diff options
Diffstat (limited to 'src/boost/libs/math/tools/ibeta_data.cpp')
-rw-r--r-- | src/boost/libs/math/tools/ibeta_data.cpp | 303 |
1 files changed, 303 insertions, 0 deletions
diff --git a/src/boost/libs/math/tools/ibeta_data.cpp b/src/boost/libs/math/tools/ibeta_data.cpp new file mode 100644 index 00000000..667365ec --- /dev/null +++ b/src/boost/libs/math/tools/ibeta_data.cpp @@ -0,0 +1,303 @@ +// (C) Copyright John Maddock 2006. +// 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/special_functions/gamma.hpp> +#include <boost/math/special_functions/beta.hpp> +#include <boost/math/constants/constants.hpp> +#include <boost/lexical_cast.hpp> +#include <fstream> +#include <map> +#include <boost/math/tools/test_data.hpp> +#include <boost/random.hpp> +#include "mp_t.hpp" + +using namespace boost::math::tools; +using namespace boost::math; +using namespace std; + +template <class T> +struct ibeta_fraction1_t +{ + typedef std::pair<T, T> result_type; + + ibeta_fraction1_t(T a_, T b_, T x_) : a(a_), b(b_), x(x_), k(1) {} + + result_type operator()() + { + T aN; + if(k & 1) + { + int m = (k - 1) / 2; + aN = -(a + m) * (a + b + m) * x; + aN /= a + 2*m; + aN /= a + 2*m + 1; + } + else + { + int m = k / 2; + aN = m * (b - m) *x; + aN /= a + 2*m - 1; + aN /= a + 2*m; + } + ++k; + return std::make_pair(aN, T(1)); + } + +private: + T a, b, x; + int k; +}; + +// +// This function caches previous calls to beta +// just so we can speed things up a bit: +// +template <class T> +T get_beta(T a, T b) +{ + static std::map<std::pair<T, T>, T> m; + + if(a < b) + std::swap(a, b); + + std::pair<T, T> p(a, b); + typename std::map<std::pair<T, T>, T>::const_iterator i = m.find(p); + if(i != m.end()) + return i->second; + + T r = beta(a, b); + p.first = a; + p.second = b; + m[p] = r; + + return r; +} + +// +// compute the continued fraction: +// +template <class T> +T get_ibeta_fraction1(T a, T b, T x) +{ + ibeta_fraction1_t<T> f(a, b, x); + T fract = boost::math::tools::continued_fraction_a(f, boost::math::policies::digits<T, boost::math::policies::policy<> >()); + T denom = (a * (fract + 1)); + T num = pow(x, a) * pow(1 - x, b); + if(num == 0) + return 0; + else if(denom == 0) + return -1; + return num / denom; +} +// +// calculate the incomplete beta from the fraction: +// +template <class T> +std::pair<T,T> ibeta_fraction1(T a, T b, T x) +{ + T bet = get_beta(a, b); + if(x > ((a+1)/(a+b+2))) + { + T fract = get_ibeta_fraction1(b, a, 1-x); + if(fract/bet > 0.75) + { + fract = get_ibeta_fraction1(a, b, x); + return std::make_pair(fract, bet - fract); + } + return std::make_pair(bet - fract, fract); + } + T fract = get_ibeta_fraction1(a, b, x); + if(fract/bet > 0.75) + { + fract = get_ibeta_fraction1(b, a, 1-x); + return std::make_pair(bet - fract, fract); + } + return std::make_pair(fract, bet - fract); + +} +// +// calculate the regularised incomplete beta from the fraction: +// +template <class T> +std::pair<T,T> ibeta_fraction1_regular(T a, T b, T x) +{ + T bet = get_beta(a, b); + if(x > ((a+1)/(a+b+2))) + { + T fract = get_ibeta_fraction1(b, a, 1-x); + if(fract == 0) + bet = 1; // normalise so we don't get 0/0 + else if(bet == 0) + return std::make_pair(T(-1), T(-1)); // Yikes!! + if(fract / bet > 0.75) + { + fract = get_ibeta_fraction1(a, b, x); + return std::make_pair(fract / bet, 1 - (fract / bet)); + } + return std::make_pair(1 - (fract / bet), fract / bet); + } + T fract = get_ibeta_fraction1(a, b, x); + if(fract / bet > 0.75) + { + fract = get_ibeta_fraction1(b, a, 1-x); + return std::make_pair(1 - (fract / bet), fract / bet); + } + return std::make_pair(fract / bet, 1 - (fract / bet)); +} + +// +// we absolutely must trunctate the input values to float +// precision: we have to be certain that the input values +// can be represented exactly in whatever width floating +// point type we are testing, otherwise the output will +// necessarily be off. +// +float external_f; +float force_truncate(const float* f) +{ + external_f = *f; + return external_f; +} + +float truncate_to_float(mp_t r) +{ + float f = boost::math::tools::real_cast<float>(r); + return force_truncate(&f); +} + +boost::mt19937 rnd; +boost::uniform_real<float> ur_a(1.0F, 5.0F); +boost::variate_generator<boost::mt19937, boost::uniform_real<float> > gen(rnd, ur_a); +boost::uniform_real<float> ur_a2(0.0F, 100.0F); +boost::variate_generator<boost::mt19937, boost::uniform_real<float> > gen2(rnd, ur_a2); + +struct beta_data_generator +{ + boost::math::tuple<mp_t, mp_t, mp_t, mp_t, mp_t, mp_t, mp_t> operator()(mp_t ap, mp_t bp, mp_t x_) + { + float a = truncate_to_float(real_cast<float>(gen() * pow(mp_t(10), ap))); + float b = truncate_to_float(real_cast<float>(gen() * pow(mp_t(10), bp))); + float x = truncate_to_float(real_cast<float>(x_)); + std::cout << a << " " << b << " " << x << std::endl; + std::pair<mp_t, mp_t> ib_full = ibeta_fraction1(mp_t(a), mp_t(b), mp_t(x)); + std::pair<mp_t, mp_t> ib_reg = ibeta_fraction1_regular(mp_t(a), mp_t(b), mp_t(x)); + return boost::math::make_tuple(a, b, x, ib_full.first, ib_full.second, ib_reg.first, ib_reg.second); + } +}; + +// medium sized values: +struct beta_data_generator_medium +{ + boost::math::tuple<mp_t, mp_t, mp_t, mp_t, mp_t, mp_t, mp_t> operator()(mp_t x_) + { + mp_t a = gen2(); + mp_t b = gen2(); + mp_t x = x_; + a = ConvPrec(a, 22); + b = ConvPrec(b, 22); + x = ConvPrec(x, 22); + std::cout << a << " " << b << " " << x << std::endl; + //mp_t exp_beta = boost::math::beta(a, b, x); + std::pair<mp_t, mp_t> ib_full = ibeta_fraction1(mp_t(a), mp_t(b), mp_t(x)); + /*exp_beta = boost::math::tools::relative_error(ib_full.first, exp_beta); + if(exp_beta > 1e-40) + { + std::cout << exp_beta << std::endl; + }*/ + std::pair<mp_t, mp_t> ib_reg = ibeta_fraction1_regular(mp_t(a), mp_t(b), mp_t(x)); + return boost::math::make_tuple(a, b, x, ib_full.first, ib_full.second, ib_reg.first, ib_reg.second); + } +}; + +struct beta_data_generator_small +{ + boost::math::tuple<mp_t, mp_t, mp_t, mp_t, mp_t, mp_t, mp_t> operator()(mp_t x_) + { + float a = truncate_to_float(gen2()/10); + float b = truncate_to_float(gen2()/10); + float x = truncate_to_float(real_cast<float>(x_)); + std::cout << a << " " << b << " " << x << std::endl; + std::pair<mp_t, mp_t> ib_full = ibeta_fraction1(mp_t(a), mp_t(b), mp_t(x)); + std::pair<mp_t, mp_t> ib_reg = ibeta_fraction1_regular(mp_t(a), mp_t(b), mp_t(x)); + return boost::math::make_tuple(a, b, x, ib_full.first, ib_full.second, ib_reg.first, ib_reg.second); + } +}; + +struct beta_data_generator_int +{ + boost::math::tuple<mp_t, mp_t, mp_t, mp_t, mp_t, mp_t, mp_t> operator()(mp_t a, mp_t b, mp_t x_) + { + float x = truncate_to_float(real_cast<float>(x_)); + std::cout << a << " " << b << " " << x << std::endl; + std::pair<mp_t, mp_t> ib_full = ibeta_fraction1(a, b, mp_t(x)); + std::pair<mp_t, mp_t> ib_reg = ibeta_fraction1_regular(a, b, mp_t(x)); + return boost::math::make_tuple(a, b, x, ib_full.first, ib_full.second, ib_reg.first, ib_reg.second); + } +}; + +int main(int, char* []) +{ + parameter_info<mp_t> arg1, arg2, arg3, arg4, arg5; + test_data<mp_t> data; + + std::cout << "Welcome.\n" + "This program will generate spot tests for the incomplete beta functions:\n" + " beta(a, b, x) and ibeta(a, b, x)\n\n" + "This is not an interactive program be prepared for a long wait!!!\n\n"; + + arg1 = make_periodic_param(mp_t(-5), mp_t(6), 11); + arg2 = make_periodic_param(mp_t(-5), mp_t(6), 11); + arg3 = make_random_param(mp_t(0.0001), mp_t(1), 10); + arg4 = make_random_param(mp_t(0.0001), mp_t(1), 100 /*500*/); + arg5 = make_periodic_param(mp_t(1), mp_t(41), 10); + + arg1.type |= dummy_param; + arg2.type |= dummy_param; + arg3.type |= dummy_param; + arg4.type |= dummy_param; + arg5.type |= dummy_param; + + // comment out all but one of the following when running + // or this program will take forever to complete! + //data.insert(beta_data_generator(), arg1, arg2, arg3); + //data.insert(beta_data_generator_medium(), arg4); + //data.insert(beta_data_generator_small(), arg4); + data.insert(beta_data_generator_int(), arg5, arg5, arg3); + + test_data<mp_t>::const_iterator i, j; + i = data.begin(); + j = data.end(); + while(i != j) + { + mp_t v1 = beta((*i)[0], (*i)[1], (*i)[2]); + mp_t v2 = relative_error(v1, (*i)[3]); + std::string s = boost::lexical_cast<std::string>((*i)[3]); + mp_t v3 = boost::lexical_cast<mp_t>(s); + mp_t v4 = relative_error(v3, (*i)[3]); + if(v2 > 1e-40) + { + std::cout << v2 << std::endl; + } + if(v4 > 1e-60) + { + std::cout << v4 << std::endl; + } + ++ i; + } + + std::cout << "Enter name of test data file [default=ibeta_data.ipp]"; + std::string line; + std::getline(std::cin, line); + boost::algorithm::trim(line); + if(line == "") + line = "ibeta_data.ipp"; + std::ofstream ofs(line.c_str()); + ofs << std::scientific << std::setprecision(40); + write_code(ofs, data, "ibeta_data"); + + return 0; +} + + |