mirror of
https://github.com/saitohirga/WSJT-X.git
synced 2024-11-15 16:42:12 -05:00
216 lines
8.0 KiB
C++
216 lines
8.0 KiB
C++
// Copyright John Maddock 2006.
|
|
// Copyright Paul A. Bristow 2007, 2009
|
|
// 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)
|
|
|
|
#include <boost/math/concepts/real_concept.hpp>
|
|
#define BOOST_TEST_MAIN
|
|
#include <boost/test/unit_test.hpp>
|
|
#include <boost/test/floating_point_comparison.hpp>
|
|
#include <boost/math/special_functions/math_fwd.hpp>
|
|
#include <boost/math/constants/constants.hpp>
|
|
#include <boost/type_traits/is_floating_point.hpp>
|
|
#include <boost/array.hpp>
|
|
#include "functor.hpp"
|
|
|
|
#include "handle_test_result.hpp"
|
|
#include "table_type.hpp"
|
|
|
|
#ifndef SC_
|
|
#define SC_(x) static_cast<typename table_type<T>::type>(BOOST_JOIN(x, L))
|
|
#endif
|
|
|
|
template <class Real, class T>
|
|
void do_test_erf(const T& data, const char* type_name, const char* test_name)
|
|
{
|
|
typedef Real value_type;
|
|
|
|
typedef value_type (*pg)(value_type);
|
|
#ifdef ERF_FUNCTION_TO_TEST
|
|
pg funcp = ERF_FUNCTION_TO_TEST;
|
|
#elif defined(BOOST_MATH_NO_DEDUCED_FUNCTION_POINTERS)
|
|
pg funcp = boost::math::erf<value_type>;
|
|
#else
|
|
pg funcp = boost::math::erf;
|
|
#endif
|
|
|
|
boost::math::tools::test_result<value_type> result;
|
|
|
|
std::cout << "Testing " << test_name << " with type " << type_name
|
|
<< "\n~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\n";
|
|
|
|
//
|
|
// test erf against data:
|
|
//
|
|
#if !(defined(ERROR_REPORTING_MODE) && !defined(ERF_FUNCTION_TO_TEST))
|
|
result = boost::math::tools::test_hetero<Real>(
|
|
data,
|
|
bind_func<Real>(funcp, 0),
|
|
extract_result<Real>(1));
|
|
handle_test_result(result, data[result.worst()], result.worst(), type_name, "erf", test_name);
|
|
#endif
|
|
//
|
|
// test erfc against data:
|
|
//
|
|
#ifdef ERFC_FUNCTION_TO_TEST
|
|
funcp = ERFC_FUNCTION_TO_TEST;
|
|
#elif defined(BOOST_MATH_NO_DEDUCED_FUNCTION_POINTERS)
|
|
funcp = boost::math::erfc<value_type>;
|
|
#else
|
|
funcp = boost::math::erfc;
|
|
#endif
|
|
#if !(defined(ERROR_REPORTING_MODE) && !defined(ERFC_FUNCTION_TO_TEST))
|
|
result = boost::math::tools::test_hetero<Real>(
|
|
data,
|
|
bind_func<Real>(funcp, 0),
|
|
extract_result<Real>(2));
|
|
handle_test_result(result, data[result.worst()], result.worst(), type_name, "erfc", test_name);
|
|
std::cout << std::endl;
|
|
#endif
|
|
}
|
|
|
|
|
|
template <class Real, class T>
|
|
void do_test_erf_inv(const T& data, const char* type_name, const char* test_name)
|
|
{
|
|
#if !(defined(ERROR_REPORTING_MODE) && !defined(ERF_INV_FUNCTION_TO_TEST))
|
|
typedef Real value_type;
|
|
|
|
typedef value_type (*pg)(value_type);
|
|
|
|
boost::math::tools::test_result<value_type> result;
|
|
std::cout << "Testing " << test_name << " with type " << type_name
|
|
<< "\n~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\n";
|
|
//
|
|
// test erf_inv against data:
|
|
//
|
|
#ifdef ERF_INV_FUNCTION_TO_TEST
|
|
pg funcp = ERF_INV_FUNCTION_TO_TEST;
|
|
#elif defined(BOOST_MATH_NO_DEDUCED_FUNCTION_POINTERS)
|
|
pg funcp = boost::math::erf_inv<value_type>;
|
|
#else
|
|
pg funcp = boost::math::erf_inv;
|
|
#endif
|
|
result = boost::math::tools::test_hetero<Real>(
|
|
data,
|
|
bind_func<Real>(funcp, 0),
|
|
extract_result<Real>(1));
|
|
handle_test_result(result, data[result.worst()], result.worst(), type_name, "erf_inv", test_name);
|
|
std::cout << std::endl;
|
|
#endif
|
|
}
|
|
|
|
template <class Real, class T>
|
|
void do_test_erfc_inv(const T& data, const char* type_name, const char* test_name)
|
|
{
|
|
#if !(defined(ERROR_REPORTING_MODE) && !defined(ERFC_INV_FUNCTION_TO_TEST))
|
|
typedef Real value_type;
|
|
|
|
typedef value_type (*pg)(value_type);
|
|
|
|
boost::math::tools::test_result<value_type> result;
|
|
std::cout << "Testing " << test_name << " with type " << type_name
|
|
<< "\n~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\n";
|
|
//
|
|
// test erfc_inv against data:
|
|
//
|
|
#ifdef ERFC_INV_FUNCTION_TO_TEST
|
|
pg funcp = ERFC_INV_FUNCTION_TO_TEST;
|
|
#elif defined(BOOST_MATH_NO_DEDUCED_FUNCTION_POINTERS)
|
|
pg funcp = boost::math::erfc_inv<value_type>;
|
|
#else
|
|
pg funcp = boost::math::erfc_inv;
|
|
#endif
|
|
result = boost::math::tools::test_hetero<Real>(
|
|
data,
|
|
bind_func<Real>(funcp, 0),
|
|
extract_result<Real>(1));
|
|
handle_test_result(result, data[result.worst()], result.worst(), type_name, "erfc_inv", test_name);
|
|
std::cout << std::endl;
|
|
#endif
|
|
}
|
|
|
|
template <class T>
|
|
void test_erf(T, const char* name)
|
|
{
|
|
//
|
|
// The actual test data is rather verbose, so it's in a separate file
|
|
//
|
|
// The contents are as follows, each row of data contains
|
|
// three items, input value a, input value b and erf(a, b):
|
|
//
|
|
# include "erf_small_data.ipp"
|
|
|
|
do_test_erf<T>(erf_small_data, name, "Erf Function: Small Values");
|
|
|
|
# include "erf_data.ipp"
|
|
|
|
do_test_erf<T>(erf_data, name, "Erf Function: Medium Values");
|
|
|
|
# include "erf_large_data.ipp"
|
|
|
|
do_test_erf<T>(erf_large_data, name, "Erf Function: Large Values");
|
|
|
|
# include "erf_inv_data.ipp"
|
|
|
|
do_test_erf_inv<T>(erf_inv_data, name, "Inverse Erf Function");
|
|
|
|
# include "erfc_inv_data.ipp"
|
|
|
|
do_test_erfc_inv<T>(erfc_inv_data, name, "Inverse Erfc Function");
|
|
|
|
# include "erfc_inv_big_data.ipp"
|
|
|
|
if(std::numeric_limits<T>::min_exponent <= -4500)
|
|
{
|
|
do_test_erfc_inv<T>(erfc_inv_big_data, name, "Inverse Erfc Function: extreme values");
|
|
}
|
|
}
|
|
|
|
template <class T>
|
|
void test_spots(T, const char* t)
|
|
{
|
|
std::cout << "Testing basic sanity checks for type " << t << std::endl;
|
|
//
|
|
// basic sanity checks, tolerance is 10 epsilon expressed as a percentage:
|
|
//
|
|
T tolerance = boost::math::tools::epsilon<T>() * 1000;
|
|
BOOST_CHECK_CLOSE(::boost::math::erfc(static_cast<T>(0.125)), static_cast<T>(0.85968379519866618260697055347837660181302041685015L), tolerance);
|
|
BOOST_CHECK_CLOSE(::boost::math::erfc(static_cast<T>(0.5)), static_cast<T>(0.47950012218695346231725334610803547126354842424204L), tolerance);
|
|
BOOST_CHECK_CLOSE(::boost::math::erfc(static_cast<T>(1)), static_cast<T>(0.15729920705028513065877936491739074070393300203370L), tolerance);
|
|
BOOST_CHECK_CLOSE(::boost::math::erfc(static_cast<T>(5)), static_cast<T>(1.5374597944280348501883434853833788901180503147234e-12L), tolerance);
|
|
BOOST_CHECK_CLOSE(::boost::math::erfc(static_cast<T>(-0.125)), static_cast<T>(1.1403162048013338173930294465216233981869795831498L), tolerance);
|
|
BOOST_CHECK_CLOSE(::boost::math::erfc(static_cast<T>(-0.5)), static_cast<T>(1.5204998778130465376827466538919645287364515757580L), tolerance);
|
|
BOOST_CHECK_CLOSE(::boost::math::erfc(static_cast<T>(0)), static_cast<T>(1), tolerance);
|
|
|
|
BOOST_CHECK_CLOSE(::boost::math::erf(static_cast<T>(0.125)), static_cast<T>(0.14031620480133381739302944652162339818697958314985L), tolerance);
|
|
BOOST_CHECK_CLOSE(::boost::math::erf(static_cast<T>(0.5)), static_cast<T>(0.52049987781304653768274665389196452873645157575796L), tolerance);
|
|
BOOST_CHECK_CLOSE(::boost::math::erf(static_cast<T>(1)), static_cast<T>(0.84270079294971486934122063508260925929606699796630L), tolerance);
|
|
BOOST_CHECK_CLOSE(::boost::math::erf(static_cast<T>(5)), static_cast<T>(0.9999999999984625402055719651498116565146166211099L), tolerance);
|
|
BOOST_CHECK_CLOSE(::boost::math::erf(static_cast<T>(-0.125)), static_cast<T>(-0.14031620480133381739302944652162339818697958314985L), tolerance);
|
|
BOOST_CHECK_CLOSE(::boost::math::erf(static_cast<T>(-0.5)), static_cast<T>(-0.52049987781304653768274665389196452873645157575796L), tolerance);
|
|
BOOST_CHECK_CLOSE(::boost::math::erf(static_cast<T>(0)), static_cast<T>(0), tolerance);
|
|
|
|
tolerance = boost::math::tools::epsilon<T>() * 100 * 200; // 200 eps %.
|
|
#if defined(__CYGWIN__)
|
|
// some platforms long double is only reliably accurate to double precision:
|
|
if(sizeof(T) == sizeof(long double))
|
|
tolerance = boost::math::tools::epsilon<double>() * 100 * 200; // 200 eps %.
|
|
#endif
|
|
|
|
for(T i = -0.95f; i < 1; i += 0.125f)
|
|
{
|
|
T inv = boost::math::erf_inv(i);
|
|
T b = boost::math::erf(inv);
|
|
BOOST_CHECK_CLOSE(b, i, tolerance);
|
|
}
|
|
for(T j = 0.125f; j < 2; j += 0.125f)
|
|
{
|
|
T inv = boost::math::erfc_inv(j);
|
|
T b = boost::math::erfc(inv);
|
|
BOOST_CHECK_CLOSE(b, j, tolerance);
|
|
}
|
|
}
|
|
|