[ceph.git] / ceph / src / boost / libs / math / doc / sf / erf.qbk

[section:error_function Error Functions]

[h4 Synopsis]

``
#include <boost/math/special_functions/erf.hpp>
``

   namespace boost{ namespace math{
   
   template <class T>
   ``__sf_result`` erf(T z);
   
   template <class T, class ``__Policy``>
   ``__sf_result`` erf(T z, const ``__Policy``&);
   
   template <class T>
   ``__sf_result`` erfc(T z);
   
   template <class T, class ``__Policy``>
   ``__sf_result`` erfc(T z, const ``__Policy``&);
   
   }} // namespaces
   
The return type of these functions is computed using the __arg_promotion_rules:
the return type is `double` if T is an integer type, and T otherwise.

[optional_policy]

[h4 Description]

   template <class T>
   ``__sf_result`` erf(T z);
   
   template <class T, class ``__Policy``>
   ``__sf_result`` erf(T z, const ``__Policy``&);
   
Returns the [@http://en.wikipedia.org/wiki/Error_function error function]
[@http://functions.wolfram.com/GammaBetaErf/Erf/ erf] of z:

[equation erf1]

[graph erf]

   template <class T>
   ``__sf_result`` erfc(T z);
   
   template <class T, class ``__Policy``>
   ``__sf_result`` erfc(T z, const ``__Policy``&);
   
Returns the complement of the [@http://functions.wolfram.com/GammaBetaErf/Erfc/ error function] of z:

[equation erf2]

[graph erfc]

[h4 Accuracy]

The following table shows the peak errors (in units of epsilon) 
found on various platforms with various floating point types, 
along with comparisons to the __gsl, __glibc, __hpc and __cephes libraries.
Unless otherwise specified any floating point type that is narrower
than the one shown will have __zero_error.

[table_erf]

[table_erfc]

[h4 Testing]

The tests for these functions come in two parts:
basic sanity checks use spot values calculated using
[@http://functions.wolfram.com/webMathematica/FunctionEvaluation.jsp?name=Erf Mathworld's online evaluator],
while accuracy checks use high-precision test values calculated at 1000-bit precision with
[@http://shoup.net/ntl/doc/RR.txt NTL::RR] and this implementation. 
Note that the generic and type-specific
versions of these functions use differing implementations internally, so this
gives us reasonably independent test data.  Using our test data to test other
"known good" implementations also provides an additional sanity check. 

[h4 Implementation]

All versions of these functions first use the usual reflection formulas
to make their arguments positive:

   erf(-z) = 1 - erf(z);
   
   erfc(-z) = 2 - erfc(z);  // preferred when -z < -0.5
   
   erfc(-z) = 1 + erf(z);   // preferred when -0.5 <= -z < 0

The generic versions of these functions are implemented in terms of
the incomplete gamma function.

When the significand (mantissa) size is recognised
(currently for 53, 64 and 113-bit reals, plus single-precision 24-bit handled via promotion to double)
then a series of rational approximations [jm_rationals] are used.

For `z <= 0.5` then a rational approximation to erf is used, based on the 
observation that erf is an odd function and therefore erf is calculated using:

   erf(z) = z * (C + R(z*z));
   
where the rational approximation R(z*z) is optimised for absolute error:
as long as its absolute error is small enough compared to the constant C, then any 
round-off error incurred during the computation of R(z*z) will effectively 
disappear from the result.  As a result the error for erf and erfc in this
region is very low: the last bit is incorrect in only a very small number of 
cases.

For `z > 0.5` we observe that over a small interval \[a, b) then:

   erfc(z) * exp(z*z) * z ~ c
   
for some constant c.

Therefore for `z > 0.5` we calculate erfc using:

   erfc(z) = exp(-z*z) * (C + R(z - B)) / z;
   
Again R(z - B) is optimised for absolute error, and the constant `C` is
the average of `erfc(z) * exp(z*z) * z` taken at the endpoints of the range.
Once again, as long as the absolute error in R(z - B) is small
compared to `c` then `c + R(z - B)` will be correctly rounded, and the error
in the result will depend only on the accuracy of the exp function.  In practice,
in all but a very small number of cases, the error is confined to the last bit
of the result.  The constant `B` is chosen so that the left hand end of the range
of the rational approximation is 0.

For large `z` over a range \[a, +[infin]\] the above approximation is modified to:

   erfc(z) = exp(-z*z) * (C + R(1 / z)) / z;

[endsect]
[/ :error_function The Error Functions]

[/ 
  Copyright 2006 John Maddock and Paul A. Bristow.
  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).
]
Commit	Line	Data
7c673cae FG	1	[section:error_function Error Functions]
	2
	3	[h4 Synopsis]
	4
	5	``
	6	#include <boost/math/special_functions/erf.hpp>
	7	``
	8
	9	namespace boost{ namespace math{
	10
	11	template <class T>
	12	``__sf_result`` erf(T z);
	13
	14	template <class T, class ``__Policy``>
	15	``__sf_result`` erf(T z, const ``__Policy``&);
	16
	17	template <class T>
	18	``__sf_result`` erfc(T z);
	19
	20	template <class T, class ``__Policy``>
	21	``__sf_result`` erfc(T z, const ``__Policy``&);
	22
	23	}} // namespaces
	24
	25	The return type of these functions is computed using the __arg_promotion_rules:
	26	the return type is `double` if T is an integer type, and T otherwise.
	27
	28	[optional_policy]
	29
	30	[h4 Description]
	31
	32	template <class T>
	33	``__sf_result`` erf(T z);
	34
	35	template <class T, class ``__Policy``>
	36	``__sf_result`` erf(T z, const ``__Policy``&);
	37
	38	Returns the [@http://en.wikipedia.org/wiki/Error_function error function]
	39	[@http://functions.wolfram.com/GammaBetaErf/Erf/ erf] of z:
	40
	41	[equation erf1]
	42
	43	[graph erf]
	44
	45	template <class T>
	46	``__sf_result`` erfc(T z);
	47
	48	template <class T, class ``__Policy``>
	49	``__sf_result`` erfc(T z, const ``__Policy``&);
	50
	51	Returns the complement of the [@http://functions.wolfram.com/GammaBetaErf/Erfc/ error function] of z:
	52
	53	[equation erf2]
	54
	55	[graph erfc]
	56
	57	[h4 Accuracy]
	58
	59	The following table shows the peak errors (in units of epsilon)
	60	found on various platforms with various floating point types,
	61	along with comparisons to the __gsl, __glibc, __hpc and __cephes libraries.
	62	Unless otherwise specified any floating point type that is narrower
	63	than the one shown will have __zero_error.
	64
65	[table_erf]
66
67	[table_erfc]
68
69	[h4 Testing]
70
71	The tests for these functions come in two parts:
72	basic sanity checks use spot values calculated using
73	[@http://functions.wolfram.com/webMathematica/FunctionEvaluation.jsp?name=Erf Mathworld's online evaluator],
74	while accuracy checks use high-precision test values calculated at 1000-bit precision with
75	[@http://shoup.net/ntl/doc/RR.txt NTL::RR] and this implementation.
76	Note that the generic and type-specific
77	versions of these functions use differing implementations internally, so this
78	gives us reasonably independent test data. Using our test data to test other
79	"known good" implementations also provides an additional sanity check.
80
81	[h4 Implementation]
82
83	All versions of these functions first use the usual reflection formulas
84	to make their arguments positive:
85
86	erf(-z) = 1 - erf(z);
87
88	erfc(-z) = 2 - erfc(z); // preferred when -z < -0.5
89
90	erfc(-z) = 1 + erf(z); // preferred when -0.5 <= -z < 0
91
92	The generic versions of these functions are implemented in terms of
93	the incomplete gamma function.
94
95	When the significand (mantissa) size is recognised
96	(currently for 53, 64 and 113-bit reals, plus single-precision 24-bit handled via promotion to double)
97	then a series of rational approximations [jm_rationals] are used.
98
99	For `z <= 0.5` then a rational approximation to erf is used, based on the
100	observation that erf is an odd function and therefore erf is calculated using:
101
102	erf(z) = z * (C + R(z*z));
103
104	where the rational approximation R(z*z) is optimised for absolute error:
105	as long as its absolute error is small enough compared to the constant C, then any
106	round-off error incurred during the computation of R(z*z) will effectively
107	disappear from the result. As a result the error for erf and erfc in this
108	region is very low: the last bit is incorrect in only a very small number of
109	cases.
110
111	For `z > 0.5` we observe that over a small interval \[a, b) then:
112
113	erfc(z) * exp(zz) z ~ c
114
115	for some constant c.
116
117	Therefore for `z > 0.5` we calculate erfc using:
118
119	erfc(z) = exp(-zz) (C + R(z - B)) / z;
120
121	Again R(z - B) is optimised for absolute error, and the constant `C` is
122	the average of `erfc(z) * exp(zz) z` taken at the endpoints of the range.
123	Once again, as long as the absolute error in R(z - B) is small
124	compared to `c` then `c + R(z - B)` will be correctly rounded, and the error
125	in the result will depend only on the accuracy of the exp function. In practice,
126	in all but a very small number of cases, the error is confined to the last bit
127	of the result. The constant `B` is chosen so that the left hand end of the range
128	of the rational approximation is 0.
129
130	For large `z` over a range \[a, +[infin]\] the above approximation is modified to:
131
132	erfc(z) = exp(-zz) (C + R(1 / z)) / z;
133
134	[endsect]
135	[/ :error_function The Error Functions]
136
137	[/
138	Copyright 2006 John Maddock and Paul A. Bristow.
139	Distributed under the Boost Software License, Version 1.0.
140	(See accompanying file LICENSE_1_0.txt or copy at
141	http://www.boost.org/LICENSE_1_0.txt).
142	]