MathFunctionsImpl.h
Go to the documentation of this file.
1 // This file is part of Eigen, a lightweight C++ template library
2 // for linear algebra.
3 //
4 // Copyright (C) 2014 Pedro Gonnet (pedro.gonnet@gmail.com)
5 // Copyright (C) 2016 Gael Guennebaud <gael.guennebaud@inria.fr>
6 //
7 // This Source Code Form is subject to the terms of the Mozilla
8 // Public License v. 2.0. If a copy of the MPL was not distributed
9 // with this file, You can obtain one at http://mozilla.org/MPL/2.0/.
10 
11 #ifndef EIGEN_MATHFUNCTIONSIMPL_H
12 #define EIGEN_MATHFUNCTIONSIMPL_H
13 
14 namespace Eigen {
15 
16 namespace internal {
17 
25 template<typename T>
26 T generic_fast_tanh_float(const T& a_x)
27 {
28  // Clamp the inputs to the range [-9, 9] since anything outside
29  // this range is +/-1.0f in single-precision.
30  const T plus_9 = pset1<T>(9.f);
31  const T minus_9 = pset1<T>(-9.f);
32  // NOTE GCC prior to 6.3 might improperly optimize this max/min
33  // step such that if a_x is nan, x will be either 9 or -9,
34  // and tanh will return 1 or -1 instead of nan.
35  // This is supposed to be fixed in gcc6.3,
36  // see: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=72867
37  const T x = pmax(minus_9,pmin(plus_9,a_x));
38  // The monomial coefficients of the numerator polynomial (odd).
39  const T alpha_1 = pset1<T>(4.89352455891786e-03f);
40  const T alpha_3 = pset1<T>(6.37261928875436e-04f);
41  const T alpha_5 = pset1<T>(1.48572235717979e-05f);
42  const T alpha_7 = pset1<T>(5.12229709037114e-08f);
43  const T alpha_9 = pset1<T>(-8.60467152213735e-11f);
44  const T alpha_11 = pset1<T>(2.00018790482477e-13f);
45  const T alpha_13 = pset1<T>(-2.76076847742355e-16f);
46 
47  // The monomial coefficients of the denominator polynomial (even).
48  const T beta_0 = pset1<T>(4.89352518554385e-03f);
49  const T beta_2 = pset1<T>(2.26843463243900e-03f);
50  const T beta_4 = pset1<T>(1.18534705686654e-04f);
51  const T beta_6 = pset1<T>(1.19825839466702e-06f);
52 
53  // Since the polynomials are odd/even, we need x^2.
54  const T x2 = pmul(x, x);
55 
56  // Evaluate the numerator polynomial p.
57  T p = pmadd(x2, alpha_13, alpha_11);
58  p = pmadd(x2, p, alpha_9);
59  p = pmadd(x2, p, alpha_7);
60  p = pmadd(x2, p, alpha_5);
61  p = pmadd(x2, p, alpha_3);
62  p = pmadd(x2, p, alpha_1);
63  p = pmul(x, p);
64 
65  // Evaluate the denominator polynomial p.
66  T q = pmadd(x2, beta_6, beta_4);
67  q = pmadd(x2, q, beta_2);
68  q = pmadd(x2, q, beta_0);
69 
70  // Divide the numerator by the denominator.
71  return pdiv(p, q);
72 }
73 
74 } // end namespace internal
75 
76 } // end namespace Eigen
77 
78 #endif // EIGEN_MATHFUNCTIONSIMPL_H
static int f(const TensorMap< Tensor< int, 3 > > &tensor)
Definition: LDLT.h:16
T generic_fast_tanh_float(const T &a_x)
EIGEN_DEVICE_FUNC Packet pmin(const Packet &a, const Packet &b)
EIGEN_DEVICE_FUNC const Scalar & q
EIGEN_DEVICE_FUNC Packet pdiv(const Packet &a, const Packet &b)
EIGEN_STRONG_INLINE Packet4f pmadd(const Packet4f &a, const Packet4f &b, const Packet4f &c)
EIGEN_DEVICE_FUNC Packet pmul(const Packet &a, const Packet &b)
EIGEN_DEVICE_FUNC Packet pmax(const Packet &a, const Packet &b)


hebiros
Author(s): Xavier Artache , Matthew Tesch
autogenerated on Thu Sep 3 2020 04:08:25