/*** *math.h - definitions and declarations for math library * * Copyright (c) 1985-2001, Microsoft Corporation. All rights reserved. * *Purpose: * This file contains constant definitions and external subroutine * declarations for the math subroutine library. * [ANSI/System V] * * [Public] * ****/ #if _MSC_VER > 1000 #pragma once #endif #ifndef _INC_MATH #define _INC_MATH #if !defined(_WIN32) #error ERROR: Only Win32 target supported! #endif #ifdef _MSC_VER /* * Currently, all MS C compilers for Win32 platforms default to 8 byte * alignment. */ #pragma pack(push,8) #endif /* _MSC_VER */ #ifdef __cplusplus extern "C" { #endif #ifndef __assembler /* Protect from assembler */ /* Define _CRTIMP */ #ifndef _CRTIMP #ifdef _DLL #define _CRTIMP __declspec(dllimport) #else /* ndef _DLL */ #define _CRTIMP #endif /* _DLL */ #endif /* _CRTIMP */ /* Define __cdecl for non-Microsoft compilers */ #if ( !defined(_MSC_VER) && !defined(__cdecl) ) #define __cdecl #endif /* Definition of _exception struct - this struct is passed to the matherr * routine when a floating point exception is detected */ #ifndef _EXCEPTION_DEFINED struct _exception { int type; /* exception type - see below */ char *name; /* name of function where error occured */ double arg1; /* first argument to function */ double arg2; /* second argument (if any) to function */ double retval; /* value to be returned by function */ } ; #define _EXCEPTION_DEFINED #endif /* Definition of a _complex struct to be used by those who use cabs and * want type checking on their argument */ #ifndef _COMPLEX_DEFINED struct _complex { double x,y; /* real and imaginary parts */ } ; #if !__STDC__ && !defined (__cplusplus) /* Non-ANSI name for compatibility */ #define complex _complex #endif #define _COMPLEX_DEFINED #endif #endif /* __assembler */ /* Constant definitions for the exception type passed in the _exception struct */ #define _DOMAIN 1 /* argument domain error */ #define _SING 2 /* argument singularity */ #define _OVERFLOW 3 /* overflow range error */ #define _UNDERFLOW 4 /* underflow range error */ #define _TLOSS 5 /* total loss of precision */ #define _PLOSS 6 /* partial loss of precision */ #define EDOM 33 #define ERANGE 34 /* Definitions of _HUGE and HUGE_VAL - respectively the XENIX and ANSI names * for a value returned in case of error by a number of the floating point * math routines */ #ifndef __assembler /* Protect from assembler */ _CRTIMP extern double _HUGE; #endif /* __assembler */ #define HUGE_VAL _HUGE #ifdef _USE_MATH_DEFINES /* Define _USE_MATH_DEFINES before including math.h to expose these macro * definitions for common math constants. These are placed under an #ifdef * since these commonly-defined names are not part of the C/C++ standards. */ /* Definitions of useful mathematical constants * M_E - e * M_LOG2E - log2(e) * M_LOG10E - log10(e) * M_LN2 - ln(2) * M_LN10 - ln(10) * M_PI - pi * M_PI_2 - pi/2 * M_PI_4 - pi/4 * M_1_PI - 1/pi * M_2_PI - 2/pi * M_2_SQRTPI - 2/sqrt(pi) * M_SQRT2 - sqrt(2) * M_SQRT1_2 - 1/sqrt(2) */ #define M_E 2.71828182845904523536 #define M_LOG2E 1.44269504088896340736 #define M_LOG10E 0.434294481903251827651 #define M_LN2 0.693147180559945309417 #define M_LN10 2.30258509299404568402 #define M_PI 3.14159265358979323846 #define M_PI_2 1.57079632679489661923 #define M_PI_4 0.785398163397448309616 #define M_1_PI 0.318309886183790671538 #define M_2_PI 0.636619772367581343076 #define M_2_SQRTPI 1.12837916709551257390 #define M_SQRT2 1.41421356237309504880 #define M_SQRT1_2 0.707106781186547524401 #endif /* _USE_MATH_DEFINES */ /* Function prototypes */ #if !defined(__assembler) /* Protect from assembler */ #if defined(_M_MRX000) _CRTIMP int __cdecl abs(int); _CRTIMP double __cdecl acos(double); _CRTIMP double __cdecl asin(double); _CRTIMP double __cdecl atan(double); _CRTIMP double __cdecl atan2(double, double); _CRTIMP double __cdecl cos(double); _CRTIMP double __cdecl cosh(double); _CRTIMP double __cdecl exp(double); _CRTIMP double __cdecl fabs(double); _CRTIMP double __cdecl fmod(double, double); _CRTIMP long __cdecl labs(long); _CRTIMP double __cdecl log(double); _CRTIMP double __cdecl log10(double); _CRTIMP double __cdecl pow(double, double); _CRTIMP double __cdecl sin(double); _CRTIMP double __cdecl sinh(double); _CRTIMP double __cdecl tan(double); _CRTIMP double __cdecl tanh(double); _CRTIMP double __cdecl sqrt(double); #else int __cdecl abs(int); double __cdecl acos(double); double __cdecl asin(double); double __cdecl atan(double); double __cdecl atan2(double, double); double __cdecl cos(double); double __cdecl cosh(double); double __cdecl exp(double); double __cdecl fabs(double); double __cdecl fmod(double, double); long __cdecl labs(long); double __cdecl log(double); double __cdecl log10(double); double __cdecl pow(double, double); double __cdecl sin(double); double __cdecl sinh(double); double __cdecl tan(double); double __cdecl tanh(double); double __cdecl sqrt(double); #endif _CRTIMP double __cdecl atof(const char *); _CRTIMP double __cdecl _cabs(struct _complex); #if defined(_M_ALPHA) double __cdecl ceil(double); double __cdecl floor(double); #else _CRTIMP double __cdecl ceil(double); _CRTIMP double __cdecl floor(double); #endif _CRTIMP double __cdecl frexp(double, int *); _CRTIMP double __cdecl _hypot(double, double); _CRTIMP double __cdecl _j0(double); _CRTIMP double __cdecl _j1(double); _CRTIMP double __cdecl _jn(int, double); _CRTIMP double __cdecl ldexp(double, int); int __cdecl _matherr(struct _exception *); _CRTIMP double __cdecl modf(double, double *); _CRTIMP double __cdecl _y0(double); _CRTIMP double __cdecl _y1(double); _CRTIMP double __cdecl _yn(int, double); #if defined(_M_IX86) _CRTIMP int __cdecl _set_SSE2_enable(int); #endif #if defined(_M_MRX000) /* MIPS fast prototypes for float */ /* ANSI C, 4.5 Mathematics */ /* 4.5.2 Trigonometric functions */ _CRTIMP float __cdecl acosf( float ); _CRTIMP float __cdecl asinf( float ); _CRTIMP float __cdecl atanf( float ); _CRTIMP float __cdecl atan2f( float , float ); _CRTIMP float __cdecl cosf( float ); _CRTIMP float __cdecl sinf( float ); _CRTIMP float __cdecl tanf( float ); /* 4.5.3 Hyperbolic functions */ _CRTIMP float __cdecl coshf( float ); _CRTIMP float __cdecl sinhf( float ); _CRTIMP float __cdecl tanhf( float ); /* 4.5.4 Exponential and logarithmic functions */ _CRTIMP float __cdecl expf( float ); _CRTIMP float __cdecl logf( float ); _CRTIMP float __cdecl log10f( float ); _CRTIMP float __cdecl modff( float , float* ); /* 4.5.5 Power functions */ _CRTIMP float __cdecl powf( float , float ); float __cdecl sqrtf( float ); /* 4.5.6 Nearest integer, absolute value, and remainder functions */ float __cdecl ceilf( float ); float __cdecl fabsf( float ); float __cdecl floorf( float ); _CRTIMP float __cdecl fmodf( float , float ); _CRTIMP float __cdecl hypotf(float, float); #endif /* _M_MRX000 */ #if defined(_M_ALPHA) /* ALPHA fast prototypes for float */ /* ANSI C, 4.5 Mathematics */ /* 4.5.2 Trigonometric functions */ float __cdecl acosf( float ); float __cdecl asinf( float ); float __cdecl atanf( float ); float __cdecl atan2f( float , float ); float __cdecl cosf( float ); float __cdecl sinf( float ); float __cdecl tanf( float ); /* 4.5.3 Hyperbolic functions */ float __cdecl coshf( float ); float __cdecl sinhf( float ); float __cdecl tanhf( float ); /* 4.5.4 Exponential and logarithmic functions */ float __cdecl expf( float ); float __cdecl logf( float ); float __cdecl log10f( float ); _CRTIMP float __cdecl modff( float , float* ); /* 4.5.5 Power functions */ float __cdecl powf( float , float ); float __cdecl sqrtf( float ); /* 4.5.6 Nearest integer, absolute value, and remainder functions */ float __cdecl ceilf( float ); float __cdecl fabsf( float ); float __cdecl floorf( float ); float __cdecl fmodf( float , float ); _CRTIMP float __cdecl _hypotf(float, float); #endif /* _M_ALPHA */ #if defined(_M_IA64) /* ANSI C, 4.5 Mathematics */ /* 4.5.2 Trigonometric functions */ float __cdecl acosf( float ); float __cdecl asinf( float ); float __cdecl atanf( float ); float __cdecl atan2f( float , float ); float __cdecl cosf( float ); float __cdecl sinf( float ); float __cdecl tanf( float ); /* 4.5.3 Hyperbolic functions */ float __cdecl coshf( float ); float __cdecl sinhf( float ); float __cdecl tanhf( float ); /* 4.5.4 Exponential and logarithmic functions */ float __cdecl expf( float ); float __cdecl logf( float ); float __cdecl log10f( float ); float __cdecl modff( float , float* ); /* 4.5.5 Power functions */ float __cdecl powf( float , float ); float __cdecl sqrtf( float ); /* 4.5.6 Nearest integer, absolute value, and remainder functions */ float __cdecl ceilf( float ); float __cdecl fabsf( float ); float __cdecl floorf( float ); float __cdecl fmodf( float , float ); float __cdecl hypotf(float, float); #endif /* _M_IA64 */ /* Macros defining long double functions to be their double counterparts * (long double is synonymous with double in this implementation). */ #ifndef __cplusplus #define acosl(x) ((long double)acos((double)(x))) #define asinl(x) ((long double)asin((double)(x))) #define atanl(x) ((long double)atan((double)(x))) #define atan2l(x,y) ((long double)atan2((double)(x), (double)(y))) #define _cabsl _cabs #define ceill(x) ((long double)ceil((double)(x))) #define cosl(x) ((long double)cos((double)(x))) #define coshl(x) ((long double)cosh((double)(x))) #define expl(x) ((long double)exp((double)(x))) #define fabsl(x) ((long double)fabs((double)(x))) #define floorl(x) ((long double)floor((double)(x))) #define fmodl(x,y) ((long double)fmod((double)(x), (double)(y))) #define frexpl(x,y) ((long double)frexp((double)(x), (y))) #define _hypotl(x,y) ((long double)_hypot((double)(x), (double)(y))) #define ldexpl(x,y) ((long double)ldexp((double)(x), (y))) #define logl(x) ((long double)log((double)(x))) #define log10l(x) ((long double)log10((double)(x))) #define _matherrl _matherr #define modfl(x,y) ((long double)modf((double)(x), (double *)(y))) #define powl(x,y) ((long double)pow((double)(x), (double)(y))) #define sinl(x) ((long double)sin((double)(x))) #define sinhl(x) ((long double)sinh((double)(x))) #define sqrtl(x) ((long double)sqrt((double)(x))) #define tanl(x) ((long double)tan((double)(x))) #define tanhl(x) ((long double)tanh((double)(x))) #else /* __cplusplus */ inline long double acosl(long double _X) {return (acos((double)_X)); } inline long double asinl(long double _X) {return (asin((double)_X)); } inline long double atanl(long double _X) {return (atan((double)_X)); } inline long double atan2l(long double _X, long double _Y) {return (atan2((double)_X, (double)_Y)); } inline long double ceill(long double _X) {return (ceil((double)_X)); } inline long double cosl(long double _X) {return (cos((double)_X)); } inline long double coshl(long double _X) {return (cosh((double)_X)); } inline long double expl(long double _X) {return (exp((double)_X)); } inline long double fabsl(long double _X) {return (fabs((double)_X)); } inline long double floorl(long double _X) {return (floor((double)_X)); } inline long double fmodl(long double _X, long double _Y) {return (fmod((double)_X, (double)_Y)); } inline long double frexpl(long double _X, int *_Y) {return (frexp((double)_X, _Y)); } inline long double ldexpl(long double _X, int _Y) {return (ldexp((double)_X, _Y)); } inline long double logl(long double _X) {return (log((double)_X)); } inline long double log10l(long double _X) {return (log10((double)_X)); } inline long double modfl(long double _X, long double *_Y) {double _Di, _Df = modf((double)_X, &_Di); *_Y = (long double)_Di; return (_Df); } inline long double powl(long double _X, long double _Y) {return (pow((double)_X, (double)_Y)); } inline long double sinl(long double _X) {return (sin((double)_X)); } inline long double sinhl(long double _X) {return (sinh((double)_X)); } inline long double sqrtl(long double _X) {return (sqrt((double)_X)); } inline long double tanl(long double _X) {return (tan((double)_X)); } inline long double tanhl(long double _X) {return (tanh((double)_X)); } inline float frexpf(float _X, int *_Y) {return ((float)frexp((double)_X, _Y)); } inline float ldexpf(float _X, int _Y) {return ((float)ldexp((double)_X, _Y)); } #if !defined(_M_MRX000) && !defined(_M_ALPHA) && !defined(_M_IA64) inline float acosf(float _X) {return ((float)acos((double)_X)); } inline float asinf(float _X) {return ((float)asin((double)_X)); } inline float atanf(float _X) {return ((float)atan((double)_X)); } inline float atan2f(float _X, float _Y) {return ((float)atan2((double)_X, (double)_Y)); } inline float ceilf(float _X) {return ((float)ceil((double)_X)); } inline float cosf(float _X) {return ((float)cos((double)_X)); } inline float coshf(float _X) {return ((float)cosh((double)_X)); } inline float expf(float _X) {return ((float)exp((double)_X)); } inline float fabsf(float _X) {return ((float)fabs((double)_X)); } inline float floorf(float _X) {return ((float)floor((double)_X)); } inline float fmodf(float _X, float _Y) {return ((float)fmod((double)_X, (double)_Y)); } inline float logf(float _X) {return ((float)log((double)_X)); } inline float log10f(float _X) {return ((float)log10((double)_X)); } inline float modff(float _X, float *_Y) { double _Di, _Df = modf((double)_X, &_Di); *_Y = (float)_Di; return ((float)_Df); } inline float powf(float _X, float _Y) {return ((float)pow((double)_X, (double)_Y)); } inline float sinf(float _X) {return ((float)sin((double)_X)); } inline float sinhf(float _X) {return ((float)sinh((double)_X)); } inline float sqrtf(float _X) {return ((float)sqrt((double)_X)); } inline float tanf(float _X) {return ((float)tan((double)_X)); } inline float tanhf(float _X) {return ((float)tanh((double)_X)); } #endif /* !defined(_M_MRX000) && !defined(_M_ALPHA) && !defined(_M_IA64) */ #endif /* __cplusplus */ #endif /* __assembler */ #if !__STDC__ /* Non-ANSI names for compatibility */ #define DOMAIN _DOMAIN #define SING _SING #define OVERFLOW _OVERFLOW #define UNDERFLOW _UNDERFLOW #define TLOSS _TLOSS #define PLOSS _PLOSS #define matherr _matherr #ifndef __assembler /* Protect from assembler */ _CRTIMP extern double HUGE; _CRTIMP double __cdecl cabs(struct _complex); _CRTIMP double __cdecl hypot(double, double); _CRTIMP double __cdecl j0(double); _CRTIMP double __cdecl j1(double); _CRTIMP double __cdecl jn(int, double); int __cdecl matherr(struct _exception *); _CRTIMP double __cdecl y0(double); _CRTIMP double __cdecl y1(double); _CRTIMP double __cdecl yn(int, double); #endif /* __assembler */ #endif /* __STDC__ */ #ifdef __cplusplus } extern "C++" { template inline _Ty _Pow_int(_Ty _X, int _Y) {unsigned int _N; if (_Y >= 0) _N = _Y; else _N = -_Y; for (_Ty _Z = _Ty(1); ; _X *= _X) {if ((_N & 1) != 0) _Z *= _X; if ((_N >>= 1) == 0) return (_Y < 0 ? _Ty(1) / _Z : _Z); }} #ifndef _MSC_EXTENSIONS inline long __cdecl abs(long _X) {return (labs(_X)); } inline double __cdecl abs(double _X) {return (fabs(_X)); } inline double __cdecl pow(double _X, int _Y) {return (_Pow_int(_X, _Y)); } inline double __cdecl pow(int _X, int _Y) {return (_Pow_int(_X, _Y)); } inline float __cdecl abs(float _X) {return (fabsf(_X)); } inline float __cdecl acos(float _X) {return (acosf(_X)); } inline float __cdecl asin(float _X) {return (asinf(_X)); } inline float __cdecl atan(float _X) {return (atanf(_X)); } inline float __cdecl atan2(float _Y, float _X) {return (atan2f(_Y, _X)); } inline float __cdecl ceil(float _X) {return (ceilf(_X)); } inline float __cdecl cos(float _X) {return (cosf(_X)); } inline float __cdecl cosh(float _X) {return (coshf(_X)); } inline float __cdecl exp(float _X) {return (expf(_X)); } inline float __cdecl fabs(float _X) {return (fabsf(_X)); } inline float __cdecl floor(float _X) {return (floorf(_X)); } inline float __cdecl fmod(float _X, float _Y) {return (fmodf(_X, _Y)); } inline float __cdecl frexp(float _X, int * _Y) {return (frexpf(_X, _Y)); } inline float __cdecl ldexp(float _X, int _Y) {return (ldexpf(_X, _Y)); } inline float __cdecl log(float _X) {return (logf(_X)); } inline float __cdecl log10(float _X) {return (log10f(_X)); } inline float __cdecl modf(float _X, float * _Y) {return (modff(_X, _Y)); } inline float __cdecl pow(float _X, float _Y) {return (powf(_X, _Y)); } inline float __cdecl pow(float _X, int _Y) {return (_Pow_int(_X, _Y)); } inline float __cdecl sin(float _X) {return (sinf(_X)); } inline float __cdecl sinh(float _X) {return (sinhf(_X)); } inline float __cdecl sqrt(float _X) {return (sqrtf(_X)); } inline float __cdecl tan(float _X) {return (tanf(_X)); } inline float __cdecl tanh(float _X) {return (tanhf(_X)); } inline long double __cdecl abs(long double _X) {return (fabsl(_X)); } inline long double __cdecl acos(long double _X) {return (acosl(_X)); } inline long double __cdecl asin(long double _X) {return (asinl(_X)); } inline long double __cdecl atan(long double _X) {return (atanl(_X)); } inline long double __cdecl atan2(long double _Y, long double _X) {return (atan2l(_Y, _X)); } inline long double __cdecl ceil(long double _X) {return (ceill(_X)); } inline long double __cdecl cos(long double _X) {return (cosl(_X)); } inline long double __cdecl cosh(long double _X) {return (coshl(_X)); } inline long double __cdecl exp(long double _X) {return (expl(_X)); } inline long double __cdecl fabs(long double _X) {return (fabsl(_X)); } inline long double __cdecl floor(long double _X) {return (floorl(_X)); } inline long double __cdecl fmod(long double _X, long double _Y) {return (fmodl(_X, _Y)); } inline long double __cdecl frexp(long double _X, int * _Y) {return (frexpl(_X, _Y)); } inline long double __cdecl ldexp(long double _X, int _Y) {return (ldexpl(_X, _Y)); } inline long double __cdecl log(long double _X) {return (logl(_X)); } inline long double __cdecl log10(long double _X) {return (log10l(_X)); } inline long double __cdecl modf(long double _X, long double * _Y) {return (modfl(_X, _Y)); } inline long double __cdecl pow(long double _X, long double _Y) {return (powl(_X, _Y)); } inline long double __cdecl pow(long double _X, int _Y) {return (_Pow_int(_X, _Y)); } inline long double __cdecl sin(long double _X) {return (sinl(_X)); } inline long double __cdecl sinh(long double _X) {return (sinhl(_X)); } inline long double __cdecl sqrt(long double _X) {return (sqrtl(_X)); } inline long double __cdecl tan(long double _X) {return (tanl(_X)); } inline long double __cdecl tanh(long double _X) {return (tanhl(_X)); } #endif /* _MSC_EXTENSIONS */ } #endif /* __cplusplus */ #ifdef _MSC_VER #pragma pack(pop) #endif /* _MSC_VER */ #endif /* _INC_MATH */