sergiotarxz/nuttx
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity
d20bd62e08
nuttx/libs/libc/math/lib_erf.c

70 lines
2.3 KiB
C
Raw Normal View History

Standardize the width of all comment boxes in C files
2015-10-03 01:43:18 +02:00
/****************************************************************************
This commit moves all of the libraries under a common directory called libs/. This most certainly break libcxx and uClibc++ for now. Squashed commit of the following: libs/libxx: Fix some confusing in naming. If the directory is called libxx, then the library must be libxx.a (unless perhaps LIBCXX is selected). libs/: Fix paths in moved library directories. libs: Brute force move of libc, libnx, and libxx to libs. Cannot yet build it in that configuration.
2018-05-29 21:21:26 +02:00
* libs/libc/math/lib_erf.c
Add error function to math library. From Brennan Ashton.
2015-01-26 20:58:47 +01:00
*
libs: libc: update licenses to Apache Gregory Nutt is the copyright holder for those files and he has submitted the SGA as a result we can migrate the licenses to Apache. Signed-off-by: Alin Jerpelea <alin.jerpelea@sony.com>
2021-05-26 10:05:59 +02:00
* Licensed to the Apache Software Foundation (ASF) under one or more
* contributor license agreements. See the NOTICE file distributed with
* this work for additional information regarding copyright ownership. The
* ASF licenses this file to you under the Apache License, Version 2.0 (the
* "License"); you may not use this file except in compliance with the
* License. You may obtain a copy of the License at
Add error function to math library. From Brennan Ashton.
2015-01-26 20:58:47 +01:00
*
libs: libc: update licenses to Apache Gregory Nutt is the copyright holder for those files and he has submitted the SGA as a result we can migrate the licenses to Apache. Signed-off-by: Alin Jerpelea <alin.jerpelea@sony.com>
2021-05-26 10:05:59 +02:00
* http://www.apache.org/licenses/LICENSE-2.0
Add error function to math library. From Brennan Ashton.
2015-01-26 20:58:47 +01:00
*
libs: libc: update licenses to Apache Gregory Nutt is the copyright holder for those files and he has submitted the SGA as a result we can migrate the licenses to Apache. Signed-off-by: Alin Jerpelea <alin.jerpelea@sony.com>
2021-05-26 10:05:59 +02:00
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
* WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
* License for the specific language governing permissions and limitations
* under the License.
Add error function to math library. From Brennan Ashton.
2015-01-26 20:58:47 +01:00
*
****************************************************************************/
Standardize the width of all comment boxes in C files
2015-10-03 01:43:18 +02:00
/****************************************************************************
Add error function to math library. From Brennan Ashton.
2015-01-26 20:58:47 +01:00
* Included Files
Standardize the width of all comment boxes in C files
2015-10-03 01:43:18 +02:00
****************************************************************************/
Add error function to math library. From Brennan Ashton.
2015-01-26 20:58:47 +01:00
#include <nuttx/config.h>
#include <nuttx/compiler.h>
#include <math.h>
This commit fixes the following libc/math issues: 1) asin[f l]() use Newton’s method to converge on a solution. But Newton’s method converges very slowly (> 500,000 iterations) for values of x close to 1.0; and, in the case of asinl(), sometimes fails to converge (loops forever). The attached patch uses an trig identity for values of x > sqrt(2). The resultant functions converge in no more than 5 iterations, 6 for asinl(). 2) The NuttX erf[f l]() functions are based on Chebyshev fitting to a good guess. The problem there’s a bug in the implementation that causes the functions to blow up with x near -3.0. This patch fixes that problem. It should be noted that this method returns the error function erf(x) with fractional error less than 1.2E-07 and that’s fine for the float version erff(), but the same method is used for double and long double version which will yield only slightly better precision. This patch doesn't address the issue of lower precision for erf() and erfl(). 3) a faster version of copysignf() for floats is included.
2016-07-30 23:43:56 +02:00
#ifdef CONFIG_HAVE_DOUBLE
/****************************************************************************
* Pre-processor Definitions
****************************************************************************/
#define A1 0.254829592
#define A2 (-0.284496736)
#define A3 1.421413741
#define A4 (-1.453152027)
#define A5 1.061405429
#define P 0.3275911
Standardize the width of all comment boxes in C files
2015-10-03 01:43:18 +02:00
/****************************************************************************
Add error function to math library. From Brennan Ashton.
2015-01-26 20:58:47 +01:00
* Public Functions
Standardize the width of all comment boxes in C files
2015-10-03 01:43:18 +02:00
****************************************************************************/
Add error function to math library. From Brennan Ashton.
2015-01-26 20:58:47 +01:00
This commit fixes the following libc/math issues: 1) asin[f l]() use Newton’s method to converge on a solution. But Newton’s method converges very slowly (> 500,000 iterations) for values of x close to 1.0; and, in the case of asinl(), sometimes fails to converge (loops forever). The attached patch uses an trig identity for values of x > sqrt(2). The resultant functions converge in no more than 5 iterations, 6 for asinl(). 2) The NuttX erf[f l]() functions are based on Chebyshev fitting to a good guess. The problem there’s a bug in the implementation that causes the functions to blow up with x near -3.0. This patch fixes that problem. It should be noted that this method returns the error function erf(x) with fractional error less than 1.2E-07 and that’s fine for the float version erff(), but the same method is used for double and long double version which will yield only slightly better precision. This patch doesn't address the issue of lower precision for erf() and erfl(). 3) a faster version of copysignf() for floats is included.
2016-07-30 23:43:56 +02:00
/****************************************************************************
* Name: erf
*
* Description:
* This implementation comes from the Handbook of Mathmatical Functions
* The implementations in this book are not protected by copyright.
* erf comes from formula 7.1.26
*
****************************************************************************/
Add error function to math library. From Brennan Ashton.
2015-01-26 20:58:47 +01:00
double erf(double x)
{
double t;
This commit fixes the following libc/math issues: 1) asin[f l]() use Newton’s method to converge on a solution. But Newton’s method converges very slowly (> 500,000 iterations) for values of x close to 1.0; and, in the case of asinl(), sometimes fails to converge (loops forever). The attached patch uses an trig identity for values of x > sqrt(2). The resultant functions converge in no more than 5 iterations, 6 for asinl(). 2) The NuttX erf[f l]() functions are based on Chebyshev fitting to a good guess. The problem there’s a bug in the implementation that causes the functions to blow up with x near -3.0. This patch fixes that problem. It should be noted that this method returns the error function erf(x) with fractional error less than 1.2E-07 and that’s fine for the float version erff(), but the same method is used for double and long double version which will yield only slightly better precision. This patch doesn't address the issue of lower precision for erf() and erfl(). 3) a faster version of copysignf() for floats is included.
2016-07-30 23:43:56 +02:00
double z;
Remove dangling whitespace
2015-10-04 23:28:54 +02:00
This commit fixes the following libc/math issues: 1) asin[f l]() use Newton’s method to converge on a solution. But Newton’s method converges very slowly (> 500,000 iterations) for values of x close to 1.0; and, in the case of asinl(), sometimes fails to converge (loops forever). The attached patch uses an trig identity for values of x > sqrt(2). The resultant functions converge in no more than 5 iterations, 6 for asinl(). 2) The NuttX erf[f l]() functions are based on Chebyshev fitting to a good guess. The problem there’s a bug in the implementation that causes the functions to blow up with x near -3.0. This patch fixes that problem. It should be noted that this method returns the error function erf(x) with fractional error less than 1.2E-07 and that’s fine for the float version erff(), but the same method is used for double and long double version which will yield only slightly better precision. This patch doesn't address the issue of lower precision for erf() and erfl(). 3) a faster version of copysignf() for floats is included.
2016-07-30 23:43:56 +02:00
z = fabs(x);
t = 1.0 / (1.0 + P * z);
libs: fix nxstyle errors fix erros reported by nxstyle Signed-off-by: Alin Jerpelea <alin.jerpelea@sony.com>
2021-04-02 07:44:54 +02:00
t = 1.0 -
(((((A5 * t + A4) * t) + A3) * t + A2) * t + A1) * t * exp(-z * z);
This commit fixes the following libc/math issues: 1) asin[f l]() use Newton’s method to converge on a solution. But Newton’s method converges very slowly (> 500,000 iterations) for values of x close to 1.0; and, in the case of asinl(), sometimes fails to converge (loops forever). The attached patch uses an trig identity for values of x > sqrt(2). The resultant functions converge in no more than 5 iterations, 6 for asinl(). 2) The NuttX erf[f l]() functions are based on Chebyshev fitting to a good guess. The problem there’s a bug in the implementation that causes the functions to blow up with x near -3.0. This patch fixes that problem. It should be noted that this method returns the error function erf(x) with fractional error less than 1.2E-07 and that’s fine for the float version erff(), but the same method is used for double and long double version which will yield only slightly better precision. This patch doesn't address the issue of lower precision for erf() and erfl(). 3) a faster version of copysignf() for floats is included.
2016-07-30 23:43:56 +02:00
return copysign(t, x);
Add error function to math library. From Brennan Ashton.
2015-01-26 20:58:47 +01:00
}
This commit fixes the following libc/math issues: 1) asin[f l]() use Newton’s method to converge on a solution. But Newton’s method converges very slowly (> 500,000 iterations) for values of x close to 1.0; and, in the case of asinl(), sometimes fails to converge (loops forever). The attached patch uses an trig identity for values of x > sqrt(2). The resultant functions converge in no more than 5 iterations, 6 for asinl(). 2) The NuttX erf[f l]() functions are based on Chebyshev fitting to a good guess. The problem there’s a bug in the implementation that causes the functions to blow up with x near -3.0. This patch fixes that problem. It should be noted that this method returns the error function erf(x) with fractional error less than 1.2E-07 and that’s fine for the float version erff(), but the same method is used for double and long double version which will yield only slightly better precision. This patch doesn't address the issue of lower precision for erf() and erfl(). 3) a faster version of copysignf() for floats is included.
2016-07-30 23:43:56 +02:00
#endif /* CONFIG_HAVE_DOUBLE */
Reference in New Issue Copy Permalink