luke
/
ngspice
mirror of https://git.code.sf.net/p/ngspice/ngspice
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

math functions in one central place maths/misc

This commit is contained in:
dwarning 2008-05-11 20:27:28 +00:00
parent 77925f7f48
commit a28ad1a197
14 changed files with 203 additions and 245 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
ChangeLog
View File

@ -1,3 +1,9 @@
2008-05-11 Dietmar Warning
* src/misc/missing_math.c,h, src/math/misc/*.*: move math function to one place
* src/include/ngspice.h, src/main.c, src/Makefile.am, src/maths/Makefile.am,
src/maths/misc/Makefile.am, src/misc/Makefile.am, configure.in: organization of libmathmisc.a
* src/maths/misc/erfc.c: better erfc for lossy transmission line
2008-05-10 Holger Vogt
* src/frontend/resource.c: Memory information is now stemming from the
/proc file system (LINUX) or using GlobalMemoryStatusEx and

8
configure.in
View File

@ -630,10 +630,6 @@ if test "$enable_cider" = "yes"; then
$CIDERDIR/input/libciderinput.a \
$CIDERDIR/support/libcidersuprt.a"
CIDERMATH="misc"
CIDERMATHDIR="maths/misc/libmathmisc.a"
NUMDEV=" spicelib/devices/nbjt/libnbjt.a \
spicelib/devices/nbjt2/libnbjt2.a \
spicelib/devices/numd/libnumd.a \
@ -650,16 +646,12 @@ CIDERSCRIPTS="devload devaxis ciderinit"
else
CIDERLIB=""
CIDERSIM=""
CIDERMATH=""
CIDERMATHDIR=""
NUMDEV=""
NUMDEVDIR=""
CIDERSCRIPTS=""
fi
AC_SUBST(CIDERDIR)
AC_SUBST(CIDERSIM)
AC_SUBST(CIDERMATH)
AC_SUBST(CIDERMATHDIR)
AC_SUBST(NUMDEV)
AC_SUBST(NUMDEVDIR)
AC_SUBST(CIDERSCRIPTS)

4
src/Makefile.am
View File

@ -109,7 +109,7 @@ ngspice_LDADD = \
@CIDERSIM@ \
maths/deriv/libderiv.a \
maths/cmaths/libcmaths.a \
@CIDERMATHDIR@ \
maths/misc/libmathmisc.a \
maths/poly/libpoly.a \
maths/ni/libni.a \
maths/sparse/libsparse.a \
@ -142,6 +142,7 @@ ngnutmeg_LDADD = \
@NUMPARAMLIB@ \
frontend/help/libhlp.a \
maths/cmaths/libcmaths.a \
maths/misc/libmathmisc.a \
maths/poly/libpoly.a \
misc/libmisc.a \
spicelib/parser/libinp.a
@ -173,6 +174,7 @@ ngsconvert_SOURCES = ngsconvert.c
ngsconvert_LDADD = \
frontend/libfte.a \
frontend/parser/libparser.a \
maths/misc/libmathmisc.a \
misc/libmisc.a
## proc2mod:

1
src/include/ngspice.h
View File

@ -31,6 +31,7 @@
#include <math.h>
#include <stdio.h>
#include "missing_math.h"
#ifdef STDC_HEADERS
# include <stdlib.h>

2
src/maths/Makefile.am
View File

@ -1,6 +1,6 @@
## Process this file with automake
SUBDIRS = cmaths ni sparse poly deriv @CIDERMATH@
SUBDIRS = cmaths ni sparse poly deriv misc
DIST_SUBDIRS = cmaths ni sparse poly deriv misc
MAINTAINERCLEANFILES = Makefile.in

4
src/maths/misc/Makefile.am
View File

@ -8,6 +8,10 @@ libmathmisc_a_SOURCES = \
bernoull.h \
bernoull.c \
erfc.c \
equality.c \
isnan.c \
logb.c \
scalb.c \
norm.h \
norm.c

4
src/maths/misc/accuracy.c
View File

@ -69,6 +69,9 @@ evalAccLimits(void)
double xhold, dif; /* Introduced to avoid numerical trap if
using non IEEE754 FPU */
#ifndef CIDER
double Acc, BMin, BMax, ExpLim, MuLim, MutLim;
#endif
/* First we compute accuracy */
@ -149,7 +152,6 @@ evalAccLimits(void)
muLim *= 2.0;
MutLim = muLim;
}
/*

62
src/maths/misc/equality.c Normal file
View File

@ -0,0 +1,62 @@
/**********
Copyright 1991 Regents of the University of California. All rights reserved.
**********/
#include <assert.h>
#include "ngspice.h"
#ifdef _MSC_VER
typedef __int64 long64;
#else
typedef long long long64;
#endif
#define Abs(x) ((x) < 0 ? -(x) : (x))
/* From Bruce Dawson, Comparing floating point numbers,
http://www.cygnus-software.com/papers/comparingfloats/Comparing%20floating%20point%20numbers.htm
Original this function is named AlmostEqual2sComplement but we leave it to AlmostEqualUlps
and can leave the code (measure.c, dctran.c) unchanged. The transformation to the 2's complement
prevent problems around 0.0.
One Ulp is equivalent to a maxRelativeError of between 1/4,000,000,000,000,000 and 1/8,000,000,000,000,000.
Practical: 3 < maxUlps < some hundred's (or thousand's) - depending on numerical requirements.
*/
bool AlmostEqualUlps(double A, double B, int maxUlps)
{
long64 aInt, bInt, intDiff;
if (A == B)
return TRUE;
/* If not - the entire method can not work */
assert(sizeof(double) == sizeof(long64));
/* Make sure maxUlps is non-negative and small enough that the */
/* default NAN won't compare as equal to anything. */
assert(maxUlps > 0 && maxUlps < 4 * 1024 * 1024);
aInt = *(long64*)&A;
/* Make aInt lexicographically ordered as a twos-complement int */
if (aInt < 0)
#ifdef _MSC_VER
aInt = 0x8000000000000000 - aInt;
#else
aInt = 0x8000000000000000LL - aInt;
#endif
bInt = *(long64*)&B;
/* Make bInt lexicographically ordered as a twos-complement int */
if (bInt < 0)
#ifdef _MSC_VER
bInt = 0x8000000000000000 - bInt;
#else
bInt = 0x8000000000000000LL - bInt;
#endif
#ifdef _MSC_VER
intDiff = Abs(aInt - bInt);
#else
intDiff = llabs(aInt - bInt);
#endif
/* printf("A:%e B:%e aInt:%d bInt:%d diff:%d\n", A, B, aInt, bInt, intDiff); */
if (intDiff <= maxUlps)
return TRUE;
return FALSE;
}

53
src/maths/misc/isnan.c Normal file
View File

@ -0,0 +1,53 @@
/**********
Copyright 1991 Regents of the University of California. All rights reserved.
**********/
#include "ngspice.h"
#ifndef HAVE_ISNAN
/* isnan (originally) for SOI devices in MINGW32 hvogt (dev.c) */
union ieee754_double
{
double d;
/* This is the IEEE 754 double-precision format. */
struct
{
/* Together these comprise the mantissa. */
unsigned int mantissa1:32;
unsigned int mantissa0:20;
unsigned int exponent:11;
unsigned int negative:1;
} ieee;
struct
{
/* Together these comprise the mantissa. */
unsigned int mantissa1:32;
unsigned int mantissa0:19;
unsigned int quiet_nan:1;
unsigned int exponent:11;
unsigned int negative:1;
} ieee_nan;
};
int
isnan(double value)
{
union ieee754_double u;
u.d = value;
/* IEEE 754 NaN's have the maximum possible
exponent and a nonzero mantissa. */
return ((u.ieee.exponent & 0x7ff) == 0x7ff &&
(u.ieee.mantissa0 != 0 || u.ieee.mantissa1 != 0));
}
/*
* end isnan.c
*/
#else /* HAVE_ISNAN */
int Dummy_Symbol_4;
#endif /* HAVE_ISNAN */

36
src/maths/misc/logb.c Normal file
View File

@ -0,0 +1,36 @@
/**********
Copyright 1991 Regents of the University of California. All rights reserved.
**********/
#include "ngspice.h"
#ifndef HAVE_LOGB
double
logb(double x)
{
double y = 0.0;
if (x != 0.0)
{
if (x < 0.0)
x = - x;
while (x > 2.0)
{
y += 1.0;
x /= 2.0;
}
while (x < 1.0)
{
y -= 1.0;
x *= 2.0;
}
}
else
y = 0.0;
return y;
}
#else
int Dummy_Symbol_3;
#endif

34
src/maths/misc/scalb.c Normal file
View File

@ -0,0 +1,34 @@
/**********
Copyright 1991 Regents of the University of California. All rights reserved.
**********/
#include "ngspice.h"
#ifndef HAVE_SCALB
# ifdef HAVE_SCALBN
# define scalb scalbn
#else /* Chris Inbody */
double
scalb(double x, int n)
{
double y, z = 1.0, k = 2.0;
if (n < 0) {
n = -n;
k = 0.5;
}
if (x != 0.0)
for (y = 1.0; n; n >>= 1) {
y *= k;
if (n & 1)
z *= y;
}
return x * z;
}
# endif /* HAVE_SCALBN */
#else /* HAVE_SCALB */
int Dummy_Symbol_1;
#endif /* HAVE_SCALB */

2
src/misc/Makefile.am
View File

@ -14,8 +14,6 @@ libmisc_a_SOURCES = \
dup2.h \
ivars.c \
ivars.h \
missing_math.c \
missing_math.h \
mktemp.c \
mktemp.h \
printnum.c \

203
src/misc/missing_math.c
View File

@ -1,203 +0,0 @@
/**********
Copyright 1991 Regents of the University of California. All rights reserved.
$Id$
**********/
/*
* Missing math functions
*/
#include <assert.h>
#include "config.h"
#include "ngspice.h"
#include "missing_math.h"
#ifdef _MSC_VER
typedef __int64 long64;
#else
typedef long long long64;
#endif
#define Abs(x) ((x) < 0 ? -(x) : (x))
/* From Bruce Dawson, Comparing floating point numbers,
http://www.cygnus-software.com/papers/comparingfloats/Comparing%20floating%20point%20numbers.htm
Original this function is named AlmostEqual2sComplement but we leave it to AlmostEqualUlps
and can leave the code (measure.c, dctran.c) unchanged. The transformation to the 2's complement
prevent problems around 0.0.
One Ulp is equivalent to a maxRelativeError of between 1/4,000,000,000,000,000 and 1/8,000,000,000,000,000.
Practical: 3 < maxUlps < some hundred's (or thousand's) - depending on numerical requirements.
*/
bool AlmostEqualUlps(double A, double B, int maxUlps)
{
long64 aInt, bInt, intDiff;
if (A == B)
return TRUE;
/* If not - the entire method can not work */
assert(sizeof(double) == sizeof(long64));
/* Make sure maxUlps is non-negative and small enough that the */
/* default NAN won't compare as equal to anything. */
assert(maxUlps > 0 && maxUlps < 4 * 1024 * 1024);
aInt = *(long64*)&A;
/* Make aInt lexicographically ordered as a twos-complement int */
if (aInt < 0)
#ifdef _MSC_VER
aInt = 0x8000000000000000 - aInt;
#else
aInt = 0x8000000000000000LL - aInt;
#endif
bInt = *(long64*)&B;
/* Make bInt lexicographically ordered as a twos-complement int */
if (bInt < 0)
#ifdef _MSC_VER
bInt = 0x8000000000000000 - bInt;
#else
bInt = 0x8000000000000000LL - bInt;
#endif
#ifdef _MSC_VER
intDiff = Abs(aInt - bInt);
#else
intDiff = llabs(aInt - bInt);
#endif
/* printf("A:%e B:%e aInt:%d bInt:%d diff:%d\n", A, B, aInt, bInt, intDiff); */
if (intDiff <= maxUlps)
return TRUE;
return FALSE;
}
#ifndef HAVE_LOGB
double
logb(double x)
{
double y = 0.0;
if (x != 0.0)
{
if (x < 0.0)
x = - x;
while (x > 2.0)
{
y += 1.0;
x /= 2.0;
}
while (x < 1.0)
{
y -= 1.0;
x *= 2.0;
}
}
else
y = 0.0;
return y;
}
#endif
#ifndef HAVE_SCALB
# ifdef HAVE_SCALBN
# define scalb scalbn
#else /* Chris Inbody */
double
scalb(double x, int n)
{
double y, z = 1.0, k = 2.0;
if (n < 0) {
n = -n;
k = 0.5;
}
if (x != 0.0)
for (y = 1.0; n; n >>= 1) {
y *= k;
if (n & 1)
z *= y;
}
return x * z;
}
# endif /* HAVE_SCALBN */
#endif /* HAVE_SCALB */
#ifndef HAVE_ERFC
/* From C. Hastings, Jr., Approximations for digital computers,
Princeton Univ. Press, 1955.
Approximation accurate to within 1.5E-7
(making some assumptions about your machine's floating point mechanism)
*/
double
erfc(double x)
{
double t, z;
t = 1/(1 + 0.3275911*x);
z = 1.061405429;
z = -1.453152027 + t * z;
z = 1.421413741 + t * z;
z = -0.284496736 + t * z;
z = 0.254829592 + t * z;
z = exp(-x*x) * t * z;
return(z);
}
#endif
#ifndef HAVE_ISNAN
/* isnan (originally) for SOI devices in MINGW32 hvogt (dev.c) */
union ieee754_double
{
double d;
/* This is the IEEE 754 double-precision format. */
struct
{
/* Together these comprise the mantissa. */
unsigned int mantissa1:32;
unsigned int mantissa0:20;
unsigned int exponent:11;
unsigned int negative:1;
} ieee;
struct
{
/* Together these comprise the mantissa. */
unsigned int mantissa1:32;
unsigned int mantissa0:19;
unsigned int quiet_nan:1;
unsigned int exponent:11;
unsigned int negative:1;
} ieee_nan;
};
int
isnan(double value)
{
union ieee754_double u;
u.d = value;
/* IEEE 754 NaN's have the maximum possible
exponent and a nonzero mantissa. */
return ((u.ieee.exponent & 0x7ff) == 0x7ff &&
(u.ieee.mantissa0 != 0 || u.ieee.mantissa1 != 0));
}
/*
* end isnan.c
*/
#endif /* HAVE_ISNAN */

29
src/misc/missing_math.h
View File

@ -1,29 +0,0 @@
/*************
* Header file for missing_math.c
* 1999 E. Rouat
************/
#ifndef MISSING_MATH_H_INCLUDED
#define MISSING_MATH_H_INCLUDED
bool AlmostEqualUlps(double, double, int);
#ifndef HAVE_ERFC
double erfc(double);
#endif
#ifndef HAVE_LOGB
double logb(double);
#endif
#ifndef HAVE_SCALB
# ifndef HAVE_SCALBN
double scalb(double, int);
# endif
#endif
#ifndef HAVE_ISNAN
int isnan(double);
#endif
#endif /* MISSING_MATH_H_INCLUDED */