/* * // Copyright (c) Radzivon Bartoshyk 7/2025. All rights reserved. * // * // Redistribution and use in source and binary forms, with or without modification, * // are permitted provided that the following conditions are met: * // * // 1. Redistributions of source code must retain the above copyright notice, this * // list of conditions and the following disclaimer. * // * // 2. Redistributions in binary form must reproduce the above copyright notice, * // this list of conditions and the following disclaimer in the documentation * // and/or other materials provided with the distribution. * // * // 3. Neither the name of the copyright holder nor the names of its * // contributors may be used to endorse or promote products derived from * // this software without specific prior written permission. * // * // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" * // AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * // IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE * // DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE * // FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * // DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR * // SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER * // CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, * // OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE * // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ use crate::bessel::alpha1::{ bessel_1_asympt_alpha, bessel_1_asympt_alpha_fast, bessel_1_asympt_alpha_hard, }; use crate::bessel::beta1::{ bessel_1_asympt_beta, bessel_1_asympt_beta_fast, bessel_1_asympt_beta_hard, }; use crate::bessel::i0::bessel_rsqrt_hard; use crate::bessel::y1_coeffs::Y1_COEFFS_REMEZ; use crate::bessel::y1_coeffs_taylor::Y1_COEFFS; use crate::bessel::y1f_coeffs::{Y1_ZEROS, Y1_ZEROS_VALUES}; use crate::common::f_fmla; use crate::double_double::DoubleDouble; use crate::dyadic_float::{DyadicFloat128, DyadicSign}; use crate::logs::log_dd_fast; use crate::polyeval::{f_polyeval12, f_polyeval13, f_polyeval15, f_polyeval22, f_polyeval24}; use crate::rounding::CpuCeil; use crate::sin_helper::{cos_dd_small, cos_dd_small_fast, cos_f128_small}; use crate::sincos_reduce::{AngleReduced, rem2pi_any, rem2pi_f128}; /// Bessel of the second kind of order 1 ( Y1 ) pub fn f_y1(x: f64) -> f64 { let ix = x.to_bits(); if ix >= 0x7ffu64 << 52 || ix == 0 { // |x| == NaN, x == inf, |x| == 0, x < 0 if ix.wrapping_shl(1) == 0 { // |x| == 0 return f64::NEG_INFINITY; } if x.is_infinite() { if x.is_sign_negative() { return f64::NAN; } return 0.; } return x + f64::NAN; // x == NaN } let xb = x.to_bits(); if xb <= 0x4049c00000000000u64 { // x <= 51.5 if xb <= 0x4000000000000000u64 { // x <= 2 if xb <= 0x3ff75c28f5c28f5cu64 { // x <= 1.46 return y1_near_zero_fast(x); } // transient zone from 1.46 to 2 have bad behavior for log poly already, // and not yet good to be easily covered, thus it use its own poly return y1_transient_zone_fast(x); } return y1_small_argument_fast(x); } y1_asympt_fast(x) } /** Generated by SageMath: Evaluates: y2 = -J1(x)*log(x) + 1/x * (1 - sum((-1)^m*(H(m)+H(m-1))/(2^m*m!*(m-1)!)*x^(2*m)) Y1(x) = 2/pi*(-y2(x)+(euler_gamma - log(2))*J1(x)) expressed as: Y1(x)=log(x)*W1(x) - Z1(x) - 2/(pi*x) ```python from sage.all import * R = LaurentSeriesRing(RealField(300), 'x', default_prec=300) x = R.gen() N = 16 # Number of terms (adjust as needed) gamma = RealField(300)(euler_gamma) d2 = RealField(300)(2) pi = RealField(300).pi() log2 = RealField(300)(2).log() def j_series(n, x): return sum([(-1)**m * (x/2)**(ZZ(n) + ZZ(2)*ZZ(m)) / (ZZ(m).factorial() * (ZZ(m) + ZZ(n)).factorial()) for m in range(N)]) J1_series = j_series(1, x) def harmony(m): return sum(RealField(300)(1)/RealField(300)(k) for k in range(1, m+1)) def z_series(x): return sum([(-1)**m * (x)**(ZZ(2)*ZZ(m)) / (ZZ(2)**(2*m) * ZZ(m).factorial() * (ZZ(m) - ZZ(1)).factorial()) * (harmony(m) + harmony(m - 1)) for m in range(1, N)]) W1 = d2/pi * J1_series Z1 = -(d2/(x*pi) * z_series(x) + d2/pi * gamma * J1_series(x) - d2/pi * log2 * J1_series(x)) def y1_full(x): return d2/pi * (J1_series(x) * x.log() - 1/x * ( 1 - z_series(x)) + (gamma - log2) * J1_series(x)) # see the series print(W0) print(Z0) ``` See ./notes/bessel_y1_taylor.ipynb for generation **/ #[inline] fn y1_near_zero_fast(x: f64) -> f64 { const W: [(u64, u64); 15] = [ (0xbc76b01ec5417056, 0x3fd45f306dc9c883), (0x3c46b01ec5417056, 0xbfa45f306dc9c883), (0xbbfe40290701eb1e, 0x3f5b2995e7b7b604), (0xbba67fe4a5feb897, 0xbf021bb945252402), (0xbb0334914cdd2ba9, 0x3e9cf9286ea1d337), (0x3aab4f3c6d42c1f4, 0xbe2ee7a29824147f), (0xba407fb57ef4dc2c, 0x3db78be9987d036d), (0x39d2921e91b07dd0, 0xbd3ae90af76a4d0f), (0x395a28c8620dc90e, 0x3cb7eb97f85e7d62), (0xb8cf83f52abe45c5, 0xbc31028e3376648a), (0xb8441050c68ca435, 0x3ba3cb1e7d0c17e7), (0xb7ab072548a1aa43, 0xbb133191ed9f1eef), (0xb6f05192c2d9b6ee, 0x3a7f7f4b5e8ef7b0), (0x367ad65afe306d57, 0xb9e626e36cb3515d), (0xb5ea1c4136f8f230, 0x394b01153dce6810), ]; let x2 = DoubleDouble::from_exact_mult(x, x); let w0 = f_polyeval12( x2.hi, f64::from_bits(W[3].1), f64::from_bits(W[4].1), f64::from_bits(W[5].1), f64::from_bits(W[6].1), f64::from_bits(W[7].1), f64::from_bits(W[8].1), f64::from_bits(W[9].1), f64::from_bits(W[10].1), f64::from_bits(W[11].1), f64::from_bits(W[12].1), f64::from_bits(W[13].1), f64::from_bits(W[14].1), ); let mut w = DoubleDouble::mul_f64_add(x2, w0, DoubleDouble::from_bit_pair(W[2])); w = DoubleDouble::mul_add(x2, w, DoubleDouble::from_bit_pair(W[1])); w = DoubleDouble::mul_add(x2, w, DoubleDouble::from_bit_pair(W[0])); w = DoubleDouble::quick_mult_f64(w, x); const Z: [(u64, u64); 15] = [ (0x3c61d7eb2e54cda1, 0x3fc91866143cbc8a), (0xbc2f9f7a0ce54a40, 0xbfabd3975c75b4a7), (0xbbf7659313f45e8c, 0x3f6835b97894be5b), (0x3b9cbcd40f1be7b9, 0xbf12c7dbffcde97d), (0xbb495d78778645b4, 0x3eb0a780ac776eac), (0xbae15be86455c1ab, 0xbe432e5a4ddeea30), (0xba5ad966c12f1e3c, 0x3dcf0ce34d2066a6), (0x39e9717155dc7521, 0xbd52a4e1aea45c18), (0x394f447fe5de1290, 0x3cd1474ade9154ac), (0xb8e1699d9009a7fc, 0xbc4978ba84f218c0), (0xb8505502096ead17, 0x3bbe9598c016378b), (0x37942b6c36b2c5f1, 0xbb2e7e5fcfc4b7b1), (0x37210853b78bd08a, 0x3a99a6c1266c116d), (0xb686c9639c9d976e, 0xba02738998fe7337), (0xb603b739ee04b9fe, 0x3966f58cd41b6d08), ]; let z0 = f_polyeval12( x2.hi, f64::from_bits(Z[3].1), f64::from_bits(Z[4].1), f64::from_bits(Z[5].1), f64::from_bits(Z[6].1), f64::from_bits(Z[7].1), f64::from_bits(Z[8].1), f64::from_bits(Z[9].1), f64::from_bits(Z[10].1), f64::from_bits(Z[11].1), f64::from_bits(Z[12].1), f64::from_bits(Z[13].1), f64::from_bits(Z[14].1), ); let mut z = DoubleDouble::mul_f64_add(x2, z0, DoubleDouble::from_bit_pair(Z[2])); z = DoubleDouble::mul_add(x2, z, DoubleDouble::from_bit_pair(Z[1])); z = DoubleDouble::mul_add(x2, z, DoubleDouble::from_bit_pair(Z[0])); z = DoubleDouble::quick_mult_f64(z, x); let w_log = log_dd_fast(x); const MINUS_TWO_OVER_PI: DoubleDouble = DoubleDouble::from_bit_pair((0x3c86b01ec5417056, 0xbfe45f306dc9c883)); let m_two_over_pi_div_x: DoubleDouble; #[cfg(any( all( any(target_arch = "x86", target_arch = "x86_64"), target_feature = "fma" ), target_arch = "aarch64" ))] { m_two_over_pi_div_x = DoubleDouble::div_dd_f64(MINUS_TWO_OVER_PI, x) } #[cfg(not(any( all( any(target_arch = "x86", target_arch = "x86_64"), target_feature = "fma" ), target_arch = "aarch64" )))] { use crate::double_double::two_product_compatible; m_two_over_pi_div_x = if two_product_compatible(x) { DoubleDouble::div_dd_f64(MINUS_TWO_OVER_PI, x) } else { DoubleDouble::div_safe_dd_f64(MINUS_TWO_OVER_PI, x) }; } if m_two_over_pi_div_x.hi.is_infinite() { return f64::NEG_INFINITY; } let zvp = DoubleDouble::mul_add(w, w_log, -z); let p = DoubleDouble::add(m_two_over_pi_div_x, zvp); let err = f_fmla( p.hi, f64::from_bits(0x3c30000000000000), // 2^-60 f64::from_bits(0x3be0000000000000), // 2^-65 ); let ub = p.hi + (p.lo + err); let lb = p.hi + (p.lo - err); if ub == lb { return p.to_f64(); } y1_near_zero(x, w_log) } /** Generated by SageMath: Evaluates: y2 = -J1(x)*log(x) + 1/x * (1 - sum((-1)^m*(H(m)+H(m-1))/(2^m*m!*(m-1)!)*x^(2*m)) Y1(x) = 2/pi*(-y2(x)+(euler_gamma - log(2))*J1(x)) expressed as: Y1(x)=log(x)*W1(x) - Z1(x) - 2/(pi*x) ```python from sage.all import * R = LaurentSeriesRing(RealField(300), 'x', default_prec=300) x = R.gen() N = 16 # Number of terms (adjust as needed) gamma = RealField(300)(euler_gamma) d2 = RealField(300)(2) pi = RealField(300).pi() log2 = RealField(300)(2).log() def j_series(n, x): return sum([(-1)**m * (x/2)**(ZZ(n) + ZZ(2)*ZZ(m)) / (ZZ(m).factorial() * (ZZ(m) + ZZ(n)).factorial()) for m in range(N)]) J1_series = j_series(1, x) def harmony(m): return sum(RealField(300)(1)/RealField(300)(k) for k in range(1, m+1)) def z_series(x): return sum([(-1)**m * (x)**(ZZ(2)*ZZ(m)) / (ZZ(2)**(2*m) * ZZ(m).factorial() * (ZZ(m) - ZZ(1)).factorial()) * (harmony(m) + harmony(m - 1)) for m in range(1, N)]) W1 = d2/pi * J1_series Z1 = -(d2/(x*pi) * z_series(x) + d2/pi * gamma * J1_series(x) - d2/pi * log2 * J1_series(x)) def y1_full(x): return d2/pi * (J1_series(x) * x.log() - 1/x * ( 1 - z_series(x)) + (gamma - log2) * J1_series(x)) # see the series print(W0) print(Z0) ``` See ./notes/bessel_y1_taylor.ipynb for generation **/ #[cold] #[inline(never)] fn y1_near_zero(x: f64, w_log: DoubleDouble) -> f64 { const W: [(u64, u64); 15] = [ (0xbc76b01ec5417056, 0x3fd45f306dc9c883), (0x3c46b01ec5417056, 0xbfa45f306dc9c883), (0xbbfe40290701eb1e, 0x3f5b2995e7b7b604), (0xbba67fe4a5feb897, 0xbf021bb945252402), (0xbb0334914cdd2ba9, 0x3e9cf9286ea1d337), (0x3aab4f3c6d42c1f4, 0xbe2ee7a29824147f), (0xba407fb57ef4dc2c, 0x3db78be9987d036d), (0x39d2921e91b07dd0, 0xbd3ae90af76a4d0f), (0x395a28c8620dc90e, 0x3cb7eb97f85e7d62), (0xb8cf83f52abe45c5, 0xbc31028e3376648a), (0xb8441050c68ca435, 0x3ba3cb1e7d0c17e7), (0xb7ab072548a1aa43, 0xbb133191ed9f1eef), (0xb6f05192c2d9b6ee, 0x3a7f7f4b5e8ef7b0), (0x367ad65afe306d57, 0xb9e626e36cb3515d), (0xb5ea1c4136f8f230, 0x394b01153dce6810), ]; let x2 = DoubleDouble::from_exact_mult(x, x); let mut w = f_polyeval15( x2, DoubleDouble::from_bit_pair(W[0]), DoubleDouble::from_bit_pair(W[1]), DoubleDouble::from_bit_pair(W[2]), DoubleDouble::from_bit_pair(W[3]), DoubleDouble::from_bit_pair(W[4]), DoubleDouble::from_bit_pair(W[5]), DoubleDouble::from_bit_pair(W[6]), DoubleDouble::from_bit_pair(W[7]), DoubleDouble::from_bit_pair(W[8]), DoubleDouble::from_bit_pair(W[9]), DoubleDouble::from_bit_pair(W[10]), DoubleDouble::from_bit_pair(W[11]), DoubleDouble::from_bit_pair(W[12]), DoubleDouble::from_bit_pair(W[13]), DoubleDouble::from_bit_pair(W[14]), ); w = DoubleDouble::quick_mult_f64(w, x); const Z: [(u64, u64); 15] = [ (0x3c61d7eb2e54cda1, 0x3fc91866143cbc8a), (0xbc2f9f7a0ce54a40, 0xbfabd3975c75b4a7), (0xbbf7659313f45e8c, 0x3f6835b97894be5b), (0x3b9cbcd40f1be7b9, 0xbf12c7dbffcde97d), (0xbb495d78778645b4, 0x3eb0a780ac776eac), (0xbae15be86455c1ab, 0xbe432e5a4ddeea30), (0xba5ad966c12f1e3c, 0x3dcf0ce34d2066a6), (0x39e9717155dc7521, 0xbd52a4e1aea45c18), (0x394f447fe5de1290, 0x3cd1474ade9154ac), (0xb8e1699d9009a7fc, 0xbc4978ba84f218c0), (0xb8505502096ead17, 0x3bbe9598c016378b), (0x37942b6c36b2c5f1, 0xbb2e7e5fcfc4b7b1), (0x37210853b78bd08a, 0x3a99a6c1266c116d), (0xb686c9639c9d976e, 0xba02738998fe7337), (0xb603b739ee04b9fe, 0x3966f58cd41b6d08), ]; let mut z = f_polyeval15( x2, DoubleDouble::from_bit_pair(Z[0]), DoubleDouble::from_bit_pair(Z[1]), DoubleDouble::from_bit_pair(Z[2]), DoubleDouble::from_bit_pair(Z[3]), DoubleDouble::from_bit_pair(Z[4]), DoubleDouble::from_bit_pair(Z[5]), DoubleDouble::from_bit_pair(Z[6]), DoubleDouble::from_bit_pair(Z[7]), DoubleDouble::from_bit_pair(Z[8]), DoubleDouble::from_bit_pair(Z[9]), DoubleDouble::from_bit_pair(Z[10]), DoubleDouble::from_bit_pair(Z[11]), DoubleDouble::from_bit_pair(Z[12]), DoubleDouble::from_bit_pair(Z[13]), DoubleDouble::from_bit_pair(Z[14]), ); z = DoubleDouble::quick_mult_f64(z, x); const MINUS_TWO_OVER_PI: DoubleDouble = DoubleDouble::from_bit_pair((0x3c86b01ec5417056, 0xbfe45f306dc9c883)); let m_two_over_pi_div_x: DoubleDouble; #[cfg(any( all( any(target_arch = "x86", target_arch = "x86_64"), target_feature = "fma" ), target_arch = "aarch64" ))] { m_two_over_pi_div_x = DoubleDouble::div_dd_f64(MINUS_TWO_OVER_PI, x) } #[cfg(not(any( all( any(target_arch = "x86", target_arch = "x86_64"), target_feature = "fma" ), target_arch = "aarch64" )))] { use crate::double_double::two_product_compatible; m_two_over_pi_div_x = if two_product_compatible(x) { DoubleDouble::div_dd_f64(MINUS_TWO_OVER_PI, x) } else { DoubleDouble::div_safe_dd_f64(MINUS_TWO_OVER_PI, x) }; } if m_two_over_pi_div_x.hi.is_infinite() { return f64::NEG_INFINITY; } let zvp = DoubleDouble::mul_add(w, w_log, -z); DoubleDouble::full_dd_add(m_two_over_pi_div_x, zvp).to_f64() } #[inline] fn y1_transient_zone_fast(x: f64) -> f64 { /* <120] poly=error[[1]]; coeffs=CoefficientList[poly,x]; TableForm[Table[Row[{"'",NumberForm[coeffs[[i+1]],{50,50}, ExponentFunction->(Null&)],"',"}],{i,0,Length[coeffs]-1}]] */ const C: [(u64, u64); 28] = [ (0x38d23216c8eac2ee, 0x3c631de77e55e9b0), (0x3c77f3a6718f8e03, 0x3fe0aa48442f0150), (0xbc52b77d09be8679, 0xbfbe56f82217b33a), (0x3c237af39fc9d759, 0xbfa0d2af4e922afe), (0xbc08241e95389bc3, 0xbf73a6dec2f9f739), (0x3c18eac6a35acd63, 0x3f7e671c82a1c956), (0x3be0a0f3c8908083, 0xbf65429d5dbc3bb0), (0x3be217fa58861600, 0x3f517abad71c26c0), (0x3bebc327d6c65514, 0xbf40b28b4ef33a56), (0xbbcb597d6bdd5992, 0x3f2ef0d150ec9934), (0x3bb1a480de696e07, 0xbf1c0758a86844be), (0xbb9c6ea6352ab84e, 0x3f0b7c88b58d9ef3), (0x3b7dbb78dfd868a9, 0xbee97ef9519bdcfd), (0xbb8d4519030f499d, 0x3f04815f08e7ad5e), (0x3bbbd70e1480e260, 0x3f10f348483b57bc), (0x3b7117cf4d2b6f3c, 0x3f231914389bb1bb), (0x3b8ca48beaf6a58d, 0x3f30b29e838345b4), (0x3bccfac65ce17cf9, 0x3f39a69e98f61897), (0xbbeae7e3065b09c9, 0x3f40b9511666fcf0), (0xbbe5cbddf691e7e6, 0x3f428cd8388e634b), (0xbbd91372412d1e1b, 0x3f414ba048d9e1d5), (0xbbb0781a70c6f715, 0x3f3acdcf66f1de95), (0xbba3ae83fd425494, 0x3f30f44ae6620bba), (0x3bc001d75da77b74, 0x3f21154a0a1f2161), (0xbb91c9afb1a1b874, 0x3f0a687b664cbac6), (0x3b8e0a06b9444963, 0x3eed7c3cbb4ba5d8), (0x3b4f0e9dfc915934, 0x3ec53ca23fdd0999), (0xbb0409258f8ffca8, 0x3e8de620acb51b2d), ]; // this poly relative to first zero const ZERO: DoubleDouble = DoubleDouble::from_bit_pair((0xbc8bd1e50d219bfd, 0x400193bed4dff243)); let mut r = DoubleDouble::full_add_f64(-ZERO, x); r = DoubleDouble::from_exact_add(r.hi, r.lo); let p0 = f_polyeval24( r.to_f64(), f64::from_bits(C[4].1), f64::from_bits(C[5].1), f64::from_bits(C[6].1), f64::from_bits(C[7].1), f64::from_bits(C[8].1), f64::from_bits(C[9].1), f64::from_bits(C[10].1), f64::from_bits(C[11].1), f64::from_bits(C[12].1), f64::from_bits(C[13].1), f64::from_bits(C[14].1), f64::from_bits(C[15].1), f64::from_bits(C[16].1), f64::from_bits(C[17].1), f64::from_bits(C[18].1), f64::from_bits(C[19].1), f64::from_bits(C[20].1), f64::from_bits(C[21].1), f64::from_bits(C[22].1), f64::from_bits(C[23].1), f64::from_bits(C[24].1), f64::from_bits(C[25].1), f64::from_bits(C[26].1), f64::from_bits(C[27].1), ); let mut p = DoubleDouble::mul_f64_add(r, p0, DoubleDouble::from_bit_pair(C[3])); p = DoubleDouble::mul_add(p, r, DoubleDouble::from_bit_pair(C[2])); p = DoubleDouble::mul_add(p, r, DoubleDouble::from_bit_pair(C[1])); p = DoubleDouble::mul_add(p, r, DoubleDouble::from_bit_pair(C[0])); let err = f_fmla( p.hi, f64::from_bits(0x3c50000000000000), // 2^-58 f64::from_bits(0x3b90000000000000), // 2^-70 ); let ub = p.hi + (p.lo + err); let lb = p.hi + (p.lo - err); if ub == lb { return p.to_f64(); } y1_transient_zone(x) } fn y1_transient_zone(x: f64) -> f64 { /* <120] poly=error[[1]]; coeffs=CoefficientList[poly,x]; TableForm[Table[Row[{"'",NumberForm[coeffs[[i+1]],{50,50}, ExponentFunction->(Null&)],"',"}],{i,0,Length[coeffs]-1}]] */ const C: [(u64, u64); 28] = [ (0x38d23216c8eac2ee, 0x3c631de77e55e9b0), (0x3c77f3a6718f8e03, 0x3fe0aa48442f0150), (0xbc52b77d09be8679, 0xbfbe56f82217b33a), (0x3c237af39fc9d759, 0xbfa0d2af4e922afe), (0xbc08241e95389bc3, 0xbf73a6dec2f9f739), (0x3c18eac6a35acd63, 0x3f7e671c82a1c956), (0x3be0a0f3c8908083, 0xbf65429d5dbc3bb0), (0x3be217fa58861600, 0x3f517abad71c26c0), (0x3bebc327d6c65514, 0xbf40b28b4ef33a56), (0xbbcb597d6bdd5992, 0x3f2ef0d150ec9934), (0x3bb1a480de696e07, 0xbf1c0758a86844be), (0xbb9c6ea6352ab84e, 0x3f0b7c88b58d9ef3), (0x3b7dbb78dfd868a9, 0xbee97ef9519bdcfd), (0xbb8d4519030f499d, 0x3f04815f08e7ad5e), (0x3bbbd70e1480e260, 0x3f10f348483b57bc), (0x3b7117cf4d2b6f3c, 0x3f231914389bb1bb), (0x3b8ca48beaf6a58d, 0x3f30b29e838345b4), (0x3bccfac65ce17cf9, 0x3f39a69e98f61897), (0xbbeae7e3065b09c9, 0x3f40b9511666fcf0), (0xbbe5cbddf691e7e6, 0x3f428cd8388e634b), (0xbbd91372412d1e1b, 0x3f414ba048d9e1d5), (0xbbb0781a70c6f715, 0x3f3acdcf66f1de95), (0xbba3ae83fd425494, 0x3f30f44ae6620bba), (0x3bc001d75da77b74, 0x3f21154a0a1f2161), (0xbb91c9afb1a1b874, 0x3f0a687b664cbac6), (0x3b8e0a06b9444963, 0x3eed7c3cbb4ba5d8), (0x3b4f0e9dfc915934, 0x3ec53ca23fdd0999), (0xbb0409258f8ffca8, 0x3e8de620acb51b2d), ]; // this poly relative to first zero const ZERO: DoubleDouble = DoubleDouble::from_bit_pair((0xbc8bd1e50d219bfd, 0x400193bed4dff243)); let r = DoubleDouble::full_add_f64(-ZERO, x); let p0 = f_polyeval13( r.to_f64(), f64::from_bits(C[15].1), f64::from_bits(C[16].1), f64::from_bits(C[17].1), f64::from_bits(C[18].1), f64::from_bits(C[19].1), f64::from_bits(C[20].1), f64::from_bits(C[21].1), f64::from_bits(C[22].1), f64::from_bits(C[23].1), f64::from_bits(C[24].1), f64::from_bits(C[25].1), f64::from_bits(C[26].1), f64::from_bits(C[27].1), ); let mut p_e = DoubleDouble::mul_f64_add(r, p0, DoubleDouble::from_bit_pair(C[14])); p_e = DoubleDouble::mul_add(p_e, r, DoubleDouble::from_bit_pair(C[13])); p_e = DoubleDouble::mul_add(p_e, r, DoubleDouble::from_bit_pair(C[12])); p_e = DoubleDouble::mul_add(p_e, r, DoubleDouble::from_bit_pair(C[11])); p_e = DoubleDouble::mul_add(p_e, r, DoubleDouble::from_bit_pair(C[10])); p_e = DoubleDouble::mul_add(p_e, r, DoubleDouble::from_bit_pair(C[9])); p_e = DoubleDouble::mul_add(p_e, r, DoubleDouble::from_bit_pair(C[8])); p_e = DoubleDouble::mul_add(p_e, r, DoubleDouble::from_bit_pair(C[7])); p_e = DoubleDouble::mul_add(p_e, r, DoubleDouble::from_bit_pair(C[6])); p_e = DoubleDouble::mul_add(p_e, r, DoubleDouble::from_bit_pair(C[5])); p_e = DoubleDouble::mul_add(p_e, r, DoubleDouble::from_bit_pair(C[4])); p_e = DoubleDouble::mul_add(p_e, r, DoubleDouble::from_bit_pair(C[3])); p_e = DoubleDouble::mul_add(p_e, r, DoubleDouble::from_bit_pair(C[2])); p_e = DoubleDouble::mul_add(p_e, r, DoubleDouble::from_bit_pair(C[1])); p_e = DoubleDouble::mul_add(p_e, r, DoubleDouble::from_bit_pair(C[0])); let p = DoubleDouble::from_exact_add(p_e.hi, p_e.lo); let err = f_fmla( p.hi, f64::from_bits(0x3c20000000000000), // 2^-61 f64::from_bits(0x3a90000000000000), // 2^-86 ); let ub = p.hi + (p.lo + err); let lb = p.hi + (p.lo - err); if ub == lb { return p.to_f64(); } y1_transient_hard(x) } #[cold] #[inline(never)] fn y1_transient_hard(x: f64) -> f64 { /* <120] poly=error[[1]]; coeffs=CoefficientList[poly,x]; TableForm[Table[Row[{"'",NumberForm[coeffs[[i+1]],{50,50}, ExponentFunction->(Null&)],"',"}],{i,0,Length[coeffs]-1}]] */ pub(crate) static C: [DyadicFloat128; 28] = [ DyadicFloat128 { sign: DyadicSign::Pos, exponent: -184, mantissa: 0x98ef3bf2_af4d8048_c85b23ab_0bb72488_u128, }, DyadicFloat128 { sign: DyadicSign::Pos, exponent: -128, mantissa: 0x85524221_780a817f_3a6718f8_e03513ec_u128, }, DyadicFloat128 { sign: DyadicSign::Neg, exponent: -131, mantissa: 0xf2b7c110_bd99d256_efa137d0_cf1f7988_u128, }, DyadicFloat128 { sign: DyadicSign::Neg, exponent: -132, mantissa: 0x86957a74_9157ef64_286301b1_453792e2_u128, }, DyadicFloat128 { sign: DyadicSign::Neg, exponent: -135, mantissa: 0x9d36f617_cfb9c982_41e95389_bc32e8c2_u128, }, DyadicFloat128 { sign: DyadicSign::Pos, exponent: -135, mantissa: 0xf338e415_0e4ab31d_58d46b59_ac6792eb_u128, }, DyadicFloat128 { sign: DyadicSign::Neg, exponent: -136, mantissa: 0xaa14eaed_e1dd7f7a_f861bb7b_fbe6acb5_u128, }, DyadicFloat128 { sign: DyadicSign::Pos, exponent: -137, mantissa: 0x8bd5d6b8_e1360121_7fa58861_5ff9b614_u128, }, DyadicFloat128 { sign: DyadicSign::Neg, exponent: -138, mantissa: 0x85945a77_99d2ac87_9b052735_5d7f5f59_u128, }, DyadicFloat128 { sign: DyadicSign::Pos, exponent: -140, mantissa: 0xf7868a87_64c99c94_d0528454_cdccd44c_u128, }, DyadicFloat128 { sign: DyadicSign::Neg, exponent: -141, mantissa: 0xe03ac543_4225edcb_6fe432d2_3f11a8b0_u128, }, DyadicFloat128 { sign: DyadicSign::Pos, exponent: -142, mantissa: 0xdbe445ac_6cf79639_159cad54_7b1eda7c_u128, }, DyadicFloat128 { sign: DyadicSign::Neg, exponent: -144, mantissa: 0xcbf7ca8c_dee7e624_48720279_75757a17_u128, }, DyadicFloat128 { sign: DyadicSign::Pos, exponent: -142, mantissa: 0xa40af847_3d6aef15_d737e785_b3163322_u128, }, DyadicFloat128 { sign: DyadicSign::Pos, exponent: -141, mantissa: 0x879a4241_dabde37a_e1c2901c_4c06b1dc_u128, }, DyadicFloat128 { sign: DyadicSign::Pos, exponent: -140, mantissa: 0x98c8a1c4_dd8dd811_17cf4d2b_6f3c3910_u128, }, DyadicFloat128 { sign: DyadicSign::Pos, exponent: -139, mantissa: 0x8594f41c_1a2da01c_a48beaf6_a58cef9f_u128, }, DyadicFloat128 { sign: DyadicSign::Pos, exponent: -139, mantissa: 0xcd34f4c7_b0c4b9cf_ac65ce17_cf940c9c_u128, }, DyadicFloat128 { sign: DyadicSign::Pos, exponent: -138, mantissa: 0x85ca88b3_37e77ca3_039f349e_c6ddb06d_u128, }, DyadicFloat128 { sign: DyadicSign::Pos, exponent: -138, mantissa: 0x9466c1c4_731a5546_84412dc3_033a8ee7_u128, }, DyadicFloat128 { sign: DyadicSign::Pos, exponent: -138, mantissa: 0x8a5d0246_cf0ea66e_c8dbed2e_1e50b14f_u128, }, DyadicFloat128 { sign: DyadicSign::Pos, exponent: -139, mantissa: 0xd66e7b37_8ef4a77c_3f2c79c8_4757a40d_u128, }, DyadicFloat128 { sign: DyadicSign::Pos, exponent: -139, mantissa: 0x87a25733_105dcfb1_45f00af6_adb11b5f_u128, }, DyadicFloat128 { sign: DyadicSign::Pos, exponent: -140, mantissa: 0x88aa5050_f90b0a00_3aebb4ef_6e7e9ce1_u128, }, DyadicFloat128 { sign: DyadicSign::Pos, exponent: -142, mantissa: 0xd343db32_65d62ee3_6504e5e4_78c55965_u128, }, DyadicFloat128 { sign: DyadicSign::Pos, exponent: -144, mantissa: 0xebe1e5da_5d2ec3c1_40d72889_2c57e483_u128, }, DyadicFloat128 { sign: DyadicSign::Pos, exponent: -146, mantissa: 0xa9e511fe_e84cc8f8_74efe48a_c9a09ebc_u128, }, DyadicFloat128 { sign: DyadicSign::Pos, exponent: -150, mantissa: 0xef310565_a8d9675f_b6d38380_1abf92a6_u128, }, ]; const ZERO: DyadicFloat128 = DyadicFloat128 { sign: DyadicSign::Pos, exponent: -126, mantissa: 0x8c9df6a6_ff921721_70d796f3_2017e155_u128, }; let r = DyadicFloat128::new_from_f64(x) - ZERO; let mut p = C[27]; for i in (0..27).rev() { p = r * p + C[i]; } p.fast_as_f64() } /// This method on small range searches for nearest zero or extremum. /// Then picks stored series expansion at the point end evaluates the poly at the point. #[inline] pub(crate) fn y1_small_argument_fast(x: f64) -> f64 { // let avg_step = 51.03 / 33.0; // let inv_step = 1.0 / avg_step; // // println!("inv_step {}", inv_step); const INV_STEP: f64 = 0.6466784244562023; let fx = x * INV_STEP; const Y1_ZEROS_COUNT: f64 = (Y1_ZEROS.len() - 1) as f64; let idx0 = unsafe { fx.min(Y1_ZEROS_COUNT).to_int_unchecked::() }; let idx1 = unsafe { fx.cpu_ceil() .min(Y1_ZEROS_COUNT) .to_int_unchecked::() }; let found_zero0 = DoubleDouble::from_bit_pair(Y1_ZEROS[idx0]); let found_zero1 = DoubleDouble::from_bit_pair(Y1_ZEROS[idx1]); let dist0 = (found_zero0.hi - x).abs(); let dist1 = (found_zero1.hi - x).abs(); let (found_zero, idx, dist) = if dist0 < dist1 { (found_zero0, idx0, dist0) } else { (found_zero1, idx1, dist1) }; if idx == 0 { return y1_near_zero_fast(x); } let close_to_zero = dist.abs() < 1e-3; let c = if close_to_zero { &Y1_COEFFS[idx - 1] } else { &Y1_COEFFS_REMEZ[idx - 1] }; let r = DoubleDouble::full_add_f64(-found_zero, x); // We hit exact zero, value, better to return it directly if dist == 0. { return f64::from_bits(Y1_ZEROS_VALUES[idx]); } let p = f_polyeval22( r.hi, f64::from_bits(c[6].1), f64::from_bits(c[7].1), f64::from_bits(c[8].1), f64::from_bits(c[9].1), f64::from_bits(c[10].1), f64::from_bits(c[11].1), f64::from_bits(c[12].1), f64::from_bits(c[13].1), f64::from_bits(c[14].1), f64::from_bits(c[15].1), f64::from_bits(c[16].1), f64::from_bits(c[17].1), f64::from_bits(c[18].1), f64::from_bits(c[19].1), f64::from_bits(c[20].1), f64::from_bits(c[21].1), f64::from_bits(c[22].1), f64::from_bits(c[23].1), f64::from_bits(c[24].1), f64::from_bits(c[25].1), f64::from_bits(c[26].1), f64::from_bits(c[27].1), ); let mut z = DoubleDouble::mul_f64_add(r, p, DoubleDouble::from_bit_pair(c[5])); z = DoubleDouble::mul_add(z, r, DoubleDouble::from_bit_pair(c[4])); z = DoubleDouble::mul_add(z, r, DoubleDouble::from_bit_pair(c[3])); z = DoubleDouble::mul_add(z, r, DoubleDouble::from_bit_pair(c[2])); z = DoubleDouble::mul_add(z, r, DoubleDouble::from_bit_pair(c[1])); z = DoubleDouble::mul_add(z, r, DoubleDouble::from_bit_pair(c[0])); let p = z; let err = f_fmla( p.hi, f64::from_bits(0x3c60000000000000), // 2^-57 f64::from_bits(0x3c20000000000000), // 2^-61 ); let ub = p.hi + (p.lo + err); let lb = p.hi + (p.lo - err); if ub == lb { return p.to_f64(); } y0_small_argument_moderate(r, c) } fn y0_small_argument_moderate(r: DoubleDouble, c0: &[(u64, u64); 28]) -> f64 { let c = &c0[15..]; let p0 = f_polyeval13( r.to_f64(), f64::from_bits(c[0].1), f64::from_bits(c[1].1), f64::from_bits(c[2].1), f64::from_bits(c[3].1), f64::from_bits(c[4].1), f64::from_bits(c[5].1), f64::from_bits(c[6].1), f64::from_bits(c[7].1), f64::from_bits(c[8].1), f64::from_bits(c[9].1), f64::from_bits(c[10].1), f64::from_bits(c[11].1), f64::from_bits(c[12].1), ); let c = c0; let mut p_e = DoubleDouble::mul_f64_add(r, p0, DoubleDouble::from_bit_pair(c[14])); p_e = DoubleDouble::mul_add(p_e, r, DoubleDouble::from_bit_pair(c[13])); p_e = DoubleDouble::mul_add(p_e, r, DoubleDouble::from_bit_pair(c[12])); p_e = DoubleDouble::mul_add(p_e, r, DoubleDouble::from_bit_pair(c[11])); p_e = DoubleDouble::mul_add(p_e, r, DoubleDouble::from_bit_pair(c[10])); p_e = DoubleDouble::mul_add(p_e, r, DoubleDouble::from_bit_pair(c[9])); p_e = DoubleDouble::mul_add(p_e, r, DoubleDouble::from_bit_pair(c[8])); p_e = DoubleDouble::mul_add(p_e, r, DoubleDouble::from_bit_pair(c[7])); p_e = DoubleDouble::mul_add(p_e, r, DoubleDouble::from_bit_pair(c[6])); p_e = DoubleDouble::mul_add(p_e, r, DoubleDouble::from_bit_pair(c[5])); p_e = DoubleDouble::mul_add(p_e, r, DoubleDouble::from_bit_pair(c[4])); p_e = DoubleDouble::mul_add(p_e, r, DoubleDouble::from_bit_pair(c[3])); p_e = DoubleDouble::mul_add(p_e, r, DoubleDouble::from_bit_pair(c[2])); p_e = DoubleDouble::mul_add(p_e, r, DoubleDouble::from_bit_pair(c[1])); p_e = DoubleDouble::mul_add(p_e, r, DoubleDouble::from_bit_pair(c[0])); let p = DoubleDouble::from_exact_add(p_e.hi, p_e.lo); let err = f_fmla( p.hi, f64::from_bits(0x3c30000000000000), // 2^-60 f64::from_bits(0x3bf0000000000000), // 2^-64 ); let ub = p.hi + (p.lo + err); let lb = p.hi + (p.lo - err); if ub == lb { return p.to_f64(); } y1_small_argument_hard(r, c) } #[cold] #[inline(never)] fn y1_small_argument_hard(r: DoubleDouble, c: &[(u64, u64); 28]) -> f64 { // if we're too close to zero taylor will converge faster and more accurate, // since remez, minimax and other almost cannot polynomial optimize near zero let mut p = DoubleDouble::from_bit_pair(c[27]); for i in (0..27).rev() { p = DoubleDouble::mul_add(r, p, DoubleDouble::from_bit_pair(c[i])); p = DoubleDouble::from_exact_add(p.hi, p.lo); } p.to_f64() } /* Evaluates: Y1 = sqrt(2/(PI*x)) * beta(x) * sin(x - 3*PI/4 - alpha(x)) Discarding 1/2*PI gives: Y1 = sqrt(2/(PI*x)) * beta(x) * (-cos(x - PI/4 - alpha(x))) */ #[inline] pub(crate) fn y1_asympt_fast(x: f64) -> f64 { const SQRT_2_OVER_PI: DoubleDouble = DoubleDouble::new( f64::from_bits(0xbc8cbc0d30ebfd15), f64::from_bits(0x3fe9884533d43651), ); const MPI_OVER_4: DoubleDouble = DoubleDouble::new( f64::from_bits(0xbc81a62633145c07), f64::from_bits(0xbfe921fb54442d18), ); let recip = if x.to_bits() > 0x7fd000000000000u64 { DoubleDouble::quick_mult_f64(DoubleDouble::from_exact_div_fma(4.0, x), 0.25) } else { DoubleDouble::from_recip(x) }; let alpha = bessel_1_asympt_alpha_fast(recip); let beta = bessel_1_asympt_beta_fast(recip); let AngleReduced { angle } = rem2pi_any(x); // Without full subtraction cancellation happens sometimes let x0pi34 = DoubleDouble::full_dd_sub(MPI_OVER_4, alpha); let r0 = DoubleDouble::full_dd_add(angle, x0pi34); let m_cos = -cos_dd_small_fast(r0); let z0 = DoubleDouble::quick_mult(beta, m_cos); let r_sqrt = DoubleDouble::from_rsqrt_fast(x); let scale = DoubleDouble::quick_mult(SQRT_2_OVER_PI, r_sqrt); let r = DoubleDouble::quick_mult(scale, z0); let p = DoubleDouble::from_exact_add(r.hi, r.lo); let err = f_fmla( p.hi, f64::from_bits(0x3c40000000000000), // 2^-60 f64::from_bits(0x3bf0000000000000), // 2^-87 ); let ub = p.hi + (p.lo + err); let lb = p.hi + (p.lo - err); if ub == lb { return p.to_f64(); } y1_asympt(x, recip, r_sqrt, angle) } /* Evaluates: Y1 = sqrt(2/(PI*x)) * beta(x) * sin(x - 3*PI/4 - alpha(x)) Discarding 1/2*PI gives: Y1 = sqrt(2/(PI*x)) * beta(x) * (-cos(x - PI/4 - alpha(x))) */ fn y1_asympt(x: f64, recip: DoubleDouble, r_sqrt: DoubleDouble, angle: DoubleDouble) -> f64 { const SQRT_2_OVER_PI: DoubleDouble = DoubleDouble::new( f64::from_bits(0xbc8cbc0d30ebfd15), f64::from_bits(0x3fe9884533d43651), ); const MPI_OVER_4: DoubleDouble = DoubleDouble::new( f64::from_bits(0xbc81a62633145c07), f64::from_bits(0xbfe921fb54442d18), ); let alpha = bessel_1_asympt_alpha(recip); let beta = bessel_1_asympt_beta(recip); // Without full subtraction cancellation happens sometimes let x0pi34 = DoubleDouble::full_dd_sub(MPI_OVER_4, alpha); let r0 = DoubleDouble::full_dd_add(angle, x0pi34); let m_cos = -cos_dd_small(r0); let z0 = DoubleDouble::quick_mult(beta, m_cos); let scale = DoubleDouble::quick_mult(SQRT_2_OVER_PI, r_sqrt); let r = DoubleDouble::quick_mult(scale, z0); let p = DoubleDouble::from_exact_add(r.hi, r.lo); let err = f_fmla( p.hi, f64::from_bits(0x3c30000000000000), // 2^-60 f64::from_bits(0x3a80000000000000), // 2^-87 ); let ub = p.hi + (p.lo + err); let lb = p.hi + (p.lo - err); if ub == lb { return p.to_f64(); } y1_asympt_hard(x) } /* Evaluates: Y1 = sqrt(2/(PI*x)) * beta(x) * sin(x - 3*PI/4 - alpha(x)) Discarding 1/2*PI gives: Y1 = sqrt(2/(PI*x)) * beta(x) * (-cos(x - PI/4 - alpha(x))) */ #[cold] #[inline(never)] fn y1_asympt_hard(x: f64) -> f64 { const SQRT_2_OVER_PI: DyadicFloat128 = DyadicFloat128 { sign: DyadicSign::Pos, exponent: -128, mantissa: 0xcc42299e_a1b28468_7e59e280_5d5c7180_u128, }; const MPI_OVER_4: DyadicFloat128 = DyadicFloat128 { sign: DyadicSign::Neg, exponent: -128, mantissa: 0xc90fdaa2_2168c234_c4c6628b_80dc1cd1_u128, }; let x_dyadic = DyadicFloat128::new_from_f64(x); let recip = DyadicFloat128::accurate_reciprocal(x); let alpha = bessel_1_asympt_alpha_hard(recip); let beta = bessel_1_asympt_beta_hard(recip); let angle = rem2pi_f128(x_dyadic); let x0pi34 = MPI_OVER_4 - alpha; let r0 = angle + x0pi34; let m_cos = cos_f128_small(r0).negated(); let z0 = beta * m_cos; let r_sqrt = bessel_rsqrt_hard(x, recip); let scale = SQRT_2_OVER_PI * r_sqrt; let p = scale * z0; p.fast_as_f64() } #[cfg(test)] mod tests { use super::*; #[test] fn test_y1() { // ULP should be less than 0.500001, but it was 0.5089558379720955, on 2.1957931471395398 result -0.0007023285780874727, using f_y1 and MPFR -0.0007023285780874729 assert_eq!(f_y1(0.0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000007291282546733975), -873124540555277200000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000.); assert_eq!(f_y1(2.1957931471395398), -0.0007023285780874727); assert_eq!( f_y1(f64::from_bits(0x571a31ffe2ff7e9fu64)), f64::from_bits(0x32e58532f95056ffu64) ); assert_eq!( f_y1(f64::from_bits(0x400193bed4dff243)), 0.00000000000000002513306678922122 ); assert_eq!( f_y1(f64::from_bits(0x3ffc513c569fe78e)), -0.24189760895998239 ); assert_eq!( f_y1(f64::from_bits(0x4192391e4c8faa60)), -0.000000000000000002572292246748134 ); assert_eq!( f_y1(f64::from_bits(0x403e9e480605283c)), -0.00000000000000001524456280251315 ); assert_eq!( f_y1(f64::from_bits(0x40277f9138d43206)), 0.000000000000000006849807120770496 ); assert_eq!(f_y1(f64::INFINITY), 0.); assert!(f_y1(f64::NEG_INFINITY).is_nan()); assert!(f_y1(f64::NAN).is_nan()); } #[test] fn test_y1_edge_cases() { assert_eq!(f_y1(2.1904620854463985), -0.0034837351616785234); assert_eq!(f_y1(2.197142201034536), 4.5568985277260593e-7); assert_eq!(f_y1(1.4000000000000004), -0.4791469742327998); assert_eq!(f_y1(2.0002288794493848), -0.1069033735586767); } }