most of their range.
213 GSL has the greatest difficulty near the positive root of En, while Cody's
214 SPECFUN along with this implementation increase their error rates very slightly
215 over the range [4,6].
218 The following error plot are based on an exhaustive search of the functions
219 domain, MSVC-15.5 at <code class="computeroutput"><span class="keyword">double</span></code>
220 precision, and GCC-7.1/Ubuntu for <code class="computeroutput"><span class="keyword">long</span>
221 <span class="keyword">double</span></code> and <code class="computeroutput"><span class="identifier">__float128</span></code>.
223 <div class="blockquote"><blockquote class="blockquote"><p>
224 <span class="inlinemediaobject"><img src="../../../graphs/exponential_integral_ei__double.svg" align="middle"></span>
226 </p></blockquote></div>
227 <div class="blockquote"><blockquote class="blockquote"><p>
228 <span class="inlinemediaobject"><img src="../../../graphs/exponential_integral_ei__80_bit_long_double.svg" align="middle"></span>
230 </p></blockquote></div>
231 <div class="blockquote"><blockquote class="blockquote"><p>
232 <span class="inlinemediaobject"><img src="../../../graphs/exponential_integral_ei____float128.svg" align="middle"></span>
234 </p></blockquote></div>
236 <a name="math_toolkit.expint.expint_i.h3"></a>
237 <span class="phrase"><a name="math_toolkit.expint.expint_i.testing"></a></span><a class="link" href="expint_i.html#math_toolkit.expint.expint_i.testing">Testing</a>
240 The tests for these functions come in two parts: basic sanity checks use
241 spot values calculated using <a href="http://functions.wolfram.com/webMathematica/FunctionEvaluation.jsp?name=ExpIntegralEi" target="_top">Mathworld's
242 online evaluator</a>, while accuracy checks use high-precision test values
243 calculated at 1000-bit precision with <a href="http://shoup.net/ntl/doc/RR.txt" target="_top">NTL::RR</a>
244 and this implementation. Note that the generic and type-specific versions
245 of these functions use differing implementations internally, so this gives
246 us reasonably independent test data. Using our test data to test other "known
247 good" implementations also provides an additional sanity check.
250 <a name="math_toolkit.expint.expint_i.h4"></a>
251 <span class="phrase"><a name="math_toolkit.expint.expint_i.implementation"></a></span><a class="link" href="expint_i.html#math_toolkit.expint.expint_i.implementation">Implementation</a>
254 For x < 0 this function just calls <a class="link" href="expint_n.html" title="Exponential Integral En">zeta</a>(1,
255 -x): which in turn is implemented in terms of rational approximations when
256 the type of x has 113 or fewer bits of precision.
259 For x > 0 the generic version is implemented using the infinte series:
261 <div class="blockquote"><blockquote class="blockquote"><p>
262 <span class="inlinemediaobject"><img src="../../../equations/expint_i_2.svg"></span>
264 </p></blockquote></div>
266 However, when the precision of the argument type is known at compile time
267 and is 113 bits or less, then rational approximations <a class="link" href="../sf_implementation.html#math_toolkit.sf_implementation.rational_approximations_used">devised
271 For 0 < z < 6 a root-preserving approximation of the form:
273 <div class="blockquote"><blockquote class="blockquote"><p>
274 <span class="inlinemediaobject"><img src="../../../equations/expint_i_3.svg"></span>
276 </p></blockquote></div>
278 is used, where z<sub>0</sub> is the positive root of the function, and R(z/3 - 1) is
279 a minimax rational approximation rescaled so that it is evaluated over [-1,1].
280 Note that while the rational approximation over [0,6] converges rapidly to
281 the minimax solution it is rather ill-conditioned in practice. Cody and Thacher
282 <a href="#ftn.math_toolkit.expint.expint_i.f0" class="footnote" name="math_toolkit.expint.expint_i.f0"><sup class="footnote">[5]</sup></a> experienced the same issue and converted the polynomials into
283 Chebeshev form to ensure stable computation. By experiment we found that
284 the polynomials are just as stable in polynomial as Chebyshev form, <span class="emphasis"><em>provided</em></span>
285 they are computed over the interval [-1,1].
288 Over the a series of intervals <span class="emphasis"><em>[a, b]</em></span> and <span class="emphasis"><em>[b,
289 INF]</em></span> the rational approximation takes the form:
291 <div class="blockquote"><blockquote class="blockquote"><p>
292 <span class="inlinemediaobject"><img src="../../../equations/expint_i_4.svg"></span>
294 </p></blockquote></div>
296 where <span class="emphasis"><em>c</em></span> is a constant, and <span class="emphasis"><em>R(t)</em></span>
297 is a minimax solution optimised for low absolute error compared to <span class="emphasis"><em>c</em></span>.
298 Variable <span class="emphasis"><em>t</em></span> is <code class="computeroutput"><span class="number">1</span><span class="special">/</span><span class="identifier">z</span></code> when
299 the range in infinite and <code class="computeroutput"><span class="number">2</span><span class="identifier">z</span><span class="special">/(</span><span class="identifier">b</span><span class="special">-</span><span class="identifier">a</span><span class="special">)</span> <span class="special">-</span>
300 <span class="special">(</span><span class="number">2</span><span class="identifier">a</span><span class="special">/(</span><span class="identifier">b</span><span class="special">-</span><span class="identifier">a</span><span class="special">)</span>
301 <span class="special">+</span> <span class="number">1</span><span class="special">)</span></code> otherwise: this has the effect of scaling
302 z to the interval [-1,1]. As before rational approximations over arbitrary
303 intervals were found to be ill-conditioned: Cody and Thacher solved this
304 issue by converting the polynomials to their J-Fraction equivalent. However,
305 as long as the interval of evaluation was [-1,1] and the number of terms
306 carefully chosen, it was found that the polynomials <span class="emphasis"><em>could</em></span>
307 be evaluated to suitable precision: error rates are typically 2 to 3 epsilon
308 which is comparible to the error rate that Cody and Thacher achieved using
309 J-Fractions, but marginally more efficient given that fewer divisions are
312 <div class="footnotes">
313 <br><hr style="width:100; text-align:left;margin-left: 0">
314 <div id="ftn.math_toolkit.expint.expint_i.f0" class="footnote"><p><a href="#math_toolkit.expint.expint_i.f0" class="para"><sup class="para">[5] </sup></a>
315 W. J. Cody and H. C. Thacher, Jr., Rational Chebyshev approximations for
316 the exponential integral E<sub>1</sub>(x), Math. Comp. 22 (1968), 641-649, and W.
317 J. Cody and H. C. Thacher, Jr., Chebyshev approximations for the exponential
318 integral Ei(x), Math. Comp. 23 (1969), 289-303.
322 <table xmlns:rev="http://www.cs.rpi.edu/~gregod/boost/tools/doc/revision" width="100%"><tr>
323 <td align="left"></td>
324 <td align="right"><div class="copyright-footer">Copyright © 2006-2019 Nikhar
325 Agrawal, Anton Bikineev, Paul A. Bristow, Marco Guazzone, Christopher Kormanyos,
326 Hubert Holin, Bruno Lalande, John Maddock, Jeremy Murphy, Matthew Pulver, Johan
327 Råde, Gautam Sewani, Benjamin Sobotta, Nicholas Thompson, Thijs van den Berg,
328 Daryle Walker and Xiaogang Zhang<p>
329 Distributed under the Boost Software License, Version 1.0. (See accompanying
330 file LICENSE_1_0.txt or copy at <a href="http://www.boost.org/LICENSE_1_0.txt" target="_top">http://www.boost.org/LICENSE_1_0.txt</a>)
335 <div class="spirit-nav">
336 <a accesskey="p" href="expint_n.html"><img src="../../../../../../doc/src/images/prev.png" alt="Prev"></a><a accesskey="u" href="../expint.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="../hypergeometric.html"><img src="../../../../../../doc/src/images/next.png" alt="Next"></a>