| 1 | // Copyright Nick Thompson, 2017 |
| 2 | // Use, modification and distribution are subject to the |
| 3 | // Boost Software License, Version 1.0. |
| 4 | // (See accompanying file LICENSE_1_0.txt |
| 5 | // or copy at http://www.boost.org/LICENSE_1_0.txt) |
| 6 | |
| 7 | /* |
| 8 | * This class performs tanh-sinh quadrature on the real line. |
| 9 | * Tanh-sinh quadrature is exponentially convergent for integrands in Hardy spaces, |
| 10 | * (see https://en.wikipedia.org/wiki/Hardy_space for a formal definition), and is optimal for a random function from that class. |
| 11 | * |
| 12 | * The tanh-sinh quadrature is one of a class of so called "double exponential quadratures"-there is a large family of them, |
| 13 | * but this one seems to be the most commonly used. |
| 14 | * |
| 15 | * As always, there are caveats: For instance, if the function you want to integrate is not holomorphic on the unit disk, |
| 16 | * then the rapid convergence will be spoiled. In this case, a more appropriate quadrature is (say) Romberg, which does not |
| 17 | * require the function to be holomorphic, only differentiable up to some order. |
| 18 | * |
| 19 | * In addition, if you are integrating a periodic function over a period, the trapezoidal rule is better. |
| 20 | * |
| 21 | * References: |
| 22 | * |
| 23 | * 1) Mori, Masatake. "Quadrature formulas obtained by variable transformation and the DE-rule." Journal of Computational and Applied Mathematics 12 (1985): 119-130. |
| 24 | * 2) Bailey, David H., Karthik Jeyabalan, and Xiaoye S. Li. "A comparison of three high-precision quadrature schemes." Experimental Mathematics 14.3 (2005): 317-329. |
| 25 | * 3) Press, William H., et al. "Numerical recipes third edition: the art of scientific computing." Cambridge University Press 32 (2007): 10013-2473. |
| 26 | * |
| 27 | */ |
| 28 | |
| 29 | #ifndef BOOST_MATH_QUADRATURE_TANH_SINH_HPP |
| 30 | #define BOOST_MATH_QUADRATURE_TANH_SINH_HPP |
| 31 | |
| 32 | #include <cmath> |
| 33 | #include <limits> |
| 34 | #include <memory> |
| 35 | #include <boost/math/quadrature/detail/tanh_sinh_detail.hpp> |
| 36 | |
| 37 | namespace boost{ namespace math{ namespace quadrature { |
| 38 | |
| 39 | template<class Real, class Policy = policies::policy<> > |
| 40 | class tanh_sinh |
| 41 | { |
| 42 | public: |
| 43 | tanh_sinh(size_t max_refinements = 15, const Real& min_complement = tools::min_value<Real>() * 4) |
| 44 | : m_imp(std::make_shared<detail::tanh_sinh_detail<Real, Policy>>(max_refinements, min_complement)) {} |
| 45 | |
| 46 | template<class F> |
| 47 | auto integrate(const F f, Real a, Real b, Real tolerance = tools::root_epsilon<Real>(), Real* error = nullptr, Real* L1 = nullptr, std::size_t* levels = nullptr) const ->decltype(std::declval<F>()(std::declval<Real>())); |
| 48 | template<class F> |
| 49 | auto integrate(const F f, Real a, Real b, Real tolerance = tools::root_epsilon<Real>(), Real* error = nullptr, Real* L1 = nullptr, std::size_t* levels = nullptr) const ->decltype(std::declval<F>()(std::declval<Real>(), std::declval<Real>())); |
| 50 | |
| 51 | template<class F> |
| 52 | auto integrate(const F f, Real tolerance = tools::root_epsilon<Real>(), Real* error = nullptr, Real* L1 = nullptr, std::size_t* levels = nullptr) const ->decltype(std::declval<F>()(std::declval<Real>())); |
| 53 | template<class F> |
| 54 | auto integrate(const F f, Real tolerance = tools::root_epsilon<Real>(), Real* error = nullptr, Real* L1 = nullptr, std::size_t* levels = nullptr) const ->decltype(std::declval<F>()(std::declval<Real>(), std::declval<Real>())); |
| 55 | |
| 56 | private: |
| 57 | std::shared_ptr<detail::tanh_sinh_detail<Real, Policy>> m_imp; |
| 58 | }; |
| 59 | |
| 60 | template<class Real, class Policy> |
| 61 | template<class F> |
| 62 | auto tanh_sinh<Real, Policy>::integrate(const F f, Real a, Real b, Real tolerance, Real* error, Real* L1, std::size_t* levels) const ->decltype(std::declval<F>()(std::declval<Real>())) |
| 63 | { |
| 64 | BOOST_MATH_STD_USING |
| 65 | using boost::math::constants::half; |
| 66 | using boost::math::quadrature::detail::tanh_sinh_detail; |
| 67 | |
| 68 | static const char* function = "tanh_sinh<%1%>::integrate" ; |
| 69 | |
| 70 | typedef decltype(std::declval<F>()(std::declval<Real>())) result_type; |
| 71 | static_assert(!std::is_integral<result_type>::value, |
| 72 | "The return type cannot be integral, it must be either a real or complex floating point type." ); |
| 73 | if (!(boost::math::isnan)(a) && !(boost::math::isnan)(b)) |
| 74 | { |
| 75 | |
| 76 | // Infinite limits: |
| 77 | if ((a <= -tools::max_value<Real>()) && (b >= tools::max_value<Real>())) |
| 78 | { |
| 79 | auto u = [&](const Real& t, const Real& tc)->result_type |
| 80 | { |
| 81 | Real t_sq = t*t; |
| 82 | Real inv; |
| 83 | if (t > 0.5f) |
| 84 | inv = 1 / ((2 - tc) * tc); |
| 85 | else if(t < -0.5) |
| 86 | inv = 1 / ((2 + tc) * -tc); |
| 87 | else |
| 88 | inv = 1 / (1 - t_sq); |
| 89 | return f(t*inv)*(1 + t_sq)*inv*inv; |
| 90 | }; |
| 91 | Real limit = sqrt(tools::min_value<Real>()) * 4; |
| 92 | return m_imp->integrate(u, error, L1, function, limit, limit, tolerance, levels); |
| 93 | } |
| 94 | |
| 95 | // Right limit is infinite: |
| 96 | if ((boost::math::isfinite)(a) && (b >= tools::max_value<Real>())) |
| 97 | { |
| 98 | auto u = [&](const Real& t, const Real& tc)->result_type |
| 99 | { |
| 100 | Real z, arg; |
| 101 | if (t > -0.5f) |
| 102 | z = 1 / (t + 1); |
| 103 | else |
| 104 | z = -1 / tc; |
| 105 | if (t < 0.5) |
| 106 | arg = 2 * z + a - 1; |
| 107 | else |
| 108 | arg = a + tc / (2 - tc); |
| 109 | return f(arg)*z*z; |
| 110 | }; |
| 111 | Real left_limit = sqrt(tools::min_value<Real>()) * 4; |
| 112 | result_type Q = Real(2) * m_imp->integrate(u, error, L1, function, left_limit, tools::min_value<Real>(), tolerance, levels); |
| 113 | if (L1) |
| 114 | { |
| 115 | *L1 *= 2; |
| 116 | } |
| 117 | if (error) |
| 118 | { |
| 119 | *error *= 2; |
| 120 | } |
| 121 | |
| 122 | return Q; |
| 123 | } |
| 124 | |
| 125 | if ((boost::math::isfinite)(b) && (a <= -tools::max_value<Real>())) |
| 126 | { |
| 127 | auto v = [&](const Real& t, const Real& tc)->result_type |
| 128 | { |
| 129 | Real z; |
| 130 | if (t > -0.5) |
| 131 | z = 1 / (t + 1); |
| 132 | else |
| 133 | z = -1 / tc; |
| 134 | Real arg; |
| 135 | if (t < 0.5) |
| 136 | arg = 2 * z - 1; |
| 137 | else |
| 138 | arg = tc / (2 - tc); |
| 139 | return f(b - arg) * z * z; |
| 140 | }; |
| 141 | |
| 142 | Real left_limit = sqrt(tools::min_value<Real>()) * 4; |
| 143 | result_type Q = Real(2) * m_imp->integrate(v, error, L1, function, left_limit, tools::min_value<Real>(), tolerance, levels); |
| 144 | if (L1) |
| 145 | { |
| 146 | *L1 *= 2; |
| 147 | } |
| 148 | if (error) |
| 149 | { |
| 150 | *error *= 2; |
| 151 | } |
| 152 | return Q; |
| 153 | } |
| 154 | |
| 155 | if ((boost::math::isfinite)(a) && (boost::math::isfinite)(b)) |
| 156 | { |
| 157 | if (a == b) |
| 158 | { |
| 159 | return result_type(0); |
| 160 | } |
| 161 | if (b < a) |
| 162 | { |
| 163 | return -this->integrate(f, b, a, tolerance, error, L1, levels); |
| 164 | } |
| 165 | Real avg = (a + b)*half<Real>(); |
| 166 | Real diff = (b - a)*half<Real>(); |
| 167 | Real avg_over_diff_m1 = a / diff; |
| 168 | Real avg_over_diff_p1 = b / diff; |
| 169 | bool have_small_left = fabs(a) < 0.5f; |
| 170 | bool have_small_right = fabs(b) < 0.5f; |
| 171 | Real left_min_complement = float_next(avg_over_diff_m1) - avg_over_diff_m1; |
| 172 | Real min_complement_limit = (std::max)(tools::min_value<Real>(), float_next(Real(tools::min_value<Real>() / diff))); |
| 173 | if (left_min_complement < min_complement_limit) |
| 174 | left_min_complement = min_complement_limit; |
| 175 | Real right_min_complement = avg_over_diff_p1 - float_prior(avg_over_diff_p1); |
| 176 | if (right_min_complement < min_complement_limit) |
| 177 | right_min_complement = min_complement_limit; |
| 178 | // |
| 179 | // These asserts will fail only if rounding errors on |
| 180 | // type Real have accumulated so much error that it's |
| 181 | // broken our internal logic. Should that prove to be |
| 182 | // a persistent issue, we might need to add a bit of fudge |
| 183 | // factor to move left_min_complement and right_min_complement |
| 184 | // further from the end points of the range. |
| 185 | // |
| 186 | BOOST_MATH_ASSERT((left_min_complement * diff + a) > a); |
| 187 | BOOST_MATH_ASSERT((b - right_min_complement * diff) < b); |
| 188 | auto u = [&](Real z, Real zc)->result_type |
| 189 | { |
| 190 | Real position; |
| 191 | if (z < -0.5) |
| 192 | { |
| 193 | if(have_small_left) |
| 194 | return f(diff * (avg_over_diff_m1 - zc)); |
| 195 | position = a - diff * zc; |
| 196 | } |
| 197 | else if (z > 0.5) |
| 198 | { |
| 199 | if(have_small_right) |
| 200 | return f(diff * (avg_over_diff_p1 - zc)); |
| 201 | position = b - diff * zc; |
| 202 | } |
| 203 | else |
| 204 | position = avg + diff*z; |
| 205 | BOOST_MATH_ASSERT(position != a); |
| 206 | BOOST_MATH_ASSERT(position != b); |
| 207 | return f(position); |
| 208 | }; |
| 209 | result_type Q = diff*m_imp->integrate(u, error, L1, function, left_min_complement, right_min_complement, tolerance, levels); |
| 210 | |
| 211 | if (L1) |
| 212 | { |
| 213 | *L1 *= diff; |
| 214 | } |
| 215 | if (error) |
| 216 | { |
| 217 | *error *= diff; |
| 218 | } |
| 219 | return Q; |
| 220 | } |
| 221 | } |
| 222 | return policies::raise_domain_error(function, "The domain of integration is not sensible; please check the bounds." , a, Policy()); |
| 223 | } |
| 224 | |
| 225 | template<class Real, class Policy> |
| 226 | template<class F> |
| 227 | auto tanh_sinh<Real, Policy>::integrate(const F f, Real a, Real b, Real tolerance, Real* error, Real* L1, std::size_t* levels) const ->decltype(std::declval<F>()(std::declval<Real>(), std::declval<Real>())) |
| 228 | { |
| 229 | BOOST_MATH_STD_USING |
| 230 | using boost::math::constants::half; |
| 231 | using boost::math::quadrature::detail::tanh_sinh_detail; |
| 232 | |
| 233 | static const char* function = "tanh_sinh<%1%>::integrate" ; |
| 234 | |
| 235 | if ((boost::math::isfinite)(a) && (boost::math::isfinite)(b)) |
| 236 | { |
| 237 | if (b <= a) |
| 238 | { |
| 239 | return policies::raise_domain_error(function, "Arguments to integrate are in wrong order; integration over [a,b] must have b > a." , a, Policy()); |
| 240 | } |
| 241 | auto u = [&](Real z, Real zc)->Real |
| 242 | { |
| 243 | if (z < 0) |
| 244 | return f((a - b) * zc / 2 + a, (b - a) * zc / 2); |
| 245 | else |
| 246 | return f((a - b) * zc / 2 + b, (b - a) * zc / 2); |
| 247 | }; |
| 248 | Real diff = (b - a)*half<Real>(); |
| 249 | Real left_min_complement = tools::min_value<Real>() * 4; |
| 250 | Real right_min_complement = tools::min_value<Real>() * 4; |
| 251 | Real Q = diff*m_imp->integrate(u, error, L1, function, left_min_complement, right_min_complement, tolerance, levels); |
| 252 | |
| 253 | if (L1) |
| 254 | { |
| 255 | *L1 *= diff; |
| 256 | } |
| 257 | if (error) |
| 258 | { |
| 259 | *error *= diff; |
| 260 | } |
| 261 | return Q; |
| 262 | } |
| 263 | return policies::raise_domain_error(function, "The domain of integration is not sensible; please check the bounds." , a, Policy()); |
| 264 | } |
| 265 | |
| 266 | template<class Real, class Policy> |
| 267 | template<class F> |
| 268 | auto tanh_sinh<Real, Policy>::integrate(const F f, Real tolerance, Real* error, Real* L1, std::size_t* levels) const ->decltype(std::declval<F>()(std::declval<Real>())) |
| 269 | { |
| 270 | using boost::math::quadrature::detail::tanh_sinh_detail; |
| 271 | static const char* function = "tanh_sinh<%1%>::integrate" ; |
| 272 | Real min_complement = tools::epsilon<Real>(); |
| 273 | return m_imp->integrate([&](const Real& arg, const Real&) { return f(arg); }, error, L1, function, min_complement, min_complement, tolerance, levels); |
| 274 | } |
| 275 | |
| 276 | template<class Real, class Policy> |
| 277 | template<class F> |
| 278 | auto tanh_sinh<Real, Policy>::integrate(const F f, Real tolerance, Real* error, Real* L1, std::size_t* levels) const ->decltype(std::declval<F>()(std::declval<Real>(), std::declval<Real>())) |
| 279 | { |
| 280 | using boost::math::quadrature::detail::tanh_sinh_detail; |
| 281 | static const char* function = "tanh_sinh<%1%>::integrate" ; |
| 282 | Real min_complement = tools::min_value<Real>() * 4; |
| 283 | return m_imp->integrate(f, error, L1, function, min_complement, min_complement, tolerance, levels); |
| 284 | } |
| 285 | |
| 286 | } |
| 287 | } |
| 288 | } |
| 289 | #endif |
| 290 | |