//////////////////////////////////////////////////////////////// // Copyright 2013 - 2022 John Maddock. // Copyright 2022 Christopher Kormanyos. // Distributed under the Boost Software License, // Version 1.0. (See accompanying file LICENSE_1_0.txt // or copy at https://www.boost.org/LICENSE_1_0.txt) #ifndef BOOST_MP_CPP_BIN_FLOAT_HPP #define BOOST_MP_CPP_BIN_FLOAT_HPP #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include // // Some includes we need from Boost.Math, since we rely on that library to provide these functions: // #ifdef BOOST_MP_MATH_AVAILABLE #include #include #include #include #include #include #endif #ifdef BOOST_HAS_FLOAT128 #include #endif namespace boost { namespace multiprecision { namespace backends { #ifdef BOOST_MSVC #pragma warning(push) #pragma warning(disable : 4522 6326) // multiple assignment operators specified, comparison of two constants #endif namespace detail { template inline typename std::enable_if::value, bool>::type is_negative(U) { return false; } template inline typename std::enable_if< !boost::multiprecision::detail::is_unsigned::value, bool>::type is_negative(S s) { return s < 0; } template ::value == number_kind_floating_point> struct is_cpp_bin_float_implicitly_constructible_from_type { static constexpr bool value = false; }; template struct is_cpp_bin_float_implicitly_constructible_from_type { static constexpr bool value = (std::numeric_limits::digits <= static_cast(bit_count)) && (std::numeric_limits::radix == 2) && std::numeric_limits::is_specialized #ifdef BOOST_HAS_FLOAT128 && !std::is_same::value #endif && (std::is_floating_point::value || is_number::value); }; template ::value == number_kind_floating_point> struct is_cpp_bin_float_explicitly_constructible_from_type { static constexpr bool value = false; }; template struct is_cpp_bin_float_explicitly_constructible_from_type { static constexpr bool value = (std::numeric_limits::digits > static_cast(bit_count)) && (std::numeric_limits::radix == 2) && std::numeric_limits::is_specialized #ifdef BOOST_HAS_FLOAT128 && !std::is_same::value #endif ; }; } // namespace detail template class cpp_bin_float { public: static constexpr unsigned bit_count = DigitBase == digit_base_2 ? Digits : (Digits * 1000uL) / 301uL + (((Digits * 1000uL) % 301) ? 2u : 1u); using rep_type = cpp_int_backend::value ? bit_count : 0, bit_count, std::is_void::value ? unsigned_magnitude : signed_magnitude, unchecked, Allocator>; using double_rep_type = cpp_int_backend::value ? 2 * bit_count : 0, 2 * bit_count, std::is_void::value ? unsigned_magnitude : signed_magnitude, unchecked, Allocator>; using signed_types = typename rep_type::signed_types; using unsigned_types = typename rep_type::unsigned_types; using float_types = std::tuple; using exponent_type = Exponent; static constexpr exponent_type max_exponent_limit = (std::numeric_limits::max)()- 2 * static_cast(bit_count); static constexpr exponent_type min_exponent_limit = (std::numeric_limits::min)() + 2 * static_cast(bit_count); static_assert(MinExponent >= min_exponent_limit, "Template parameter MinExponent is too negative for our internal logic to function correctly, sorry!"); static_assert(MaxExponent <= max_exponent_limit, "Template parameter MaxExponent is too large for our internal logic to function correctly, sorry!"); static_assert(MinExponent <= 0, "Template parameter MinExponent can not be positive!"); static_assert(MaxExponent >= 0, "Template parameter MaxExponent can not be negative!"); static constexpr exponent_type max_exponent = MaxExponent == 0 ? max_exponent_limit : MaxExponent; static constexpr exponent_type min_exponent = MinExponent == 0 ? min_exponent_limit : MinExponent; static constexpr exponent_type exponent_zero = max_exponent + 1; static constexpr exponent_type exponent_infinity = max_exponent + 2; static constexpr exponent_type exponent_nan = max_exponent + 3; private: rep_type m_data; exponent_type m_exponent; bool m_sign; public: cpp_bin_float() noexcept(noexcept(rep_type())) : m_data(), m_exponent(exponent_zero), m_sign(false) {} cpp_bin_float(const cpp_bin_float& o) noexcept(noexcept(rep_type(std::declval()))) : m_data(o.m_data), m_exponent(o.m_exponent), m_sign(o.m_sign) {} template cpp_bin_float(const cpp_bin_float& o, typename std::enable_if<(bit_count >= cpp_bin_float::bit_count)>::type const* = nullptr) { *this = o; } template explicit cpp_bin_float(const cpp_bin_float& o, typename std::enable_if< !(bit_count >= cpp_bin_float::bit_count)>::type const* = nullptr) : m_exponent(o.exponent()), m_sign(o.sign()) { *this = o; } // rvalue copy: template cpp_bin_float(cpp_bin_float&& o, typename std::enable_if<(bit_count >= cpp_bin_float::bit_count)>::type const* = nullptr)noexcept(noexcept(rep_type(std::declval()))) { *this = std::move(o); } template explicit cpp_bin_float(cpp_bin_float&& o, typename std::enable_if< !(bit_count >= cpp_bin_float::bit_count)>::type const* = nullptr) noexcept(noexcept(rep_type(std::declval()))) : m_exponent(o.exponent()), m_sign(o.sign()) { *this = std::move(o); } template cpp_bin_float(const Float& f, typename std::enable_if(bit_count)>::value>::type const* = nullptr) : m_data(), m_exponent(0), m_sign(false) { this->assign_float(f); } template explicit cpp_bin_float(const Float& f, typename std::enable_if(bit_count)>::value>::type const* = nullptr) : m_data(), m_exponent(0), m_sign(false) { this->assign_float(f); } #ifdef BOOST_HAS_FLOAT128 template cpp_bin_float(const Float& f, typename std::enable_if< std::is_same::value && (static_cast(bit_count) >= 113)>::type const* = nullptr) : m_data(), m_exponent(0), m_sign(false) { this->assign_float(f); } template explicit cpp_bin_float(const Float& f, typename std::enable_if< std::is_same::value && (static_cast(bit_count) < 113)>::type const* = nullptr) : m_data(), m_exponent(0), m_sign(false) { this->assign_float(f); } #endif cpp_bin_float& operator=(const cpp_bin_float& o) noexcept(noexcept(std::declval() = std::declval())) { m_data = o.m_data; m_exponent = o.m_exponent; m_sign = o.m_sign; return *this; } template cpp_bin_float& operator=(const cpp_bin_float& o) noexcept(noexcept(std::declval() = std::declval())) { m_data = o.bits(); m_sign = o.sign(); if (o.exponent() == cpp_bin_float::exponent_zero) m_exponent = exponent_zero; else if (o.exponent() == cpp_bin_float::exponent_nan) m_exponent = exponent_nan; else if (o.exponent() == cpp_bin_float::exponent_infinity) m_exponent = exponent_infinity; else if (o.exponent() > cpp_bin_float::max_exponent) { // Overflow: exponent() = cpp_bin_float::exponent_infinity; bits() = static_cast(0u); } else if (o.exponent() < cpp_bin_float::min_exponent) { // Underflow: exponent() = cpp_bin_float::exponent_zero; bits() = static_cast(0u); } else m_exponent = o.exponent(); return *this; } // rvalue copy: template cpp_bin_float& operator=(cpp_bin_float&& o) noexcept(noexcept(std::declval() = std::declval())) { m_data = std::move(o.bits()); m_sign = o.sign(); if (o.exponent() == cpp_bin_float::exponent_zero) m_exponent = exponent_zero; else if (o.exponent() == cpp_bin_float::exponent_nan) m_exponent = exponent_nan; else if (o.exponent() == cpp_bin_float::exponent_infinity) m_exponent = exponent_infinity; else if (o.exponent() > cpp_bin_float::max_exponent) { // Overflow: exponent() = cpp_bin_float::exponent_infinity; bits() = static_cast(0u); } else if (o.exponent() < cpp_bin_float::min_exponent) { // Underflow: exponent() = cpp_bin_float::exponent_zero; bits() = static_cast(0u); } else m_exponent = o.exponent(); return *this; } template cpp_bin_float& operator=(const cpp_bin_float& f) { switch (eval_fpclassify(f)) { case FP_ZERO: m_data = limb_type(0); m_sign = f.sign(); m_exponent = exponent_zero; break; case FP_NAN: m_data = limb_type(0); m_sign = false; m_exponent = exponent_nan; break; ; case FP_INFINITE: m_data = limb_type(0); m_sign = f.sign(); m_exponent = exponent_infinity; break; default: typename cpp_bin_float::rep_type b(f.bits()); this->exponent() = f.exponent() + (E)bit_count - (E)cpp_bin_float::bit_count; this->sign() = f.sign(); copy_and_round(*this, b); } return *this; } #ifdef BOOST_HAS_FLOAT128 template typename std::enable_if< (number_category::value == number_kind_floating_point) //&& (std::numeric_limits::digits <= static_cast(bit_count)) && ((std::numeric_limits::radix == 2) || (std::is_same::value)), cpp_bin_float&>::type operator=(const Float& f) #else template typename std::enable_if< (number_category::value == number_kind_floating_point) //&& (std::numeric_limits::digits <= static_cast(bit_count)) && (std::numeric_limits::radix == 2), cpp_bin_float&>::type operator=(const Float& f) #endif { return assign_float(f); } #ifdef BOOST_HAS_FLOAT128 template typename std::enable_if::value && (std::numeric_limits::digits > Digits), cpp_bin_float&>::type assign_float(Float f) { cpp_bin_float<113, DigitBase, Allocator, Exponent, MinExponent, MaxExponent> bf(f); return *this = bf; } template typename std::enable_if::value && (std::numeric_limits::digits <= Digits), cpp_bin_float&>::type assign_float(Float f) { using default_ops::eval_add; using bf_int_type = typename boost::multiprecision::detail::canonical::type; if (f == 0) { m_data = limb_type(0); m_sign = (signbitq(f) > 0); m_exponent = exponent_zero; return *this; } else if (isnanq(f)) { m_data = limb_type(0); m_sign = false; m_exponent = exponent_nan; return *this; } else if (isinfq(f)) { m_data = limb_type(0); m_sign = (f < 0); m_exponent = exponent_infinity; return *this; } if (f < 0) { *this = -f; this->negate(); return *this; } using ui_type = typename std::tuple_element<0, unsigned_types>::type; m_data = static_cast(0u); m_sign = false; m_exponent = 0; constexpr std::ptrdiff_t bits = sizeof(int) * CHAR_BIT - 1 < MaxExponent - 1 ? sizeof(int) * CHAR_BIT - 1 : 3; int e; f = frexpq(f, &e); while (f) { f = ldexpq(f, bits); e -= bits; int ipart = static_cast(truncq(f)); f -= ipart; m_exponent += bits; cpp_bin_float t; t = static_cast(ipart); eval_add(*this, t); } m_exponent += static_cast(e); if (m_exponent > max_exponent) { m_exponent = exponent_infinity; m_data = static_cast(0u); } else if (m_exponent < min_exponent) { m_exponent = exponent_zero; m_data = static_cast(0u); } return *this; } #endif #ifdef BOOST_HAS_FLOAT128 template typename std::enable_if::value && !std::is_same::value && (std::numeric_limits::digits > Digits), cpp_bin_float&>::type assign_float(Float f) #else template typename std::enable_if::value && (std::numeric_limits::digits > Digits), cpp_bin_float&>::type assign_float(Float f) #endif { cpp_bin_float::digits, DigitBase, Allocator, Exponent, MinExponent, MaxExponent> bf(f); return *this = bf; } #ifdef BOOST_HAS_FLOAT128 template typename std::enable_if::value && !std::is_same::value && (std::numeric_limits::digits <= Digits), cpp_bin_float&>::type assign_float(Float f) #else template typename std::enable_if::value && (std::numeric_limits::digits <= Digits), cpp_bin_float&>::type assign_float(Float f) #endif { using std::frexp; using std::ldexp; using std::signbit; using default_ops::eval_add; using bf_int_type = typename boost::multiprecision::detail::canonical::type; switch (BOOST_MP_FPCLASSIFY(f)) { case FP_ZERO: m_data = limb_type(0); m_sign = ((signbit)(f)); m_exponent = exponent_zero; return *this; case FP_NAN: m_data = limb_type(0); m_sign = false; m_exponent = exponent_nan; return *this; case FP_INFINITE: m_data = limb_type(0); m_sign = (f < 0); m_exponent = exponent_infinity; return *this; } if (f < 0) { *this = -f; this->negate(); return *this; } using ui_type = typename std::tuple_element<0, unsigned_types>::type; m_data = static_cast(0u); m_sign = false; m_exponent = 0; // // This code picks off the bits in f a few at a time and injects them into *this. // It does not do roundingm so we must have more digits precision in *this than // in the floating point value (the normal situation, unless we're emulating another // type like float16_t). // constexpr std::ptrdiff_t bits = static_cast(sizeof(int) * CHAR_BIT - 1) < static_cast(MaxExponent - 1) ? static_cast(sizeof(int) * CHAR_BIT - 1) : 3; int e; f = frexp(f, &e); while (f != static_cast(0.0F)) { f = ldexp(f, bits); e -= static_cast(bits); int ipart = boost::multiprecision::detail::itrunc(f); f -= static_cast(ipart); m_exponent += static_cast(bits); cpp_bin_float t; t = static_cast(ipart); eval_add(*this, t); } m_exponent += static_cast(e); if (m_exponent > max_exponent) { m_exponent = exponent_infinity; m_data = static_cast(0u); } else if(m_exponent < min_exponent) { m_exponent = exponent_zero; m_data = static_cast(0u); } return *this; } template typename std::enable_if< (number_category::value == number_kind_floating_point) && !std::is_floating_point::value && (number_category::value == number_kind_floating_point), cpp_bin_float&>::type assign_float(Float f) { using default_ops::eval_add; using default_ops::eval_convert_to; using default_ops::eval_get_sign; using default_ops::eval_subtract; using f_int_type = typename boost::multiprecision::detail::canonical::type ; using bf_int_type = typename boost::multiprecision::detail::canonical::type; switch (eval_fpclassify(f)) { case FP_ZERO: m_data = limb_type(0); m_sign = (eval_get_sign(f) > 0); m_exponent = exponent_zero; return *this; case FP_NAN: m_data = limb_type(0); m_sign = false; m_exponent = exponent_nan; return *this; case FP_INFINITE: m_data = limb_type(0); m_sign = eval_get_sign(f) < 0; m_exponent = exponent_infinity; return *this; } if (eval_get_sign(f) < 0) { f.negate(); assign_float(f); this->negate(); return *this; } using ui_type = typename std::tuple_element<0, unsigned_types>::type; m_data = static_cast(0u); m_sign = false; m_exponent = 0; constexpr std::ptrdiff_t bits = sizeof(int) * CHAR_BIT - 1; int e; eval_frexp(f, f, &e); while (eval_get_sign(f) != 0) { eval_ldexp(f, f, bits); e -= bits; int ipart; eval_convert_to(&ipart, f); eval_subtract(f, static_cast(ipart)); m_exponent += bits; eval_add(*this, static_cast(ipart)); } m_exponent += e; if (m_exponent > max_exponent) m_exponent = exponent_infinity; if (m_exponent < min_exponent) { m_data = limb_type(0u); m_exponent = exponent_zero; m_sign = (eval_get_sign(f) > 0); } else if (eval_get_sign(m_data) == 0) { m_exponent = exponent_zero; m_sign = (eval_get_sign(f) > 0); } return *this; } template cpp_bin_float& assign_float(const number& f) { return assign_float(f.backend()); } template typename std::enable_if::value, cpp_bin_float&>::type operator=(const I& i) { using default_ops::eval_bit_test; if (!i) { m_data = static_cast(0); m_exponent = exponent_zero; m_sign = false; } else { using ui_type = typename boost::multiprecision::detail::make_unsigned::type ; ui_type fi = static_cast(boost::multiprecision::detail::unsigned_abs(i)); using ar_type = typename boost::multiprecision::detail::canonical::type; m_data = static_cast(fi); std::size_t shift = msb(fi); if (shift > max_exponent) { m_exponent = exponent_infinity; m_data = static_cast(0); } else if (shift >= bit_count) { m_exponent = static_cast(shift); m_data = static_cast(fi >> (shift + 1 - bit_count)); } else { m_exponent = static_cast(shift); eval_left_shift(m_data, bit_count - shift - 1); } BOOST_MP_ASSERT((m_exponent == exponent_infinity) || eval_bit_test(m_data, bit_count - 1)); m_sign = detail::is_negative(i); } return *this; } cpp_bin_float& operator=(const char* s); void swap(cpp_bin_float& o) noexcept { m_data.swap(o.m_data); std::swap(m_exponent, o.m_exponent); std::swap(m_sign, o.m_sign); } std::string str(std::streamsize dig, std::ios_base::fmtflags f) const; void negate() { if (m_exponent != exponent_nan) m_sign = !m_sign; } int compare(const cpp_bin_float& o) const noexcept { if (m_sign != o.m_sign) return (m_exponent == exponent_zero) && (m_exponent == o.m_exponent) ? 0 : m_sign ? -1 : 1; int result; if (m_exponent == exponent_nan) return -1; else if (m_exponent != o.m_exponent) { if (m_exponent == exponent_zero) result = -1; else if (o.m_exponent == exponent_zero) result = 1; else result = m_exponent > o.m_exponent ? 1 : -1; } else result = m_data.compare(o.m_data); if (m_sign) result = -result; return result; } template int compare(const A& o) const noexcept { cpp_bin_float b; b = o; return compare(b); } rep_type& bits() { return m_data; } const rep_type& bits() const { return m_data; } exponent_type& exponent() { return m_exponent; } const exponent_type& exponent() const { return m_exponent; } bool& sign() { return m_sign; } const bool& sign() const { return m_sign; } void check_invariants() { using default_ops::eval_bit_test; using default_ops::eval_is_zero; if ((m_exponent <= max_exponent) && (m_exponent >= min_exponent)) { BOOST_MP_ASSERT(eval_bit_test(m_data, bit_count - 1)); } else { BOOST_MP_ASSERT(m_exponent > max_exponent); BOOST_MP_ASSERT(m_exponent <= exponent_nan); BOOST_MP_ASSERT(eval_is_zero(m_data)); } } #ifndef BOOST_MP_STANDALONE template void serialize(Archive& ar, const unsigned int /*version*/) { ar& boost::make_nvp("data", m_data); ar& boost::make_nvp("exponent", m_exponent); ar& boost::make_nvp("sign", m_sign); } #endif }; #ifdef BOOST_MSVC #pragma warning(pop) #endif template inline void copy_and_round(cpp_bin_float& res, Int& arg, std::ptrdiff_t bits_to_keep = cpp_bin_float::bit_count) { // Precondition: exponent of res must have been set before this function is called // as we may need to adjust it based on how many bits_to_keep in arg are set. using default_ops::eval_bit_test; using default_ops::eval_get_sign; using default_ops::eval_increment; using default_ops::eval_left_shift; using default_ops::eval_lsb; using default_ops::eval_msb; using default_ops::eval_right_shift; // cancellation may have resulted in arg being all zeros: if (eval_get_sign(arg) == 0) { res.exponent() = cpp_bin_float::exponent_zero; res.sign() = false; res.bits() = static_cast(0u); return; } std::ptrdiff_t msb = static_cast(eval_msb(arg)); if (static_cast(bits_to_keep) > msb + 1) { // Must have had cancellation in subtraction, // or be converting from a narrower type, so shift left: res.bits() = arg; eval_left_shift(res.bits(), static_cast(bits_to_keep - msb - 1)); res.exponent() -= static_cast(bits_to_keep - msb - 1); } else if (static_cast(bits_to_keep) < msb + 1) { // We have more bits_to_keep than we need, so round as required, // first get the rounding bit: bool roundup = eval_bit_test(arg, static_cast(msb - bits_to_keep)); // Then check for a tie: if (roundup && (msb - bits_to_keep == static_cast(eval_lsb(arg)))) { // Ties round towards even: if (!eval_bit_test(arg, static_cast(msb - bits_to_keep + 1))) roundup = false; } // Shift off the bits_to_keep we don't need: eval_right_shift(arg, static_cast(msb - bits_to_keep + 1)); res.exponent() += static_cast(msb - bits_to_keep + 1); if (roundup) { eval_increment(arg); if (bits_to_keep) { if (eval_bit_test(arg, static_cast(bits_to_keep))) { // This happens very very rairly, all the bits left after // truncation must be 1's and we're rounding up an order of magnitude: eval_right_shift(arg, 1u); ++res.exponent(); } } else { // We get here when bits_to_keep is zero but we're rounding up, // as a result we end up with a single digit that is a 1: ++bits_to_keep; } } if (bits_to_keep != cpp_bin_float::bit_count) { // Normalize result when we're rounding to fewer bits than we can hold, only happens in conversions // to narrower types: eval_left_shift(arg, static_cast(static_cast(cpp_bin_float::bit_count) - bits_to_keep)); res.exponent() -= static_cast(static_cast(cpp_bin_float::bit_count) - bits_to_keep); } res.bits() = arg; } else { res.bits() = arg; } if (!bits_to_keep && !res.bits().limbs()[0]) { // We're keeping zero bits and did not round up, so result is zero: res.exponent() = cpp_bin_float::exponent_zero; return; } // Result must be normalized: BOOST_MP_ASSERT(((std::ptrdiff_t )eval_msb(res.bits()) == cpp_bin_float::bit_count - 1)); if (res.exponent() > cpp_bin_float::max_exponent) { // Overflow: res.exponent() = cpp_bin_float::exponent_infinity; res.bits() = static_cast(0u); } else if (res.exponent() < cpp_bin_float::min_exponent) { // Underflow: res.exponent() = cpp_bin_float::exponent_zero; res.bits() = static_cast(0u); } } template inline void do_eval_add(cpp_bin_float& res, const BinFloat2& a, const BinFloat3& b) { if (a.exponent() < b.exponent()) { bool s = a.sign(); do_eval_add(res, b, a); if (res.sign() != s) res.negate(); return; } using default_ops::eval_add; using default_ops::eval_bit_test; using exponent_type = typename cpp_bin_float::exponent_type; typename cpp_bin_float::double_rep_type dt; // Special cases first: switch (a.exponent()) { case BinFloat2::exponent_zero: { bool s = a.sign(); res = b; res.sign() = s; return; } case BinFloat2::exponent_infinity: if (b.exponent() == cpp_bin_float::exponent_nan) res = b; else res = a; return; // result is still infinite. case cpp_bin_float::exponent_nan: res = a; return; // result is still a NaN. } switch (b.exponent()) { case BinFloat3::exponent_zero: res = a; return; case BinFloat3::exponent_infinity: res = b; if (res.sign()) res.negate(); return; // result is infinite. case BinFloat3::exponent_nan: res = b; return; // result is a NaN. } static_assert((std::numeric_limits::max)() - cpp_bin_float::bit_count > cpp_bin_float::max_exponent, "Exponent range check failed"); bool s = a.sign(); dt = a.bits(); if (a.exponent() > (std::ptrdiff_t )cpp_bin_float::bit_count + b.exponent()) { res.exponent() = a.exponent(); } else { exponent_type e_diff = a.exponent() - b.exponent(); BOOST_MP_ASSERT(e_diff >= 0); eval_left_shift(dt, static_cast(e_diff)); res.exponent() = a.exponent() - e_diff; eval_add(dt, b.bits()); } copy_and_round(res, dt); res.check_invariants(); if (res.sign() != s) res.negate(); } template inline void do_eval_subtract(BinFloat1& res, const BinFloat2& a, const BinFloat3& b) { using default_ops::eval_bit_test; using default_ops::eval_decrement; using default_ops::eval_subtract; typename BinFloat1::double_rep_type dt; // Special cases first: switch (a.exponent()) { case BinFloat2::exponent_zero: if (b.exponent() == BinFloat3::exponent_nan) res = std::numeric_limits >::quiet_NaN().backend(); else { bool s = a.sign(); res = b; if (res.exponent() == BinFloat1::exponent_zero) res.sign() = false; else if (res.sign() == s) res.negate(); } return; case BinFloat2::exponent_infinity: if ((b.exponent() == BinFloat3::exponent_nan) || (b.exponent() == BinFloat3::exponent_infinity)) res = std::numeric_limits >::quiet_NaN().backend(); else res = a; return; case BinFloat2::exponent_nan: res = a; return; // result is still a NaN. } switch (b.exponent()) { case BinFloat3::exponent_zero: res = a; return; case BinFloat3::exponent_infinity: res.exponent() = BinFloat1::exponent_infinity; res.sign() = !a.sign(); res.bits() = static_cast(0u); return; // result is a NaN. case BinFloat3::exponent_nan: res = b; return; // result is still a NaN. } bool s = a.sign(); if ((a.exponent() > b.exponent()) || ((a.exponent() == b.exponent()) && a.bits().compare(b.bits()) >= 0)) { dt = a.bits(); if (a.exponent() <= (std::ptrdiff_t )BinFloat1::bit_count + b.exponent()) { typename BinFloat1::exponent_type e_diff = a.exponent() - b.exponent(); eval_left_shift(dt, static_cast(e_diff)); res.exponent() = a.exponent() - e_diff; eval_subtract(dt, b.bits()); } else if (a.exponent() == (std::ptrdiff_t )BinFloat1::bit_count + b.exponent() + 1) { if ((eval_lsb(a.bits()) == BinFloat1::bit_count - 1) && (eval_lsb(b.bits()) != BinFloat1::bit_count - 1)) { eval_left_shift(dt, 1); eval_decrement(dt); res.exponent() = a.exponent() - 1; } else res.exponent() = a.exponent(); } else res.exponent() = a.exponent(); } else { dt = b.bits(); if (b.exponent() <= (std::ptrdiff_t )BinFloat1::bit_count + a.exponent()) { typename BinFloat1::exponent_type e_diff = a.exponent() - b.exponent(); eval_left_shift(dt, static_cast(-e_diff)); res.exponent() = b.exponent() + e_diff; eval_subtract(dt, a.bits()); } else if (b.exponent() == (std::ptrdiff_t )BinFloat1::bit_count + a.exponent() + 1) { if ((eval_lsb(a.bits()) != BinFloat1::bit_count - 1) && eval_lsb(b.bits())) { eval_left_shift(dt, 1); eval_decrement(dt); res.exponent() = b.exponent() - 1; } else res.exponent() = b.exponent(); } else res.exponent() = b.exponent(); s = !s; } copy_and_round(res, dt); if (res.exponent() == BinFloat1::exponent_zero) res.sign() = false; else if (res.sign() != s) res.negate(); res.check_invariants(); } template inline void eval_add(cpp_bin_float& res, const cpp_bin_float& a, const cpp_bin_float& b) { if (a.sign() == b.sign()) do_eval_add(res, a, b); else do_eval_subtract(res, a, b); } template inline void eval_add(cpp_bin_float& res, const cpp_bin_float& a) { return eval_add(res, res, a); } template inline void eval_subtract(cpp_bin_float& res, const cpp_bin_float& a, const cpp_bin_float& b) { if (a.sign() != b.sign()) do_eval_add(res, a, b); else do_eval_subtract(res, a, b); } template inline void eval_subtract(cpp_bin_float& res, const cpp_bin_float& a) { return eval_subtract(res, res, a); } template inline void eval_multiply(cpp_bin_float& res, const cpp_bin_float& a, const cpp_bin_float& b) { using default_ops::eval_bit_test; using default_ops::eval_multiply; // Special cases first: switch (a.exponent()) { case cpp_bin_float::exponent_zero: { if (b.exponent() == cpp_bin_float::exponent_nan) res = b; else if (b.exponent() == cpp_bin_float::exponent_infinity) res = std::numeric_limits > >::quiet_NaN().backend(); else { bool s = a.sign() != b.sign(); res = a; res.sign() = s; } return; } case cpp_bin_float::exponent_infinity: switch (b.exponent()) { case cpp_bin_float::exponent_zero: res = std::numeric_limits > >::quiet_NaN().backend(); break; case cpp_bin_float::exponent_nan: res = b; break; default: bool s = a.sign() != b.sign(); res = a; res.sign() = s; break; } return; case cpp_bin_float::exponent_nan: res = a; return; } if (b.exponent() > cpp_bin_float::max_exponent) { bool s = a.sign() != b.sign(); res = b; res.sign() = s; return; } if ((a.exponent() > 0) && (b.exponent() > 0)) { if (cpp_bin_float::max_exponent + 2 - a.exponent() < b.exponent()) { // We will certainly overflow: bool s = a.sign() != b.sign(); res.exponent() = cpp_bin_float::exponent_infinity; res.sign() = s; res.bits() = static_cast(0u); return; } } if ((a.exponent() < 0) && (b.exponent() < 0)) { if (cpp_bin_float::min_exponent - 2 - a.exponent() > b.exponent()) { // We will certainly underflow: res.exponent() = cpp_bin_float::exponent_zero; res.sign() = a.sign() != b.sign(); res.bits() = static_cast(0u); return; } } typename cpp_bin_float::double_rep_type dt; eval_multiply(dt, a.bits(), b.bits()); res.exponent() = a.exponent() + b.exponent() - (Exponent)cpp_bin_float::bit_count + 1; copy_and_round(res, dt); res.check_invariants(); res.sign() = a.sign() != b.sign(); } template inline void eval_multiply(cpp_bin_float& res, const cpp_bin_float& a) { eval_multiply(res, res, a); } template inline typename std::enable_if::value>::type eval_multiply(cpp_bin_float& res, const cpp_bin_float& a, const U& b) { using default_ops::eval_bit_test; using default_ops::eval_multiply; bool s = a.sign(); // saved for later in case a and res are the same object. // Special cases first: switch (a.exponent()) { case cpp_bin_float::exponent_zero: { res = a; res.sign() = s; return; } case cpp_bin_float::exponent_infinity: if (b == 0) res = std::numeric_limits > >::quiet_NaN().backend(); else res = a; return; case cpp_bin_float::exponent_nan: res = a; return; } typename cpp_bin_float::double_rep_type dt; using canon_ui_type = typename boost::multiprecision::detail::canonical::double_rep_type>::type; eval_multiply(dt, a.bits(), static_cast(b)); res.exponent() = a.exponent(); copy_and_round(res, dt); res.check_invariants(); res.sign() = s; } template inline typename std::enable_if::value>::type eval_multiply(cpp_bin_float& res, const U& b) { eval_multiply(res, res, b); } template inline typename std::enable_if::value && boost::multiprecision::detail::is_integral::value>::type eval_multiply(cpp_bin_float& res, const cpp_bin_float& a, const S& b) { using ui_type = typename boost::multiprecision::detail::make_unsigned::type; eval_multiply(res, a, static_cast(boost::multiprecision::detail::unsigned_abs(b))); if (b < 0) res.negate(); } template inline typename std::enable_if::value && boost::multiprecision::detail::is_integral::value>::type eval_multiply(cpp_bin_float& res, const S& b) { eval_multiply(res, res, b); } template inline void eval_divide(cpp_bin_float& res, const cpp_bin_float& u, const cpp_bin_float& v) { #ifdef BOOST_MSVC #pragma warning(push) #pragma warning(disable : 6326) // comparison of two constants #endif using default_ops::eval_bit_test; using default_ops::eval_get_sign; using default_ops::eval_increment; using default_ops::eval_qr; using default_ops::eval_subtract; // // Special cases first: // switch (u.exponent()) { case cpp_bin_float::exponent_zero: { switch (v.exponent()) { case cpp_bin_float::exponent_zero: case cpp_bin_float::exponent_nan: res = std::numeric_limits > >::quiet_NaN().backend(); return; } bool s = u.sign() != v.sign(); res = u; res.sign() = s; return; } case cpp_bin_float::exponent_infinity: { switch (v.exponent()) { case cpp_bin_float::exponent_infinity: case cpp_bin_float::exponent_nan: res = std::numeric_limits > >::quiet_NaN().backend(); return; } bool s = u.sign() != v.sign(); res = u; res.sign() = s; return; } case cpp_bin_float::exponent_nan: res = std::numeric_limits > >::quiet_NaN().backend(); return; } switch (v.exponent()) { case cpp_bin_float::exponent_zero: { bool s = u.sign() != v.sign(); res = std::numeric_limits > >::infinity().backend(); res.sign() = s; return; } case cpp_bin_float::exponent_infinity: res.exponent() = cpp_bin_float::exponent_zero; res.bits() = limb_type(0); res.sign() = u.sign() != v.sign(); return; case cpp_bin_float::exponent_nan: res = std::numeric_limits > >::quiet_NaN().backend(); return; } // We can scale u and v so that both are integers, then perform integer // division to obtain quotient q and remainder r, such that: // // q * v + r = u // // and hense: // // q + r/v = u/v // // From this, assuming q has cpp_bin_float::bit_count // bits we only need to determine whether // r/v is less than, equal to, or greater than 0.5 to determine rounding - // this we can do with a shift and comparison. // // We can set the exponent and sign of the result up front: // if ((v.exponent() < 0) && (u.exponent() > 0)) { // Check for overflow: if (cpp_bin_float::max_exponent + v.exponent() < u.exponent() - 1) { res.exponent() = cpp_bin_float::exponent_infinity; res.sign() = u.sign() != v.sign(); res.bits() = static_cast(0u); return; } } else if ((v.exponent() > 0) && (u.exponent() < 0)) { // Check for underflow: if (cpp_bin_float::min_exponent + v.exponent() > u.exponent()) { // We will certainly underflow: res.exponent() = cpp_bin_float::exponent_zero; res.sign() = u.sign() != v.sign(); res.bits() = static_cast(0u); return; } } res.exponent() = u.exponent() - v.exponent() - 1; res.sign() = u.sign() != v.sign(); // // Now get the quotient and remainder: // typename cpp_bin_float::double_rep_type t(u.bits()), t2(v.bits()), q, r; eval_left_shift(t, cpp_bin_float::bit_count); eval_qr(t, t2, q, r); // // We now have either "cpp_bin_float::bit_count" // or "cpp_bin_float::bit_count+1" significant // bits in q. // constexpr unsigned limb_bits = sizeof(limb_type) * CHAR_BIT; if (eval_bit_test(q, cpp_bin_float::bit_count)) { // // OK we have cpp_bin_float::bit_count+1 bits, // so we already have rounding info, // we just need to changes things if the last bit is 1 and either the // remainder is non-zero (ie we do not have a tie) or the quotient would // be odd if it were shifted to the correct number of bits (ie a tiebreak). // BOOST_MP_ASSERT((eval_msb(q) == cpp_bin_float::bit_count)); if ((q.limbs()[0] & 1u) && (eval_get_sign(r) || (q.limbs()[0] & 2u))) { eval_increment(q); } } else { // // We have exactly "cpp_bin_float::bit_count" bits in q. // Get rounding info, which we can get by comparing 2r with v. // We want to call copy_and_round to handle rounding and general cleanup, // so we'll left shift q and add some fake digits on the end to represent // how we'll be rounding. // using local_exponent_type = typename cpp_bin_float::exponent_type; BOOST_MP_ASSERT((eval_msb(q) == cpp_bin_float::bit_count - 1)); constexpr unsigned lshift = (cpp_bin_float::bit_count < limb_bits) ? 2 : limb_bits; eval_left_shift(q, lshift); res.exponent() -= static_cast(lshift); eval_left_shift(r, 1u); int c = r.compare(v.bits()); if (c == 0) q.limbs()[0] |= static_cast(1u) << (lshift - 1); else if (c > 0) q.limbs()[0] |= (static_cast(1u) << (lshift - 1)) + static_cast(1u); } copy_and_round(res, q); #ifdef BOOST_MSVC #pragma warning(pop) #endif } template inline void eval_divide(cpp_bin_float& res, const cpp_bin_float& arg) { eval_divide(res, res, arg); } template inline typename std::enable_if::value && (std::numeric_limits::digits <= Digits)>::type eval_divide(cpp_bin_float& res, const cpp_bin_float& u, const U& v) { #ifdef BOOST_MSVC #pragma warning(push) #pragma warning(disable : 6326) // comparison of two constants #endif using default_ops::eval_bit_test; using default_ops::eval_get_sign; using default_ops::eval_increment; using default_ops::eval_qr; using default_ops::eval_subtract; // // Special cases first: // switch (u.exponent()) { case cpp_bin_float::exponent_zero: { if (v == 0) { res = std::numeric_limits > >::quiet_NaN().backend(); return; } bool s = u.sign() != (v < 0); res = u; res.sign() = s; return; } case cpp_bin_float::exponent_infinity: res = u; return; case cpp_bin_float::exponent_nan: res = std::numeric_limits > >::quiet_NaN().backend(); return; } if (v == 0) { bool s = u.sign(); res = std::numeric_limits > >::infinity().backend(); res.sign() = s; return; } // We can scale u and v so that both are integers, then perform integer // division to obtain quotient q and remainder r, such that: // // q * v + r = u // // and hense: // // q + r/v = u/v // // From this, assuming q has "cpp_bin_float::bit_count" cpp_bin_float::bit_count, we only need to determine whether // r/v is less than, equal to, or greater than 0.5 to determine rounding - // this we can do with a shift and comparison. // // We can set the exponent and sign of the result up front: // std::ptrdiff_t gb = static_cast(msb(v)); res.exponent() = u.exponent() - static_cast(gb) - static_cast(1); res.sign() = u.sign(); // // Now get the quotient and remainder: // typename cpp_bin_float::double_rep_type t(u.bits()), q, r; eval_left_shift(t, static_cast(gb + 1)); eval_qr(t, number::double_rep_type>::canonical_value(v), q, r); // // We now have either "cpp_bin_float::bit_count" or "cpp_bin_float::bit_count+1" significant cpp_bin_float::bit_count in q. // constexpr unsigned limb_bits = sizeof(limb_type) * CHAR_BIT; if (eval_bit_test(q, cpp_bin_float::bit_count)) { // // OK we have cpp_bin_float::bit_count+1 cpp_bin_float::bit_count, so we already have rounding info, // we just need to changes things if the last bit is 1 and the // remainder is non-zero (ie we do not have a tie). // BOOST_MP_ASSERT((eval_msb(q) == cpp_bin_float::bit_count)); if ((q.limbs()[0] & 1u) && eval_get_sign(r)) { eval_increment(q); } } else { // // We have exactly "cpp_bin_float::bit_count" cpp_bin_float::bit_count in q. // Get rounding info, which we can get by comparing 2r with v. // We want to call copy_and_round to handle rounding and general cleanup, // so we'll left shift q and add some fake cpp_bin_float::bit_count on the end to represent // how we'll be rounding. // using local_exponent_type = typename cpp_bin_float::exponent_type; BOOST_MP_ASSERT((eval_msb(q) == cpp_bin_float::bit_count - 1)); constexpr unsigned lshift = cpp_bin_float::bit_count < limb_bits ? 2 : limb_bits; eval_left_shift(q, lshift); res.exponent() -= static_cast(lshift); eval_left_shift(r, 1u); int c = r.compare(number::double_rep_type>::canonical_value(v)); if (c == 0) q.limbs()[0] |= static_cast(1u) << (lshift - 1); else if (c > 0) q.limbs()[0] |= (static_cast(1u) << (lshift - 1)) + static_cast(1u); } copy_and_round(res, q); #ifdef BOOST_MSVC #pragma warning(pop) #endif } template inline typename std::enable_if::value && (std::numeric_limits::digits <= Digits)>::type eval_divide(cpp_bin_float& res, const U& v) { eval_divide(res, res, v); } template inline typename std::enable_if::value && boost::multiprecision::detail::is_integral::value && (std::numeric_limits::digits <= Digits)>::type eval_divide(cpp_bin_float& res, const cpp_bin_float& u, const S& v) { using ui_type = typename boost::multiprecision::detail::make_unsigned::type; eval_divide(res, u, static_cast(boost::multiprecision::detail::unsigned_abs(v))); if (v < 0) res.negate(); } template inline typename std::enable_if::value && boost::multiprecision::detail::is_integral::value && (std::numeric_limits::digits <= Digits)>::type eval_divide(cpp_bin_float& res, const S& v) { eval_divide(res, res, v); } template inline int eval_get_sign(const cpp_bin_float& arg) { return arg.exponent() == cpp_bin_float::exponent_zero ? 0 : arg.sign() ? -1 : 1; } template inline bool eval_is_zero(const cpp_bin_float& arg) { return arg.exponent() == cpp_bin_float::exponent_zero; } template inline bool eval_eq(const cpp_bin_float& a, cpp_bin_float& b) { if (a.exponent() == b.exponent()) { if (a.exponent() == cpp_bin_float::exponent_zero) return true; return (a.sign() == b.sign()) && (a.bits().compare(b.bits()) == 0) && (a.exponent() != cpp_bin_float::exponent_nan); } return false; } template inline void convert_to_signed_int(I* res, const cpp_bin_float& arg) { static constexpr int digits = std::numeric_limits::is_specialized ? std::numeric_limits::digits : sizeof(I) * CHAR_BIT - 1; static constexpr I max_val = std::numeric_limits::is_specialized ? (std::numeric_limits::max)() : (((I(1) << (sizeof(I) * CHAR_BIT - 2)) - 1) << 1) + 1; static constexpr I min_val = std::numeric_limits::is_specialized ? (std::numeric_limits::min)() : -max_val - 1; switch (arg.exponent()) { case cpp_bin_float::exponent_zero: *res = 0; return; case cpp_bin_float::exponent_nan: BOOST_MP_THROW_EXCEPTION(std::runtime_error("Could not convert NaN to integer.")); return; case cpp_bin_float::exponent_infinity: *res = max_val; if (arg.sign()) *res = -*res; return; } using shift_type = typename std::conditional::exponent_type) < sizeof(int), int, typename cpp_bin_float::exponent_type>::type; typename cpp_bin_float::rep_type man(arg.bits()); shift_type shift = (shift_type)cpp_bin_float::bit_count - 1 - arg.exponent(); if (shift > (shift_type)cpp_bin_float::bit_count - 1) { *res = 0; return; } if (arg.sign() && (arg.compare(min_val) <= 0)) { *res = min_val; return; } else if (!arg.sign() && (arg.compare(max_val) >= 0)) { *res = max_val; return; } if (shift < 0) { if (static_cast(cpp_bin_float::bit_count) - static_cast(shift) <= digits) { // We have more bits in long_long_type than the float, so it's OK to left shift: eval_convert_to(res, man); *res <<= -shift; } else { *res = (std::numeric_limits::max)(); return; } } else { eval_right_shift(man, static_cast(shift)); eval_convert_to(res, man); } if (arg.sign()) { *res = -*res; } } template inline void eval_convert_to(long long* res, const cpp_bin_float& arg) { convert_to_signed_int(res, arg); } #ifdef BOOST_HAS_INT128 template inline void eval_convert_to(int128_type* res, const cpp_bin_float& arg) { convert_to_signed_int(res, arg); } #endif template inline void convert_to_unsigned_int(I* res, const cpp_bin_float& arg) { static constexpr int digits = std::numeric_limits::is_specialized ? std::numeric_limits::digits : sizeof(I) * CHAR_BIT; static constexpr I max_val = std::numeric_limits::is_specialized ? (std::numeric_limits::max)() : ~static_cast(0); switch (arg.exponent()) { case cpp_bin_float::exponent_zero: *res = 0; return; case cpp_bin_float::exponent_nan: BOOST_MP_THROW_EXCEPTION(std::runtime_error("Could not convert NaN to integer.")); case cpp_bin_float::exponent_infinity: *res = max_val; return; } typename cpp_bin_float::rep_type man(arg.bits()); using shift_type = typename std::conditional::exponent_type) < sizeof(int), int, typename cpp_bin_float::exponent_type>::type; shift_type shift = (shift_type)cpp_bin_float::bit_count - 1 - arg.exponent(); if (shift > (shift_type)cpp_bin_float::bit_count - 1) { *res = 0; return; } else if (shift < 0) { if (cpp_bin_float::bit_count - shift <= digits) { // We have more bits in ulong_long_type than the float, so it's OK to left shift: eval_convert_to(res, man); *res <<= -shift; return; } *res = max_val; return; } eval_right_shift(man, shift); eval_convert_to(res, man); } template inline void eval_convert_to(unsigned long long* res, const cpp_bin_float& arg) { convert_to_unsigned_int(res, arg); } #ifdef BOOST_HAS_INT128 template inline void eval_convert_to(uint128_type* res, const cpp_bin_float& arg) { convert_to_unsigned_int(res, arg); } #endif template inline typename std::enable_if::value>::type eval_convert_to(Float* res, const cpp_bin_float& original_arg) { using conv_type = cpp_bin_float::digits, digit_base_2, void, Exponent, MinE, MaxE>; using common_exp_type = typename std::common_type::type; static constexpr int float_digits = boost::multiprecision::detail::is_float128::value ? 113 : std::numeric_limits::digits; BOOST_MP_FLOAT128_USING using std::ldexp; // // Special cases first: // switch (original_arg.exponent()) { case cpp_bin_float::exponent_zero: *res = 0; if (original_arg.sign()) *res = -*res; return; case cpp_bin_float::exponent_nan: BOOST_IF_CONSTEXPR(boost::multiprecision::detail::is_float128::value) { *res = static_cast(std::numeric_limits::quiet_NaN()); } else { *res = std::numeric_limits::quiet_NaN(); } return; case cpp_bin_float::exponent_infinity: BOOST_IF_CONSTEXPR(boost::multiprecision::detail::is_float128::value) { *res = static_cast((std::numeric_limits::infinity)()); } else { *res = (std::numeric_limits::infinity)(); } if (original_arg.sign()) *res = -*res; return; } // // Check for super large exponent that must be converted to infinity: // if (original_arg.exponent() > (boost::multiprecision::detail::is_float128::value ? 16384 : std::numeric_limits::max_exponent)) { BOOST_IF_CONSTEXPR(boost::multiprecision::detail::is_float128::value) { *res = static_cast(std::numeric_limits::infinity()); } else { *res = std::numeric_limits::has_infinity ? std::numeric_limits::infinity() : (std::numeric_limits::max)(); } if (original_arg.sign()) *res = -*res; return; } // // Figure out how many digits we will have in our result, // allowing for a possibly denormalized result: // common_exp_type digits_to_round_to = float_digits; if (original_arg.exponent() < std::numeric_limits::min_exponent - 1) { common_exp_type diff = original_arg.exponent(); diff -= boost::multiprecision::detail::is_float128::value ? -16382 : std::numeric_limits::min_exponent - 1; digits_to_round_to += diff; } if (digits_to_round_to < 0) { // Result must be zero: *res = 0; if (original_arg.sign()) *res = -*res; return; } // // Perform rounding first, then afterwards extract the digits: // cpp_bin_float(float_digits), digit_base_2, Allocator, Exponent, 0, 0> arg; typename cpp_bin_float::rep_type bits(original_arg.bits()); arg.exponent() = original_arg.exponent(); copy_and_round(arg, bits, (std::ptrdiff_t)digits_to_round_to); common_exp_type e = arg.exponent(); e -= static_cast(cpp_bin_float::bit_count) - 1; constexpr std::size_t limbs_needed = static_cast(float_digits) / (sizeof(*arg.bits().limbs()) * CHAR_BIT) + (static_cast(float_digits) % (sizeof(*arg.bits().limbs()) * CHAR_BIT) ? 1 : 0); std::size_t first_limb_needed = arg.bits().size() - limbs_needed; *res = 0; e += static_cast(first_limb_needed * sizeof(*arg.bits().limbs()) * CHAR_BIT); while (first_limb_needed < arg.bits().size()) { *res += ldexp(static_cast(arg.bits().limbs()[first_limb_needed]), static_cast(e)); ++first_limb_needed; e += static_cast(sizeof(*arg.bits().limbs()) * CHAR_BIT); } if (original_arg.sign()) *res = -*res; } template inline void eval_frexp(cpp_bin_float& res, const cpp_bin_float& arg, Exponent* e) { switch (arg.exponent()) { case cpp_bin_float::exponent_zero: case cpp_bin_float::exponent_nan: case cpp_bin_float::exponent_infinity: *e = 0; res = arg; return; } res = arg; *e = arg.exponent() + 1; res.exponent() = -1; } template inline void eval_frexp(cpp_bin_float& res, const cpp_bin_float& arg, I* pe) { Exponent e; eval_frexp(res, arg, &e); if ((e > (std::numeric_limits::max)()) || (e < (std::numeric_limits::min)())) { BOOST_MP_THROW_EXCEPTION(std::runtime_error("Exponent was outside of the range of the argument type to frexp.")); } *pe = static_cast(e); } template inline void eval_ldexp(cpp_bin_float& res, const cpp_bin_float& arg, Exponent e) { switch (arg.exponent()) { case cpp_bin_float::exponent_zero: case cpp_bin_float::exponent_nan: case cpp_bin_float::exponent_infinity: res = arg; return; } if ((e > 0) && (cpp_bin_float::max_exponent - e < arg.exponent())) { // Overflow: res = std::numeric_limits > >::infinity().backend(); res.sign() = arg.sign(); } else if ((e < 0) && (cpp_bin_float::min_exponent - e > arg.exponent())) { // Underflow: res = limb_type(0); } else { res = arg; res.exponent() += e; } } template inline typename std::enable_if::value>::type eval_ldexp(cpp_bin_float& res, const cpp_bin_float& arg, I e) { using si_type = typename boost::multiprecision::detail::make_signed::type; if (e > static_cast((std::numeric_limits::max)())) res = std::numeric_limits > >::infinity().backend(); else eval_ldexp(res, arg, static_cast(e)); } template inline typename std::enable_if::value && boost::multiprecision::detail::is_integral::value>::type eval_ldexp(cpp_bin_float& res, const cpp_bin_float& arg, I e) { if ((e > (std::numeric_limits::max)()) || (e < (std::numeric_limits::min)())) { res = std::numeric_limits > >::infinity().backend(); if (e < 0) res.negate(); } else eval_ldexp(res, arg, static_cast(e)); } /* * Sign manipulation */ template inline void eval_abs(cpp_bin_float& res, const cpp_bin_float& arg) { res = arg; res.sign() = false; } template inline void eval_abs(cpp_bin_float& res, const cpp_bin_float& arg) { res = arg; res.sign() = false; } template inline void eval_fabs(cpp_bin_float& res, const cpp_bin_float& arg) { res = arg; res.sign() = false; } template inline void eval_fabs(cpp_bin_float& res, const cpp_bin_float& arg) { res = arg; res.sign() = false; } template inline int eval_fpclassify(const cpp_bin_float& arg) { switch (arg.exponent()) { case cpp_bin_float::exponent_zero: return FP_ZERO; case cpp_bin_float::exponent_infinity: return FP_INFINITE; case cpp_bin_float::exponent_nan: return FP_NAN; } return FP_NORMAL; } template inline void eval_sqrt(cpp_bin_float& res, const cpp_bin_float& arg) { using default_ops::eval_bit_test; using default_ops::eval_increment; using default_ops::eval_integer_sqrt; switch (arg.exponent()) { case cpp_bin_float::exponent_nan: errno = EDOM; // fallthrough... case cpp_bin_float::exponent_zero: res = arg; return; case cpp_bin_float::exponent_infinity: if (arg.sign()) { res = std::numeric_limits > >::quiet_NaN().backend(); errno = EDOM; } else res = arg; return; } if (arg.sign()) { res = std::numeric_limits > >::quiet_NaN().backend(); errno = EDOM; return; } typename cpp_bin_float::double_rep_type t(arg.bits()), r, s; eval_left_shift(t, arg.exponent() & 1 ? cpp_bin_float::bit_count : cpp_bin_float::bit_count - 1); eval_integer_sqrt(s, r, t); if (!eval_bit_test(s, cpp_bin_float::bit_count)) { // We have exactly the right number of cpp_bin_float::bit_count in the result, round as required: if (s.compare(r) < 0) { eval_increment(s); } } typename cpp_bin_float::exponent_type ae = arg.exponent(); res.exponent() = ae / 2; res.sign() = false; if ((ae & 1) && (ae < 0)) --res.exponent(); copy_and_round(res, s); } template inline void eval_floor(cpp_bin_float& res, const cpp_bin_float& arg) { using default_ops::eval_increment; switch (arg.exponent()) { case cpp_bin_float::exponent_nan: errno = EDOM; // fallthrough... case cpp_bin_float::exponent_zero: case cpp_bin_float::exponent_infinity: res = arg; return; } using shift_type = typename std::conditional::exponent_type) < sizeof(int), int, typename cpp_bin_float::exponent_type>::type; shift_type shift = (shift_type)cpp_bin_float::bit_count - arg.exponent() - 1; if ((arg.exponent() > (shift_type)cpp_bin_float::max_exponent) || (shift <= 0)) { // Either arg is already an integer, or a special value: res = arg; return; } if (shift >= (shift_type)cpp_bin_float::bit_count) { res = static_cast(arg.sign() ? -1 : 0); return; } bool fractional = (shift_type)eval_lsb(arg.bits()) < shift; res = arg; eval_right_shift(res.bits(), static_cast(shift)); if (fractional && res.sign()) { eval_increment(res.bits()); const std::ptrdiff_t shift_check = static_cast(static_cast(cpp_bin_float::bit_count) - 1 - static_cast(shift)); if (static_cast(eval_msb(res.bits())) != shift_check) { // Must have extended result by one bit in the increment: --shift; ++res.exponent(); } } eval_left_shift(res.bits(), static_cast(shift)); } template inline void eval_ceil(cpp_bin_float& res, const cpp_bin_float& arg) { using default_ops::eval_increment; switch (arg.exponent()) { case cpp_bin_float::exponent_infinity: errno = EDOM; // fallthrough... case cpp_bin_float::exponent_zero: case cpp_bin_float::exponent_nan: res = arg; return; } using shift_type = typename std::conditional::exponent_type) < sizeof(int), int, typename cpp_bin_float::exponent_type>::type; shift_type shift = (shift_type)cpp_bin_float::bit_count - arg.exponent() - 1; if ((arg.exponent() > (shift_type)cpp_bin_float::max_exponent) || (shift <= 0)) { // Either arg is already an integer, or a special value: res = arg; return; } if (shift >= (shift_type)cpp_bin_float::bit_count) { bool s = arg.sign(); // takes care of signed zeros res = static_cast(arg.sign() ? 0 : 1); res.sign() = s; return; } bool fractional = (shift_type)eval_lsb(arg.bits()) < shift; res = arg; eval_right_shift(res.bits(), shift); if (fractional && !res.sign()) { eval_increment(res.bits()); if ((std::ptrdiff_t)eval_msb(res.bits()) != cpp_bin_float::bit_count - 1 - shift) { // Must have extended result by one bit in the increment: --shift; ++res.exponent(); } } eval_left_shift(res.bits(), shift); } template int eval_signbit(const cpp_bin_float& val) { return val.sign(); } template inline std::size_t hash_value(const cpp_bin_float& val) { std::size_t result = hash_value(val.bits()); boost::multiprecision::detail::hash_combine(result, val.exponent(), val.sign()); return result; } } // namespace backends namespace detail { template struct transcendental_reduction_type > { // // The type used for trigonometric reduction needs 3 times the precision of the base type. // This is double the precision of the original type, plus the largest exponent supported. // As a practical measure the largest argument supported is 1/eps, as supporting larger // arguments requires the division of argument by PI/2 to also be done at higher precision, // otherwise the result (an integer) can not be represented exactly. // // See ARGUMENT REDUCTION FOR HUGE ARGUMENTS. K C Ng. // using type = boost::multiprecision::backends::cpp_bin_float< boost::multiprecision::backends::cpp_bin_float::bit_count * 3, boost::multiprecision::backends::digit_base_2, Allocator, Exponent, MinExponent, MaxExponent>; }; #ifdef BOOST_HAS_INT128 template struct is_convertible_arithmetic > : public std::true_type {}; template struct is_convertible_arithmetic > : public std::true_type {}; #endif } // namespace detail template inline boost::multiprecision::number, ExpressionTemplates> copysign BOOST_PREVENT_MACRO_SUBSTITUTION( const boost::multiprecision::number, ExpressionTemplates>& a, const boost::multiprecision::number, ExpressionTemplates>& b) { boost::multiprecision::number, ExpressionTemplates> res(a); res.backend().sign() = b.backend().sign(); return res; } template struct number_category > : public std::integral_constant {}; template struct is_equivalent_number_type, cpp_bin_float > : public std::integral_constant {}; } // namespace multiprecision namespace math { using boost::multiprecision::copysign; using boost::multiprecision::signbit; } // namespace math } // namespace boost #include #include namespace std { // // numeric_limits [partial] specializations for the types declared in this header: // template class numeric_limits, ExpressionTemplates> > { using number_type = boost::multiprecision::number, ExpressionTemplates>; private: // // Functions to calculate cached values stored in static values: // static number_type get_min() { using ui_type = typename std::tuple_element<0, typename number_type::backend_type::unsigned_types>::type; number_type value(ui_type(1u)); value.backend().exponent() = boost::multiprecision::cpp_bin_float::min_exponent; return value; } #ifdef BOOST_MSVC #pragma warning(push) #pragma warning(disable : 4127) // conditional expression is constant #endif static number_type get_max() { number_type value; BOOST_IF_CONSTEXPR(std::is_void::value) eval_complement(value.backend().bits(), value.backend().bits()); else { // We jump through hoops here using the backend type directly just to keep VC12 happy // (ie compiler workaround, for very strange compiler bug): using boost::multiprecision::default_ops::eval_add; using boost::multiprecision::default_ops::eval_decrement; using boost::multiprecision::default_ops::eval_left_shift; using int_backend_type = typename number_type::backend_type::rep_type; using ui_type = typename std::tuple_element<0, typename int_backend_type::unsigned_types>::type; int_backend_type i; i = ui_type(1u); eval_left_shift(i, boost::multiprecision::cpp_bin_float::bit_count - 1); int_backend_type j(i); eval_decrement(i); eval_add(j, i); value.backend().bits() = j; } value.backend().exponent() = boost::multiprecision::cpp_bin_float::max_exponent; return value; } #ifdef BOOST_MSVC #pragma warning(pop) #endif static number_type get_epsilon() { using ui_type = typename std::tuple_element<0, typename number_type::backend_type::unsigned_types>::type; number_type value(ui_type(1u)); return ldexp(value, 1 - static_cast(digits)); } // What value should this be???? static number_type get_round_error() { // returns 0.5 return ldexp(number_type(1u), -1); } static number_type get_infinity() { number_type value; value.backend().exponent() = boost::multiprecision::cpp_bin_float::exponent_infinity; return value; } static number_type get_quiet_NaN() { number_type value; value.backend().exponent() = boost::multiprecision::cpp_bin_float::exponent_nan; return value; } public: static constexpr bool is_specialized = true; static number_type(min)() { // C++11 thread safe static initialization: static number_type value = get_min(); return value; } static number_type(max)() { // C++11 thread safe static initialization: static number_type value = get_max(); return value; } static constexpr number_type lowest() { return -(max)(); } static constexpr int digits = boost::multiprecision::cpp_bin_float::bit_count; static constexpr int digits10 = boost::multiprecision::detail::calc_digits10(digits)>::value; // Is this really correct??? static constexpr int max_digits10 = boost::multiprecision::detail::calc_max_digits10(digits)>::value; static constexpr bool is_signed = true; static constexpr bool is_integer = false; static constexpr bool is_exact = false; static constexpr int radix = 2; static number_type epsilon() { // C++11 thread safe static initialization: static number_type value = get_epsilon(); return value; } // What value should this be???? static number_type round_error() { // returns 0.5 // C++11 thread safe static initialization: static number_type value = get_round_error(); return value; } static constexpr typename boost::multiprecision::cpp_bin_float::exponent_type min_exponent = boost::multiprecision::cpp_bin_float::min_exponent; static constexpr typename boost::multiprecision::cpp_bin_float::exponent_type min_exponent10 = (min_exponent / 1000) * 301L; static constexpr typename boost::multiprecision::cpp_bin_float::exponent_type max_exponent = boost::multiprecision::cpp_bin_float::max_exponent; static constexpr typename boost::multiprecision::cpp_bin_float::exponent_type max_exponent10 = (max_exponent / 1000) * 301L; static constexpr bool has_infinity = true; static constexpr bool has_quiet_NaN = true; static constexpr bool has_signaling_NaN = false; #ifdef _MSC_VER #pragma warning(push) #pragma warning(disable:4996) #endif static constexpr float_denorm_style has_denorm = denorm_absent; #ifdef _MSC_VER #pragma warning(pop) #endif static constexpr bool has_denorm_loss = false; static number_type infinity() { // C++11 thread safe static initialization: static number_type value = get_infinity(); return value; } static number_type quiet_NaN() { // C++11 thread safe static initialization: static number_type value = get_quiet_NaN(); return value; } static constexpr number_type signaling_NaN() { return number_type(0); } static constexpr number_type denorm_min() { return get_min(); } static constexpr bool is_iec559 = false; static constexpr bool is_bounded = true; static constexpr bool is_modulo = false; static constexpr bool traps = true; static constexpr bool tinyness_before = false; static constexpr float_round_style round_style = round_to_nearest; }; template constexpr int numeric_limits, ExpressionTemplates> >::digits; template constexpr int numeric_limits, ExpressionTemplates> >::digits10; template constexpr int numeric_limits, ExpressionTemplates> >::max_digits10; template constexpr bool numeric_limits, ExpressionTemplates> >::is_signed; template constexpr bool numeric_limits, ExpressionTemplates> >::is_integer; template constexpr bool numeric_limits, ExpressionTemplates> >::is_exact; template constexpr int numeric_limits, ExpressionTemplates> >::radix; template constexpr typename boost::multiprecision::cpp_bin_float::exponent_type numeric_limits, ExpressionTemplates> >::min_exponent; template constexpr typename boost::multiprecision::cpp_bin_float::exponent_type numeric_limits, ExpressionTemplates> >::min_exponent10; template constexpr typename boost::multiprecision::cpp_bin_float::exponent_type numeric_limits, ExpressionTemplates> >::max_exponent; template constexpr typename boost::multiprecision::cpp_bin_float::exponent_type numeric_limits, ExpressionTemplates> >::max_exponent10; template constexpr bool numeric_limits, ExpressionTemplates> >::has_infinity; template constexpr bool numeric_limits, ExpressionTemplates> >::has_quiet_NaN; template constexpr bool numeric_limits, ExpressionTemplates> >::has_signaling_NaN; #ifdef _MSC_VER #pragma warning(push) #pragma warning(disable:4996) #endif template constexpr float_denorm_style numeric_limits, ExpressionTemplates> >::has_denorm; #ifdef _MSC_VER #pragma warning(pop) #endif template constexpr bool numeric_limits, ExpressionTemplates> >::has_denorm_loss; template constexpr bool numeric_limits, ExpressionTemplates> >::is_iec559; template constexpr bool numeric_limits, ExpressionTemplates> >::is_bounded; template constexpr bool numeric_limits, ExpressionTemplates> >::is_modulo; template constexpr bool numeric_limits, ExpressionTemplates> >::traps; template constexpr bool numeric_limits, ExpressionTemplates> >::tinyness_before; template constexpr float_round_style numeric_limits, ExpressionTemplates> >::round_style; } // namespace std #endif