diff options
Diffstat (limited to 'libphobos/src/std/math.d')
-rw-r--r-- | libphobos/src/std/math.d | 8413 |
1 files changed, 8413 insertions, 0 deletions
diff --git a/libphobos/src/std/math.d b/libphobos/src/std/math.d new file mode 100644 index 0000000..84cf4a7 --- /dev/null +++ b/libphobos/src/std/math.d @@ -0,0 +1,8413 @@ +// Written in the D programming language. + +/** + * Contains the elementary mathematical functions (powers, roots, + * and trigonometric functions), and low-level floating-point operations. + * Mathematical special functions are available in $(D std.mathspecial). + * +$(SCRIPT inhibitQuickIndex = 1;) + +$(DIVC quickindex, +$(BOOKTABLE , +$(TR $(TH Category) $(TH Members) ) +$(TR $(TDNW Constants) $(TD + $(MYREF E) $(MYREF PI) $(MYREF PI_2) $(MYREF PI_4) $(MYREF M_1_PI) + $(MYREF M_2_PI) $(MYREF M_2_SQRTPI) $(MYREF LN10) $(MYREF LN2) + $(MYREF LOG2) $(MYREF LOG2E) $(MYREF LOG2T) $(MYREF LOG10E) + $(MYREF SQRT2) $(MYREF SQRT1_2) +)) +$(TR $(TDNW Classics) $(TD + $(MYREF abs) $(MYREF fabs) $(MYREF sqrt) $(MYREF cbrt) $(MYREF hypot) + $(MYREF poly) $(MYREF nextPow2) $(MYREF truncPow2) +)) +$(TR $(TDNW Trigonometry) $(TD + $(MYREF sin) $(MYREF cos) $(MYREF tan) $(MYREF asin) $(MYREF acos) + $(MYREF atan) $(MYREF atan2) $(MYREF sinh) $(MYREF cosh) $(MYREF tanh) + $(MYREF asinh) $(MYREF acosh) $(MYREF atanh) $(MYREF expi) +)) +$(TR $(TDNW Rounding) $(TD + $(MYREF ceil) $(MYREF floor) $(MYREF round) $(MYREF lround) + $(MYREF trunc) $(MYREF rint) $(MYREF lrint) $(MYREF nearbyint) + $(MYREF rndtol) $(MYREF quantize) +)) +$(TR $(TDNW Exponentiation & Logarithms) $(TD + $(MYREF pow) $(MYREF exp) $(MYREF exp2) $(MYREF expm1) $(MYREF ldexp) + $(MYREF frexp) $(MYREF log) $(MYREF log2) $(MYREF log10) $(MYREF logb) + $(MYREF ilogb) $(MYREF log1p) $(MYREF scalbn) +)) +$(TR $(TDNW Modulus) $(TD + $(MYREF fmod) $(MYREF modf) $(MYREF remainder) +)) +$(TR $(TDNW Floating-point operations) $(TD + $(MYREF approxEqual) $(MYREF feqrel) $(MYREF fdim) $(MYREF fmax) + $(MYREF fmin) $(MYREF fma) $(MYREF nextDown) $(MYREF nextUp) + $(MYREF nextafter) $(MYREF NaN) $(MYREF getNaNPayload) + $(MYREF cmp) +)) +$(TR $(TDNW Introspection) $(TD + $(MYREF isFinite) $(MYREF isIdentical) $(MYREF isInfinity) $(MYREF isNaN) + $(MYREF isNormal) $(MYREF isSubnormal) $(MYREF signbit) $(MYREF sgn) + $(MYREF copysign) $(MYREF isPowerOf2) +)) +$(TR $(TDNW Complex Numbers) $(TD + $(MYREF abs) $(MYREF conj) $(MYREF sin) $(MYREF cos) $(MYREF expi) +)) +$(TR $(TDNW Hardware Control) $(TD + $(MYREF IeeeFlags) $(MYREF FloatingPointControl) +)) +) +) + + * The functionality closely follows the IEEE754-2008 standard for + * floating-point arithmetic, including the use of camelCase names rather + * than C99-style lower case names. All of these functions behave correctly + * when presented with an infinity or NaN. + * + * The following IEEE 'real' formats are currently supported: + * $(UL + * $(LI 64 bit Big-endian 'double' (eg PowerPC)) + * $(LI 128 bit Big-endian 'quadruple' (eg SPARC)) + * $(LI 64 bit Little-endian 'double' (eg x86-SSE2)) + * $(LI 80 bit Little-endian, with implied bit 'real80' (eg x87, Itanium)) + * $(LI 128 bit Little-endian 'quadruple' (not implemented on any known processor!)) + * $(LI Non-IEEE 128 bit Big-endian 'doubledouble' (eg PowerPC) has partial support) + * ) + * Unlike C, there is no global 'errno' variable. Consequently, almost all of + * these functions are pure nothrow. + * + * Status: + * The semantics and names of feqrel and approxEqual will be revised. + * + * Macros: + * TABLE_SV = <table border="1" cellpadding="4" cellspacing="0"> + * <caption>Special Values</caption> + * $0</table> + * SVH = $(TR $(TH $1) $(TH $2)) + * SV = $(TR $(TD $1) $(TD $2)) + * TH3 = $(TR $(TH $1) $(TH $2) $(TH $3)) + * TD3 = $(TR $(TD $1) $(TD $2) $(TD $3)) + * TABLE_DOMRG = <table border="1" cellpadding="4" cellspacing="0"> + * $(SVH Domain X, Range Y) + $(SV $1, $2) + * </table> + * DOMAIN=$1 + * RANGE=$1 + + * NAN = $(RED NAN) + * SUP = <span style="vertical-align:super;font-size:smaller">$0</span> + * GAMMA = Γ + * THETA = θ + * INTEGRAL = ∫ + * INTEGRATE = $(BIG ∫<sub>$(SMALL $1)</sub><sup>$2</sup>) + * POWER = $1<sup>$2</sup> + * SUB = $1<sub>$2</sub> + * BIGSUM = $(BIG Σ <sup>$2</sup><sub>$(SMALL $1)</sub>) + * CHOOSE = $(BIG () <sup>$(SMALL $1)</sup><sub>$(SMALL $2)</sub> $(BIG )) + * PLUSMN = ± + * INFIN = ∞ + * PLUSMNINF = ±∞ + * PI = π + * LT = < + * GT = > + * SQRT = √ + * HALF = ½ + * + * Copyright: Copyright Digital Mars 2000 - 2011. + * D implementations of tan, atan, atan2, exp, expm1, exp2, log, log10, log1p, + * log2, floor, ceil and lrint functions are based on the CEPHES math library, + * which is Copyright (C) 2001 Stephen L. Moshier $(LT)steve@moshier.net$(GT) + * and are incorporated herein by permission of the author. The author + * reserves the right to distribute this material elsewhere under different + * copying permissions. These modifications are distributed here under + * the following terms: + * License: $(HTTP www.boost.org/LICENSE_1_0.txt, Boost License 1.0). + * Authors: $(HTTP digitalmars.com, Walter Bright), Don Clugston, + * Conversion of CEPHES math library to D by Iain Buclaw and David Nadlinger + * Source: $(PHOBOSSRC std/_math.d) + */ + +/* NOTE: This file has been patched from the original DMD distribution to + * work with the GDC compiler. + */ +module std.math; + +version (Win64) +{ + version (D_InlineAsm_X86_64) + version = Win64_DMD_InlineAsm; +} + +static import core.math; +static import core.stdc.math; +static import core.stdc.fenv; +import std.traits; // CommonType, isFloatingPoint, isIntegral, isSigned, isUnsigned, Largest, Unqual + +version (LDC) +{ + import ldc.intrinsics; +} + +version (DigitalMars) +{ + version = INLINE_YL2X; // x87 has opcodes for these +} + +version (X86) version = X86_Any; +version (X86_64) version = X86_Any; +version (PPC) version = PPC_Any; +version (PPC64) version = PPC_Any; +version (MIPS32) version = MIPS_Any; +version (MIPS64) version = MIPS_Any; +version (AArch64) version = ARM_Any; +version (ARM) version = ARM_Any; + +version (D_InlineAsm_X86) +{ + version = InlineAsm_X86_Any; +} +else version (D_InlineAsm_X86_64) +{ + version = InlineAsm_X86_Any; +} + +version (X86_64) version = StaticallyHaveSSE; +version (X86) version (OSX) version = StaticallyHaveSSE; + +version (StaticallyHaveSSE) +{ + private enum bool haveSSE = true; +} +else +{ + static import core.cpuid; + private alias haveSSE = core.cpuid.sse; +} + +version (unittest) +{ + import core.stdc.stdio; // : sprintf; + + static if (real.sizeof > double.sizeof) + enum uint useDigits = 16; + else + enum uint useDigits = 15; + + /****************************************** + * Compare floating point numbers to n decimal digits of precision. + * Returns: + * 1 match + * 0 nomatch + */ + + private bool equalsDigit(real x, real y, uint ndigits) + { + if (signbit(x) != signbit(y)) + return 0; + + if (isInfinity(x) && isInfinity(y)) + return 1; + if (isInfinity(x) || isInfinity(y)) + return 0; + + if (isNaN(x) && isNaN(y)) + return 1; + if (isNaN(x) || isNaN(y)) + return 0; + + char[30] bufx; + char[30] bufy; + assert(ndigits < bufx.length); + + int ix; + int iy; + version (CRuntime_Microsoft) + alias real_t = double; + else + alias real_t = real; + ix = sprintf(bufx.ptr, "%.*Lg", ndigits, cast(real_t) x); + iy = sprintf(bufy.ptr, "%.*Lg", ndigits, cast(real_t) y); + assert(ix < bufx.length && ix > 0); + assert(ix < bufy.length && ix > 0); + + return bufx[0 .. ix] == bufy[0 .. iy]; + } +} + + + +package: +// The following IEEE 'real' formats are currently supported. +version (LittleEndian) +{ + static assert(real.mant_dig == 53 || real.mant_dig == 64 + || real.mant_dig == 113, + "Only 64-bit, 80-bit, and 128-bit reals"~ + " are supported for LittleEndian CPUs"); +} +else +{ + static assert(real.mant_dig == 53 || real.mant_dig == 106 + || real.mant_dig == 113, + "Only 64-bit and 128-bit reals are supported for BigEndian CPUs."~ + " double-double reals have partial support"); +} + +// Underlying format exposed through floatTraits +enum RealFormat +{ + ieeeHalf, + ieeeSingle, + ieeeDouble, + ieeeExtended, // x87 80-bit real + ieeeExtended53, // x87 real rounded to precision of double. + ibmExtended, // IBM 128-bit extended + ieeeQuadruple, +} + +// Constants used for extracting the components of the representation. +// They supplement the built-in floating point properties. +template floatTraits(T) +{ + // EXPMASK is a ushort mask to select the exponent portion (without sign) + // EXPSHIFT is the number of bits the exponent is left-shifted by in its ushort + // EXPBIAS is the exponent bias - 1 (exp == EXPBIAS yields ×2^-1). + // EXPPOS_SHORT is the index of the exponent when represented as a ushort array. + // SIGNPOS_BYTE is the index of the sign when represented as a ubyte array. + // RECIP_EPSILON is the value such that (smallest_subnormal) * RECIP_EPSILON == T.min_normal + enum T RECIP_EPSILON = (1/T.epsilon); + static if (T.mant_dig == 24) + { + // Single precision float + enum ushort EXPMASK = 0x7F80; + enum ushort EXPSHIFT = 7; + enum ushort EXPBIAS = 0x3F00; + enum uint EXPMASK_INT = 0x7F80_0000; + enum uint MANTISSAMASK_INT = 0x007F_FFFF; + enum realFormat = RealFormat.ieeeSingle; + version (LittleEndian) + { + enum EXPPOS_SHORT = 1; + enum SIGNPOS_BYTE = 3; + } + else + { + enum EXPPOS_SHORT = 0; + enum SIGNPOS_BYTE = 0; + } + } + else static if (T.mant_dig == 53) + { + static if (T.sizeof == 8) + { + // Double precision float, or real == double + enum ushort EXPMASK = 0x7FF0; + enum ushort EXPSHIFT = 4; + enum ushort EXPBIAS = 0x3FE0; + enum uint EXPMASK_INT = 0x7FF0_0000; + enum uint MANTISSAMASK_INT = 0x000F_FFFF; // for the MSB only + enum realFormat = RealFormat.ieeeDouble; + version (LittleEndian) + { + enum EXPPOS_SHORT = 3; + enum SIGNPOS_BYTE = 7; + } + else + { + enum EXPPOS_SHORT = 0; + enum SIGNPOS_BYTE = 0; + } + } + else static if (T.sizeof == 12) + { + // Intel extended real80 rounded to double + enum ushort EXPMASK = 0x7FFF; + enum ushort EXPSHIFT = 0; + enum ushort EXPBIAS = 0x3FFE; + enum realFormat = RealFormat.ieeeExtended53; + version (LittleEndian) + { + enum EXPPOS_SHORT = 4; + enum SIGNPOS_BYTE = 9; + } + else + { + enum EXPPOS_SHORT = 0; + enum SIGNPOS_BYTE = 0; + } + } + else + static assert(false, "No traits support for " ~ T.stringof); + } + else static if (T.mant_dig == 64) + { + // Intel extended real80 + enum ushort EXPMASK = 0x7FFF; + enum ushort EXPSHIFT = 0; + enum ushort EXPBIAS = 0x3FFE; + enum realFormat = RealFormat.ieeeExtended; + version (LittleEndian) + { + enum EXPPOS_SHORT = 4; + enum SIGNPOS_BYTE = 9; + } + else + { + enum EXPPOS_SHORT = 0; + enum SIGNPOS_BYTE = 0; + } + } + else static if (T.mant_dig == 113) + { + // Quadruple precision float + enum ushort EXPMASK = 0x7FFF; + enum ushort EXPSHIFT = 0; + enum ushort EXPBIAS = 0x3FFE; + enum realFormat = RealFormat.ieeeQuadruple; + version (LittleEndian) + { + enum EXPPOS_SHORT = 7; + enum SIGNPOS_BYTE = 15; + } + else + { + enum EXPPOS_SHORT = 0; + enum SIGNPOS_BYTE = 0; + } + } + else static if (T.mant_dig == 106) + { + // IBM Extended doubledouble + enum ushort EXPMASK = 0x7FF0; + enum ushort EXPSHIFT = 4; + enum realFormat = RealFormat.ibmExtended; + // the exponent byte is not unique + version (LittleEndian) + { + enum EXPPOS_SHORT = 7; // [3] is also an exp short + enum SIGNPOS_BYTE = 15; + } + else + { + enum EXPPOS_SHORT = 0; // [4] is also an exp short + enum SIGNPOS_BYTE = 0; + } + } + else + static assert(false, "No traits support for " ~ T.stringof); +} + +// These apply to all floating-point types +version (LittleEndian) +{ + enum MANTISSA_LSB = 0; + enum MANTISSA_MSB = 1; +} +else +{ + enum MANTISSA_LSB = 1; + enum MANTISSA_MSB = 0; +} + +// Common code for math implementations. + +// Helper for floor/ceil +T floorImpl(T)(const T x) @trusted pure nothrow @nogc +{ + alias F = floatTraits!(T); + // Take care not to trigger library calls from the compiler, + // while ensuring that we don't get defeated by some optimizers. + union floatBits + { + T rv; + ushort[T.sizeof/2] vu; + } + floatBits y = void; + y.rv = x; + + // Find the exponent (power of 2) + static if (F.realFormat == RealFormat.ieeeSingle) + { + int exp = ((y.vu[F.EXPPOS_SHORT] >> 7) & 0xff) - 0x7f; + + version (LittleEndian) + int pos = 0; + else + int pos = 3; + } + else static if (F.realFormat == RealFormat.ieeeDouble) + { + int exp = ((y.vu[F.EXPPOS_SHORT] >> 4) & 0x7ff) - 0x3ff; + + version (LittleEndian) + int pos = 0; + else + int pos = 3; + } + else static if (F.realFormat == RealFormat.ieeeExtended) + { + int exp = (y.vu[F.EXPPOS_SHORT] & 0x7fff) - 0x3fff; + + version (LittleEndian) + int pos = 0; + else + int pos = 4; + } + else static if (F.realFormat == RealFormat.ieeeQuadruple) + { + int exp = (y.vu[F.EXPPOS_SHORT] & 0x7fff) - 0x3fff; + + version (LittleEndian) + int pos = 0; + else + int pos = 7; + } + else + static assert(false, "Not implemented for this architecture"); + + if (exp < 0) + { + if (x < 0.0) + return -1.0; + else + return 0.0; + } + + exp = (T.mant_dig - 1) - exp; + + // Zero 16 bits at a time. + while (exp >= 16) + { + version (LittleEndian) + y.vu[pos++] = 0; + else + y.vu[pos--] = 0; + exp -= 16; + } + + // Clear the remaining bits. + if (exp > 0) + y.vu[pos] &= 0xffff ^ ((1 << exp) - 1); + + if ((x < 0.0) && (x != y.rv)) + y.rv -= 1.0; + + return y.rv; +} + +public: + +// Values obtained from Wolfram Alpha. 116 bits ought to be enough for anybody. +// Wolfram Alpha LLC. 2011. Wolfram|Alpha. http://www.wolframalpha.com/input/?i=e+in+base+16 (access July 6, 2011). +enum real E = 0x1.5bf0a8b1457695355fb8ac404e7a8p+1L; /** e = 2.718281... */ +enum real LOG2T = 0x1.a934f0979a3715fc9257edfe9b5fbp+1L; /** $(SUB log, 2)10 = 3.321928... */ +enum real LOG2E = 0x1.71547652b82fe1777d0ffda0d23a8p+0L; /** $(SUB log, 2)e = 1.442695... */ +enum real LOG2 = 0x1.34413509f79fef311f12b35816f92p-2L; /** $(SUB log, 10)2 = 0.301029... */ +enum real LOG10E = 0x1.bcb7b1526e50e32a6ab7555f5a67cp-2L; /** $(SUB log, 10)e = 0.434294... */ +enum real LN2 = 0x1.62e42fefa39ef35793c7673007e5fp-1L; /** ln 2 = 0.693147... */ +enum real LN10 = 0x1.26bb1bbb5551582dd4adac5705a61p+1L; /** ln 10 = 2.302585... */ +enum real PI = 0x1.921fb54442d18469898cc51701b84p+1L; /** $(_PI) = 3.141592... */ +enum real PI_2 = PI/2; /** $(PI) / 2 = 1.570796... */ +enum real PI_4 = PI/4; /** $(PI) / 4 = 0.785398... */ +enum real M_1_PI = 0x1.45f306dc9c882a53f84eafa3ea69cp-2L; /** 1 / $(PI) = 0.318309... */ +enum real M_2_PI = 2*M_1_PI; /** 2 / $(PI) = 0.636619... */ +enum real M_2_SQRTPI = 0x1.20dd750429b6d11ae3a914fed7fd8p+0L; /** 2 / $(SQRT)$(PI) = 1.128379... */ +enum real SQRT2 = 0x1.6a09e667f3bcc908b2fb1366ea958p+0L; /** $(SQRT)2 = 1.414213... */ +enum real SQRT1_2 = SQRT2/2; /** $(SQRT)$(HALF) = 0.707106... */ +// Note: Make sure the magic numbers in compiler backend for x87 match these. + + +/*********************************** + * Calculates the absolute value of a number + * + * Params: + * Num = (template parameter) type of number + * x = real number value + * z = complex number value + * y = imaginary number value + * + * Returns: + * The absolute value of the number. If floating-point or integral, + * the return type will be the same as the input; if complex or + * imaginary, the returned value will be the corresponding floating + * point type. + * + * For complex numbers, abs(z) = sqrt( $(POWER z.re, 2) + $(POWER z.im, 2) ) + * = hypot(z.re, z.im). + */ +Num abs(Num)(Num x) @safe pure nothrow +if (is(typeof(Num.init >= 0)) && is(typeof(-Num.init)) && + !(is(Num* : const(ifloat*)) || is(Num* : const(idouble*)) + || is(Num* : const(ireal*)))) +{ + static if (isFloatingPoint!(Num)) + return fabs(x); + else + return x >= 0 ? x : -x; +} + +/// ditto +auto abs(Num)(Num z) @safe pure nothrow @nogc +if (is(Num* : const(cfloat*)) || is(Num* : const(cdouble*)) + || is(Num* : const(creal*))) +{ + return hypot(z.re, z.im); +} + +/// ditto +auto abs(Num)(Num y) @safe pure nothrow @nogc +if (is(Num* : const(ifloat*)) || is(Num* : const(idouble*)) + || is(Num* : const(ireal*))) +{ + return fabs(y.im); +} + +/// ditto +@safe pure nothrow @nogc unittest +{ + assert(isIdentical(abs(-0.0L), 0.0L)); + assert(isNaN(abs(real.nan))); + assert(abs(-real.infinity) == real.infinity); + assert(abs(-3.2Li) == 3.2L); + assert(abs(71.6Li) == 71.6L); + assert(abs(-56) == 56); + assert(abs(2321312L) == 2321312L); + assert(abs(-1L+1i) == sqrt(2.0L)); +} + +@safe pure nothrow @nogc unittest +{ + import std.meta : AliasSeq; + foreach (T; AliasSeq!(float, double, real)) + { + T f = 3; + assert(abs(f) == f); + assert(abs(-f) == f); + } + foreach (T; AliasSeq!(cfloat, cdouble, creal)) + { + T f = -12+3i; + assert(abs(f) == hypot(f.re, f.im)); + assert(abs(-f) == hypot(f.re, f.im)); + } +} + +/*********************************** + * Complex conjugate + * + * conj(x + iy) = x - iy + * + * Note that z * conj(z) = $(POWER z.re, 2) - $(POWER z.im, 2) + * is always a real number + */ +auto conj(Num)(Num z) @safe pure nothrow @nogc +if (is(Num* : const(cfloat*)) || is(Num* : const(cdouble*)) + || is(Num* : const(creal*))) +{ + //FIXME + //Issue 14206 + static if (is(Num* : const(cdouble*))) + return cast(cdouble) conj(cast(creal) z); + else + return z.re - z.im*1fi; +} + +/** ditto */ +auto conj(Num)(Num y) @safe pure nothrow @nogc +if (is(Num* : const(ifloat*)) || is(Num* : const(idouble*)) + || is(Num* : const(ireal*))) +{ + return -y; +} + +/// +@safe pure nothrow @nogc unittest +{ + creal c = 7 + 3Li; + assert(conj(c) == 7-3Li); + ireal z = -3.2Li; + assert(conj(z) == -z); +} +//Issue 14206 +@safe pure nothrow @nogc unittest +{ + cdouble c = 7 + 3i; + assert(conj(c) == 7-3i); + idouble z = -3.2i; + assert(conj(z) == -z); +} +//Issue 14206 +@safe pure nothrow @nogc unittest +{ + cfloat c = 7f + 3fi; + assert(conj(c) == 7f-3fi); + ifloat z = -3.2fi; + assert(conj(z) == -z); +} + +/*********************************** + * Returns cosine of x. x is in radians. + * + * $(TABLE_SV + * $(TR $(TH x) $(TH cos(x)) $(TH invalid?)) + * $(TR $(TD $(NAN)) $(TD $(NAN)) $(TD yes) ) + * $(TR $(TD $(PLUSMN)$(INFIN)) $(TD $(NAN)) $(TD yes) ) + * ) + * Bugs: + * Results are undefined if |x| >= $(POWER 2,64). + */ + +real cos(real x) @safe pure nothrow @nogc { pragma(inline, true); return core.math.cos(x); } +//FIXME +///ditto +double cos(double x) @safe pure nothrow @nogc { return cos(cast(real) x); } +//FIXME +///ditto +float cos(float x) @safe pure nothrow @nogc { return cos(cast(real) x); } + +@safe unittest +{ + real function(real) pcos = &cos; + assert(pcos != null); +} + +/*********************************** + * Returns $(HTTP en.wikipedia.org/wiki/Sine, sine) of x. x is in $(HTTP en.wikipedia.org/wiki/Radian, radians). + * + * $(TABLE_SV + * $(TH3 x , sin(x) , invalid?) + * $(TD3 $(NAN) , $(NAN) , yes ) + * $(TD3 $(PLUSMN)0.0, $(PLUSMN)0.0, no ) + * $(TD3 $(PLUSMNINF), $(NAN) , yes ) + * ) + * + * Params: + * x = angle in radians (not degrees) + * Returns: + * sine of x + * See_Also: + * $(MYREF cos), $(MYREF tan), $(MYREF asin) + * Bugs: + * Results are undefined if |x| >= $(POWER 2,64). + */ + +real sin(real x) @safe pure nothrow @nogc { pragma(inline, true); return core.math.sin(x); } +//FIXME +///ditto +double sin(double x) @safe pure nothrow @nogc { return sin(cast(real) x); } +//FIXME +///ditto +float sin(float x) @safe pure nothrow @nogc { return sin(cast(real) x); } + +/// +@safe unittest +{ + import std.math : sin, PI; + import std.stdio : writefln; + + void someFunc() + { + real x = 30.0; + auto result = sin(x * (PI / 180)); // convert degrees to radians + writefln("The sine of %s degrees is %s", x, result); + } +} + +@safe unittest +{ + real function(real) psin = &sin; + assert(psin != null); +} + +/*********************************** + * Returns sine for complex and imaginary arguments. + * + * sin(z) = sin(z.re)*cosh(z.im) + cos(z.re)*sinh(z.im)i + * + * If both sin($(THETA)) and cos($(THETA)) are required, + * it is most efficient to use expi($(THETA)). + */ +creal sin(creal z) @safe pure nothrow @nogc +{ + const creal cs = expi(z.re); + const creal csh = coshisinh(z.im); + return cs.im * csh.re + cs.re * csh.im * 1i; +} + +/** ditto */ +ireal sin(ireal y) @safe pure nothrow @nogc +{ + return cosh(y.im)*1i; +} + +/// +@safe pure nothrow @nogc unittest +{ + assert(sin(0.0+0.0i) == 0.0); + assert(sin(2.0+0.0i) == sin(2.0L) ); +} + +/*********************************** + * cosine, complex and imaginary + * + * cos(z) = cos(z.re)*cosh(z.im) - sin(z.re)*sinh(z.im)i + */ +creal cos(creal z) @safe pure nothrow @nogc +{ + const creal cs = expi(z.re); + const creal csh = coshisinh(z.im); + return cs.re * csh.re - cs.im * csh.im * 1i; +} + +/** ditto */ +real cos(ireal y) @safe pure nothrow @nogc +{ + return cosh(y.im); +} + +/// +@safe pure nothrow @nogc unittest +{ + assert(cos(0.0+0.0i)==1.0); + assert(cos(1.3L+0.0i)==cos(1.3L)); + assert(cos(5.2Li)== cosh(5.2L)); +} + +/**************************************************************************** + * Returns tangent of x. x is in radians. + * + * $(TABLE_SV + * $(TR $(TH x) $(TH tan(x)) $(TH invalid?)) + * $(TR $(TD $(NAN)) $(TD $(NAN)) $(TD yes)) + * $(TR $(TD $(PLUSMN)0.0) $(TD $(PLUSMN)0.0) $(TD no)) + * $(TR $(TD $(PLUSMNINF)) $(TD $(NAN)) $(TD yes)) + * ) + */ + +real tan(real x) @trusted pure nothrow @nogc +{ + version (D_InlineAsm_X86) + { + asm pure nothrow @nogc + { + fld x[EBP] ; // load theta + fxam ; // test for oddball values + fstsw AX ; + sahf ; + jc trigerr ; // x is NAN, infinity, or empty + // 387's can handle subnormals +SC18: fptan ; + fstsw AX ; + sahf ; + jnp Clear1 ; // C2 = 1 (x is out of range) + + // Do argument reduction to bring x into range + fldpi ; + fxch ; +SC17: fprem1 ; + fstsw AX ; + sahf ; + jp SC17 ; + fstp ST(1) ; // remove pi from stack + jmp SC18 ; + +trigerr: + jnp Lret ; // if theta is NAN, return theta + fstp ST(0) ; // dump theta + } + return real.nan; + +Clear1: asm pure nothrow @nogc{ + fstp ST(0) ; // dump X, which is always 1 + } + +Lret: {} + } + else version (D_InlineAsm_X86_64) + { + version (Win64) + { + asm pure nothrow @nogc + { + fld real ptr [RCX] ; // load theta + } + } + else + { + asm pure nothrow @nogc + { + fld x[RBP] ; // load theta + } + } + asm pure nothrow @nogc + { + fxam ; // test for oddball values + fstsw AX ; + test AH,1 ; + jnz trigerr ; // x is NAN, infinity, or empty + // 387's can handle subnormals +SC18: fptan ; + fstsw AX ; + test AH,4 ; + jz Clear1 ; // C2 = 1 (x is out of range) + + // Do argument reduction to bring x into range + fldpi ; + fxch ; +SC17: fprem1 ; + fstsw AX ; + test AH,4 ; + jnz SC17 ; + fstp ST(1) ; // remove pi from stack + jmp SC18 ; + +trigerr: + test AH,4 ; + jz Lret ; // if theta is NAN, return theta + fstp ST(0) ; // dump theta + } + return real.nan; + +Clear1: asm pure nothrow @nogc{ + fstp ST(0) ; // dump X, which is always 1 + } + +Lret: {} + } + else + { + // Coefficients for tan(x) and PI/4 split into three parts. + static if (floatTraits!real.realFormat == RealFormat.ieeeQuadruple) + { + static immutable real[6] P = [ + 2.883414728874239697964612246732416606301E10L, + -2.307030822693734879744223131873392503321E9L, + 5.160188250214037865511600561074819366815E7L, + -4.249691853501233575668486667664718192660E5L, + 1.272297782199996882828849455156962260810E3L, + -9.889929415807650724957118893791829849557E-1L + ]; + static immutable real[7] Q = [ + 8.650244186622719093893836740197250197602E10L + -4.152206921457208101480801635640958361612E10L, + 2.758476078803232151774723646710890525496E9L, + -5.733709132766856723608447733926138506824E7L, + 4.529422062441341616231663543669583527923E5L, + -1.317243702830553658702531997959756728291E3L, + 1.0 + ]; + + enum real P1 = + 7.853981633974483067550664827649598009884357452392578125E-1L; + enum real P2 = + 2.8605943630549158983813312792950660807511260829685741796657E-18L; + enum real P3 = + 2.1679525325309452561992610065108379921905808E-35L; + } + else + { + static immutable real[3] P = [ + -1.7956525197648487798769E7L, + 1.1535166483858741613983E6L, + -1.3093693918138377764608E4L, + ]; + static immutable real[5] Q = [ + -5.3869575592945462988123E7L, + 2.5008380182335791583922E7L, + -1.3208923444021096744731E6L, + 1.3681296347069295467845E4L, + 1.0000000000000000000000E0L, + ]; + + enum real P1 = 7.853981554508209228515625E-1L; + enum real P2 = 7.946627356147928367136046290398E-9L; + enum real P3 = 3.061616997868382943065164830688E-17L; + } + + // Special cases. + if (x == 0.0 || isNaN(x)) + return x; + if (isInfinity(x)) + return real.nan; + + // Make argument positive but save the sign. + bool sign = false; + if (signbit(x)) + { + sign = true; + x = -x; + } + + // Compute x mod PI/4. + real y = floor(x / PI_4); + // Strip high bits of integer part. + real z = ldexp(y, -4); + // Compute y - 16 * (y / 16). + z = y - ldexp(floor(z), 4); + + // Integer and fraction part modulo one octant. + int j = cast(int)(z); + + // Map zeros and singularities to origin. + if (j & 1) + { + j += 1; + y += 1.0; + } + + z = ((x - y * P1) - y * P2) - y * P3; + const real zz = z * z; + + if (zz > 1.0e-20L) + y = z + z * (zz * poly(zz, P) / poly(zz, Q)); + else + y = z; + + if (j & 2) + y = -1.0 / y; + + return (sign) ? -y : y; + } +} + +@safe nothrow @nogc unittest +{ + static real[2][] vals = // angle,tan + [ + [ 0, 0], + [ .5, .5463024898], + [ 1, 1.557407725], + [ 1.5, 14.10141995], + [ 2, -2.185039863], + [ 2.5,-.7470222972], + [ 3, -.1425465431], + [ 3.5, .3745856402], + [ 4, 1.157821282], + [ 4.5, 4.637332055], + [ 5, -3.380515006], + [ 5.5,-.9955840522], + [ 6, -.2910061914], + [ 6.5, .2202772003], + [ 10, .6483608275], + + // special angles + [ PI_4, 1], + //[ PI_2, real.infinity], // PI_2 is not _exactly_ pi/2. + [ 3*PI_4, -1], + [ PI, 0], + [ 5*PI_4, 1], + //[ 3*PI_2, -real.infinity], + [ 7*PI_4, -1], + [ 2*PI, 0], + ]; + int i; + + for (i = 0; i < vals.length; i++) + { + real x = vals[i][0]; + real r = vals[i][1]; + real t = tan(x); + + //printf("tan(%Lg) = %Lg, should be %Lg\n", x, t, r); + if (!isIdentical(r, t)) assert(fabs(r-t) <= .0000001); + + x = -x; + r = -r; + t = tan(x); + //printf("tan(%Lg) = %Lg, should be %Lg\n", x, t, r); + if (!isIdentical(r, t) && !(r != r && t != t)) assert(fabs(r-t) <= .0000001); + } + // overflow + assert(isNaN(tan(real.infinity))); + assert(isNaN(tan(-real.infinity))); + // NaN propagation + assert(isIdentical( tan(NaN(0x0123L)), NaN(0x0123L) )); +} + +@system unittest +{ + assert(equalsDigit(tan(PI / 3), std.math.sqrt(3.0), useDigits)); +} + +/*************** + * Calculates the arc cosine of x, + * returning a value ranging from 0 to $(PI). + * + * $(TABLE_SV + * $(TR $(TH x) $(TH acos(x)) $(TH invalid?)) + * $(TR $(TD $(GT)1.0) $(TD $(NAN)) $(TD yes)) + * $(TR $(TD $(LT)-1.0) $(TD $(NAN)) $(TD yes)) + * $(TR $(TD $(NAN)) $(TD $(NAN)) $(TD yes)) + * ) + */ +real acos(real x) @safe pure nothrow @nogc +{ + return atan2(sqrt(1-x*x), x); +} + +/// ditto +double acos(double x) @safe pure nothrow @nogc { return acos(cast(real) x); } + +/// ditto +float acos(float x) @safe pure nothrow @nogc { return acos(cast(real) x); } + +@system unittest +{ + assert(equalsDigit(acos(0.5), std.math.PI / 3, useDigits)); +} + +/*************** + * Calculates the arc sine of x, + * returning a value ranging from -$(PI)/2 to $(PI)/2. + * + * $(TABLE_SV + * $(TR $(TH x) $(TH asin(x)) $(TH invalid?)) + * $(TR $(TD $(PLUSMN)0.0) $(TD $(PLUSMN)0.0) $(TD no)) + * $(TR $(TD $(GT)1.0) $(TD $(NAN)) $(TD yes)) + * $(TR $(TD $(LT)-1.0) $(TD $(NAN)) $(TD yes)) + * ) + */ +real asin(real x) @safe pure nothrow @nogc +{ + return atan2(x, sqrt(1-x*x)); +} + +/// ditto +double asin(double x) @safe pure nothrow @nogc { return asin(cast(real) x); } + +/// ditto +float asin(float x) @safe pure nothrow @nogc { return asin(cast(real) x); } + +@system unittest +{ + assert(equalsDigit(asin(0.5), PI / 6, useDigits)); +} + +/*************** + * Calculates the arc tangent of x, + * returning a value ranging from -$(PI)/2 to $(PI)/2. + * + * $(TABLE_SV + * $(TR $(TH x) $(TH atan(x)) $(TH invalid?)) + * $(TR $(TD $(PLUSMN)0.0) $(TD $(PLUSMN)0.0) $(TD no)) + * $(TR $(TD $(PLUSMN)$(INFIN)) $(TD $(NAN)) $(TD yes)) + * ) + */ +real atan(real x) @safe pure nothrow @nogc +{ + version (InlineAsm_X86_Any) + { + return atan2(x, 1.0L); + } + else + { + // Coefficients for atan(x) + static if (floatTraits!real.realFormat == RealFormat.ieeeQuadruple) + { + static immutable real[9] P = [ + -6.880597774405940432145577545328795037141E2L, + -2.514829758941713674909996882101723647996E3L, + -3.696264445691821235400930243493001671932E3L, + -2.792272753241044941703278827346430350236E3L, + -1.148164399808514330375280133523543970854E3L, + -2.497759878476618348858065206895055957104E2L, + -2.548067867495502632615671450650071218995E1L, + -8.768423468036849091777415076702113400070E-1L, + -6.635810778635296712545011270011752799963E-4L + ]; + static immutable real[9] Q = [ + 2.064179332321782129643673263598686441900E3L, + 8.782996876218210302516194604424986107121E3L, + 1.547394317752562611786521896296215170819E4L, + 1.458510242529987155225086911411015961174E4L, + 7.928572347062145288093560392463784743935E3L, + 2.494680540950601626662048893678584497900E3L, + 4.308348370818927353321556740027020068897E2L, + 3.566239794444800849656497338030115886153E1L, + 1.0 + ]; + } + else + { + static immutable real[5] P = [ + -5.0894116899623603312185E1L, + -9.9988763777265819915721E1L, + -6.3976888655834347413154E1L, + -1.4683508633175792446076E1L, + -8.6863818178092187535440E-1L, + ]; + static immutable real[6] Q = [ + 1.5268235069887081006606E2L, + 3.9157570175111990631099E2L, + 3.6144079386152023162701E2L, + 1.4399096122250781605352E2L, + 2.2981886733594175366172E1L, + 1.0000000000000000000000E0L, + ]; + } + + // tan(PI/8) + enum real TAN_PI_8 = 0.414213562373095048801688724209698078569672L; + // tan(3 * PI/8) + enum real TAN3_PI_8 = 2.414213562373095048801688724209698078569672L; + + // Special cases. + if (x == 0.0) + return x; + if (isInfinity(x)) + return copysign(PI_2, x); + + // Make argument positive but save the sign. + bool sign = false; + if (signbit(x)) + { + sign = true; + x = -x; + } + + // Range reduction. + real y; + if (x > TAN3_PI_8) + { + y = PI_2; + x = -(1.0 / x); + } + else if (x > TAN_PI_8) + { + y = PI_4; + x = (x - 1.0)/(x + 1.0); + } + else + y = 0.0; + + // Rational form in x^^2. + const real z = x * x; + y = y + (poly(z, P) / poly(z, Q)) * z * x + x; + + return (sign) ? -y : y; + } +} + +/// ditto +double atan(double x) @safe pure nothrow @nogc { return atan(cast(real) x); } + +/// ditto +float atan(float x) @safe pure nothrow @nogc { return atan(cast(real) x); } + +@system unittest +{ + assert(equalsDigit(atan(std.math.sqrt(3.0)), PI / 3, useDigits)); +} + +/*************** + * Calculates the arc tangent of y / x, + * returning a value ranging from -$(PI) to $(PI). + * + * $(TABLE_SV + * $(TR $(TH y) $(TH x) $(TH atan(y, x))) + * $(TR $(TD $(NAN)) $(TD anything) $(TD $(NAN)) ) + * $(TR $(TD anything) $(TD $(NAN)) $(TD $(NAN)) ) + * $(TR $(TD $(PLUSMN)0.0) $(TD $(GT)0.0) $(TD $(PLUSMN)0.0) ) + * $(TR $(TD $(PLUSMN)0.0) $(TD +0.0) $(TD $(PLUSMN)0.0) ) + * $(TR $(TD $(PLUSMN)0.0) $(TD $(LT)0.0) $(TD $(PLUSMN)$(PI))) + * $(TR $(TD $(PLUSMN)0.0) $(TD -0.0) $(TD $(PLUSMN)$(PI))) + * $(TR $(TD $(GT)0.0) $(TD $(PLUSMN)0.0) $(TD $(PI)/2) ) + * $(TR $(TD $(LT)0.0) $(TD $(PLUSMN)0.0) $(TD -$(PI)/2) ) + * $(TR $(TD $(GT)0.0) $(TD $(INFIN)) $(TD $(PLUSMN)0.0) ) + * $(TR $(TD $(PLUSMN)$(INFIN)) $(TD anything) $(TD $(PLUSMN)$(PI)/2)) + * $(TR $(TD $(GT)0.0) $(TD -$(INFIN)) $(TD $(PLUSMN)$(PI)) ) + * $(TR $(TD $(PLUSMN)$(INFIN)) $(TD $(INFIN)) $(TD $(PLUSMN)$(PI)/4)) + * $(TR $(TD $(PLUSMN)$(INFIN)) $(TD -$(INFIN)) $(TD $(PLUSMN)3$(PI)/4)) + * ) + */ +real atan2(real y, real x) @trusted pure nothrow @nogc +{ + version (InlineAsm_X86_Any) + { + version (Win64) + { + asm pure nothrow @nogc { + naked; + fld real ptr [RDX]; // y + fld real ptr [RCX]; // x + fpatan; + ret; + } + } + else + { + asm pure nothrow @nogc { + fld y; + fld x; + fpatan; + } + } + } + else + { + // Special cases. + if (isNaN(x) || isNaN(y)) + return real.nan; + if (y == 0.0) + { + if (x >= 0 && !signbit(x)) + return copysign(0, y); + else + return copysign(PI, y); + } + if (x == 0.0) + return copysign(PI_2, y); + if (isInfinity(x)) + { + if (signbit(x)) + { + if (isInfinity(y)) + return copysign(3*PI_4, y); + else + return copysign(PI, y); + } + else + { + if (isInfinity(y)) + return copysign(PI_4, y); + else + return copysign(0.0, y); + } + } + if (isInfinity(y)) + return copysign(PI_2, y); + + // Call atan and determine the quadrant. + real z = atan(y / x); + + if (signbit(x)) + { + if (signbit(y)) + z = z - PI; + else + z = z + PI; + } + + if (z == 0.0) + return copysign(z, y); + + return z; + } +} + +/// ditto +double atan2(double y, double x) @safe pure nothrow @nogc +{ + return atan2(cast(real) y, cast(real) x); +} + +/// ditto +float atan2(float y, float x) @safe pure nothrow @nogc +{ + return atan2(cast(real) y, cast(real) x); +} + +@system unittest +{ + assert(equalsDigit(atan2(1.0L, std.math.sqrt(3.0L)), PI / 6, useDigits)); +} + +/*********************************** + * Calculates the hyperbolic cosine of x. + * + * $(TABLE_SV + * $(TR $(TH x) $(TH cosh(x)) $(TH invalid?)) + * $(TR $(TD $(PLUSMN)$(INFIN)) $(TD $(PLUSMN)0.0) $(TD no) ) + * ) + */ +real cosh(real x) @safe pure nothrow @nogc +{ + // cosh = (exp(x)+exp(-x))/2. + // The naive implementation works correctly. + const real y = exp(x); + return (y + 1.0/y) * 0.5; +} + +/// ditto +double cosh(double x) @safe pure nothrow @nogc { return cosh(cast(real) x); } + +/// ditto +float cosh(float x) @safe pure nothrow @nogc { return cosh(cast(real) x); } + +@system unittest +{ + assert(equalsDigit(cosh(1.0), (E + 1.0 / E) / 2, useDigits)); +} + +/*********************************** + * Calculates the hyperbolic sine of x. + * + * $(TABLE_SV + * $(TR $(TH x) $(TH sinh(x)) $(TH invalid?)) + * $(TR $(TD $(PLUSMN)0.0) $(TD $(PLUSMN)0.0) $(TD no)) + * $(TR $(TD $(PLUSMN)$(INFIN)) $(TD $(PLUSMN)$(INFIN)) $(TD no)) + * ) + */ +real sinh(real x) @safe pure nothrow @nogc +{ + // sinh(x) = (exp(x)-exp(-x))/2; + // Very large arguments could cause an overflow, but + // the maximum value of x for which exp(x) + exp(-x)) != exp(x) + // is x = 0.5 * (real.mant_dig) * LN2. // = 22.1807 for real80. + if (fabs(x) > real.mant_dig * LN2) + { + return copysign(0.5 * exp(fabs(x)), x); + } + + const real y = expm1(x); + return 0.5 * y / (y+1) * (y+2); +} + +/// ditto +double sinh(double x) @safe pure nothrow @nogc { return sinh(cast(real) x); } + +/// ditto +float sinh(float x) @safe pure nothrow @nogc { return sinh(cast(real) x); } + +@system unittest +{ + assert(equalsDigit(sinh(1.0), (E - 1.0 / E) / 2, useDigits)); +} + +/*********************************** + * Calculates the hyperbolic tangent of x. + * + * $(TABLE_SV + * $(TR $(TH x) $(TH tanh(x)) $(TH invalid?)) + * $(TR $(TD $(PLUSMN)0.0) $(TD $(PLUSMN)0.0) $(TD no) ) + * $(TR $(TD $(PLUSMN)$(INFIN)) $(TD $(PLUSMN)1.0) $(TD no)) + * ) + */ +real tanh(real x) @safe pure nothrow @nogc +{ + // tanh(x) = (exp(x) - exp(-x))/(exp(x)+exp(-x)) + if (fabs(x) > real.mant_dig * LN2) + { + return copysign(1, x); + } + + const real y = expm1(2*x); + return y / (y + 2); +} + +/// ditto +double tanh(double x) @safe pure nothrow @nogc { return tanh(cast(real) x); } + +/// ditto +float tanh(float x) @safe pure nothrow @nogc { return tanh(cast(real) x); } + +@system unittest +{ + assert(equalsDigit(tanh(1.0), sinh(1.0) / cosh(1.0), 15)); +} + +package: + +/* Returns cosh(x) + I * sinh(x) + * Only one call to exp() is performed. + */ +creal coshisinh(real x) @safe pure nothrow @nogc +{ + // See comments for cosh, sinh. + if (fabs(x) > real.mant_dig * LN2) + { + const real y = exp(fabs(x)); + return y * 0.5 + 0.5i * copysign(y, x); + } + else + { + const real y = expm1(x); + return (y + 1.0 + 1.0/(y + 1.0)) * 0.5 + 0.5i * y / (y+1) * (y+2); + } +} + +@safe pure nothrow @nogc unittest +{ + creal c = coshisinh(3.0L); + assert(c.re == cosh(3.0L)); + assert(c.im == sinh(3.0L)); +} + +public: + +/*********************************** + * Calculates the inverse hyperbolic cosine of x. + * + * Mathematically, acosh(x) = log(x + sqrt( x*x - 1)) + * + * $(TABLE_DOMRG + * $(DOMAIN 1..$(INFIN)), + * $(RANGE 0..$(INFIN)) + * ) + * + * $(TABLE_SV + * $(SVH x, acosh(x) ) + * $(SV $(NAN), $(NAN) ) + * $(SV $(LT)1, $(NAN) ) + * $(SV 1, 0 ) + * $(SV +$(INFIN),+$(INFIN)) + * ) + */ +real acosh(real x) @safe pure nothrow @nogc +{ + if (x > 1/real.epsilon) + return LN2 + log(x); + else + return log(x + sqrt(x*x - 1)); +} + +/// ditto +double acosh(double x) @safe pure nothrow @nogc { return acosh(cast(real) x); } + +/// ditto +float acosh(float x) @safe pure nothrow @nogc { return acosh(cast(real) x); } + + +@system unittest +{ + assert(isNaN(acosh(0.9))); + assert(isNaN(acosh(real.nan))); + assert(acosh(1.0)==0.0); + assert(acosh(real.infinity) == real.infinity); + assert(isNaN(acosh(0.5))); + assert(equalsDigit(acosh(cosh(3.0)), 3, useDigits)); +} + +/*********************************** + * Calculates the inverse hyperbolic sine of x. + * + * Mathematically, + * --------------- + * asinh(x) = log( x + sqrt( x*x + 1 )) // if x >= +0 + * asinh(x) = -log(-x + sqrt( x*x + 1 )) // if x <= -0 + * ------------- + * + * $(TABLE_SV + * $(SVH x, asinh(x) ) + * $(SV $(NAN), $(NAN) ) + * $(SV $(PLUSMN)0, $(PLUSMN)0 ) + * $(SV $(PLUSMN)$(INFIN),$(PLUSMN)$(INFIN)) + * ) + */ +real asinh(real x) @safe pure nothrow @nogc +{ + return (fabs(x) > 1 / real.epsilon) + // beyond this point, x*x + 1 == x*x + ? copysign(LN2 + log(fabs(x)), x) + // sqrt(x*x + 1) == 1 + x * x / ( 1 + sqrt(x*x + 1) ) + : copysign(log1p(fabs(x) + x*x / (1 + sqrt(x*x + 1)) ), x); +} + +/// ditto +double asinh(double x) @safe pure nothrow @nogc { return asinh(cast(real) x); } + +/// ditto +float asinh(float x) @safe pure nothrow @nogc { return asinh(cast(real) x); } + +@system unittest +{ + assert(isIdentical(asinh(0.0), 0.0)); + assert(isIdentical(asinh(-0.0), -0.0)); + assert(asinh(real.infinity) == real.infinity); + assert(asinh(-real.infinity) == -real.infinity); + assert(isNaN(asinh(real.nan))); + assert(equalsDigit(asinh(sinh(3.0)), 3, useDigits)); +} + +/*********************************** + * Calculates the inverse hyperbolic tangent of x, + * returning a value from ranging from -1 to 1. + * + * Mathematically, atanh(x) = log( (1+x)/(1-x) ) / 2 + * + * $(TABLE_DOMRG + * $(DOMAIN -$(INFIN)..$(INFIN)), + * $(RANGE -1 .. 1) + * ) + * $(BR) + * $(TABLE_SV + * $(SVH x, acosh(x) ) + * $(SV $(NAN), $(NAN) ) + * $(SV $(PLUSMN)0, $(PLUSMN)0) + * $(SV -$(INFIN), -0) + * ) + */ +real atanh(real x) @safe pure nothrow @nogc +{ + // log( (1+x)/(1-x) ) == log ( 1 + (2*x)/(1-x) ) + return 0.5 * log1p( 2 * x / (1 - x) ); +} + +/// ditto +double atanh(double x) @safe pure nothrow @nogc { return atanh(cast(real) x); } + +/// ditto +float atanh(float x) @safe pure nothrow @nogc { return atanh(cast(real) x); } + + +@system unittest +{ + assert(isIdentical(atanh(0.0), 0.0)); + assert(isIdentical(atanh(-0.0),-0.0)); + assert(isNaN(atanh(real.nan))); + assert(isNaN(atanh(-real.infinity))); + assert(atanh(0.0) == 0); + assert(equalsDigit(atanh(tanh(0.5L)), 0.5, useDigits)); +} + +/***************************************** + * Returns x rounded to a long value using the current rounding mode. + * If the integer value of x is + * greater than long.max, the result is + * indeterminate. + */ +long rndtol(real x) @nogc @safe pure nothrow { pragma(inline, true); return core.math.rndtol(x); } +//FIXME +///ditto +long rndtol(double x) @safe pure nothrow @nogc { return rndtol(cast(real) x); } +//FIXME +///ditto +long rndtol(float x) @safe pure nothrow @nogc { return rndtol(cast(real) x); } + +@safe unittest +{ + long function(real) prndtol = &rndtol; + assert(prndtol != null); +} + +/***************************************** + * Returns x rounded to a long value using the FE_TONEAREST rounding mode. + * If the integer value of x is + * greater than long.max, the result is + * indeterminate. + */ +extern (C) real rndtonl(real x); + +/*************************************** + * Compute square root of x. + * + * $(TABLE_SV + * $(TR $(TH x) $(TH sqrt(x)) $(TH invalid?)) + * $(TR $(TD -0.0) $(TD -0.0) $(TD no)) + * $(TR $(TD $(LT)0.0) $(TD $(NAN)) $(TD yes)) + * $(TR $(TD +$(INFIN)) $(TD +$(INFIN)) $(TD no)) + * ) + */ +float sqrt(float x) @nogc @safe pure nothrow { pragma(inline, true); return core.math.sqrt(x); } + +/// ditto +double sqrt(double x) @nogc @safe pure nothrow { pragma(inline, true); return core.math.sqrt(x); } + +/// ditto +real sqrt(real x) @nogc @safe pure nothrow { pragma(inline, true); return core.math.sqrt(x); } + +@safe pure nothrow @nogc unittest +{ + //ctfe + enum ZX80 = sqrt(7.0f); + enum ZX81 = sqrt(7.0); + enum ZX82 = sqrt(7.0L); + + assert(isNaN(sqrt(-1.0f))); + assert(isNaN(sqrt(-1.0))); + assert(isNaN(sqrt(-1.0L))); +} + +@safe unittest +{ + float function(float) psqrtf = &sqrt; + assert(psqrtf != null); + double function(double) psqrtd = &sqrt; + assert(psqrtd != null); + real function(real) psqrtr = &sqrt; + assert(psqrtr != null); +} + +creal sqrt(creal z) @nogc @safe pure nothrow +{ + creal c; + real x,y,w,r; + + if (z == 0) + { + c = 0 + 0i; + } + else + { + const real z_re = z.re; + const real z_im = z.im; + + x = fabs(z_re); + y = fabs(z_im); + if (x >= y) + { + r = y / x; + w = sqrt(x) * sqrt(0.5 * (1 + sqrt(1 + r * r))); + } + else + { + r = x / y; + w = sqrt(y) * sqrt(0.5 * (r + sqrt(1 + r * r))); + } + + if (z_re >= 0) + { + c = w + (z_im / (w + w)) * 1.0i; + } + else + { + if (z_im < 0) + w = -w; + c = z_im / (w + w) + w * 1.0i; + } + } + return c; +} + +/** + * Calculates e$(SUPERSCRIPT x). + * + * $(TABLE_SV + * $(TR $(TH x) $(TH e$(SUPERSCRIPT x)) ) + * $(TR $(TD +$(INFIN)) $(TD +$(INFIN)) ) + * $(TR $(TD -$(INFIN)) $(TD +0.0) ) + * $(TR $(TD $(NAN)) $(TD $(NAN)) ) + * ) + */ +real exp(real x) @trusted pure nothrow @nogc +{ + version (D_InlineAsm_X86) + { + // e^^x = 2^^(LOG2E*x) + // (This is valid because the overflow & underflow limits for exp + // and exp2 are so similar). + return exp2(LOG2E*x); + } + else version (D_InlineAsm_X86_64) + { + // e^^x = 2^^(LOG2E*x) + // (This is valid because the overflow & underflow limits for exp + // and exp2 are so similar). + return exp2(LOG2E*x); + } + else + { + alias F = floatTraits!real; + static if (F.realFormat == RealFormat.ieeeDouble) + { + // Coefficients for exp(x) + static immutable real[3] P = [ + 9.99999999999999999910E-1L, + 3.02994407707441961300E-2L, + 1.26177193074810590878E-4L, + ]; + static immutable real[4] Q = [ + 2.00000000000000000009E0L, + 2.27265548208155028766E-1L, + 2.52448340349684104192E-3L, + 3.00198505138664455042E-6L, + ]; + + // C1 + C2 = LN2. + enum real C1 = 6.93145751953125E-1; + enum real C2 = 1.42860682030941723212E-6; + + // Overflow and Underflow limits. + enum real OF = 7.09782712893383996732E2; // ln((1-2^-53) * 2^1024) + enum real UF = -7.451332191019412076235E2; // ln(2^-1075) + } + else static if (F.realFormat == RealFormat.ieeeExtended) + { + // Coefficients for exp(x) + static immutable real[3] P = [ + 9.9999999999999999991025E-1L, + 3.0299440770744196129956E-2L, + 1.2617719307481059087798E-4L, + ]; + static immutable real[4] Q = [ + 2.0000000000000000000897E0L, + 2.2726554820815502876593E-1L, + 2.5244834034968410419224E-3L, + 3.0019850513866445504159E-6L, + ]; + + // C1 + C2 = LN2. + enum real C1 = 6.9314575195312500000000E-1L; + enum real C2 = 1.4286068203094172321215E-6L; + + // Overflow and Underflow limits. + enum real OF = 1.1356523406294143949492E4L; // ln((1-2^-64) * 2^16384) + enum real UF = -1.13994985314888605586758E4L; // ln(2^-16446) + } + else static if (F.realFormat == RealFormat.ieeeQuadruple) + { + // Coefficients for exp(x) - 1 + static immutable real[5] P = [ + 9.999999999999999999999999999999999998502E-1L, + 3.508710990737834361215404761139478627390E-2L, + 2.708775201978218837374512615596512792224E-4L, + 6.141506007208645008909088812338454698548E-7L, + 3.279723985560247033712687707263393506266E-10L + ]; + static immutable real[6] Q = [ + 2.000000000000000000000000000000000000150E0, + 2.368408864814233538909747618894558968880E-1L, + 3.611828913847589925056132680618007270344E-3L, + 1.504792651814944826817779302637284053660E-5L, + 1.771372078166251484503904874657985291164E-8L, + 2.980756652081995192255342779918052538681E-12L + ]; + + // C1 + C2 = LN2. + enum real C1 = 6.93145751953125E-1L; + enum real C2 = 1.428606820309417232121458176568075500134E-6L; + + // Overflow and Underflow limits. + enum real OF = 1.135583025911358400418251384584930671458833e4L; + enum real UF = -1.143276959615573793352782661133116431383730e4L; + } + else + static assert(0, "Not implemented for this architecture"); + + // Special cases. Raises an overflow or underflow flag accordingly, + // except in the case for CTFE, where there are no hardware controls. + if (isNaN(x)) + return x; + if (x > OF) + { + if (__ctfe) + return real.infinity; + else + return real.max * copysign(real.max, real.infinity); + } + if (x < UF) + { + if (__ctfe) + return 0.0; + else + return real.min_normal * copysign(real.min_normal, 0.0); + } + + // Express: e^^x = e^^g * 2^^n + // = e^^g * e^^(n * LOG2E) + // = e^^(g + n * LOG2E) + int n = cast(int) floor(LOG2E * x + 0.5); + x -= n * C1; + x -= n * C2; + + // Rational approximation for exponential of the fractional part: + // e^^x = 1 + 2x P(x^^2) / (Q(x^^2) - P(x^^2)) + const real xx = x * x; + const real px = x * poly(xx, P); + x = px / (poly(xx, Q) - px); + x = 1.0 + ldexp(x, 1); + + // Scale by power of 2. + x = ldexp(x, n); + + return x; + } +} + +/// ditto +double exp(double x) @safe pure nothrow @nogc { return exp(cast(real) x); } + +/// ditto +float exp(float x) @safe pure nothrow @nogc { return exp(cast(real) x); } + +@system unittest +{ + assert(equalsDigit(exp(3.0L), E * E * E, useDigits)); +} + +/** + * Calculates the value of the natural logarithm base (e) + * raised to the power of x, minus 1. + * + * For very small x, expm1(x) is more accurate + * than exp(x)-1. + * + * $(TABLE_SV + * $(TR $(TH x) $(TH e$(SUPERSCRIPT x)-1) ) + * $(TR $(TD $(PLUSMN)0.0) $(TD $(PLUSMN)0.0) ) + * $(TR $(TD +$(INFIN)) $(TD +$(INFIN)) ) + * $(TR $(TD -$(INFIN)) $(TD -1.0) ) + * $(TR $(TD $(NAN)) $(TD $(NAN)) ) + * ) + */ +real expm1(real x) @trusted pure nothrow @nogc +{ + version (D_InlineAsm_X86) + { + enum PARAMSIZE = (real.sizeof+3)&(0xFFFF_FFFC); // always a multiple of 4 + asm pure nothrow @nogc + { + /* expm1() for x87 80-bit reals, IEEE754-2008 conformant. + * Author: Don Clugston. + * + * expm1(x) = 2^^(rndint(y))* 2^^(y-rndint(y)) - 1 where y = LN2*x. + * = 2rndy * 2ym1 + 2rndy - 1, where 2rndy = 2^^(rndint(y)) + * and 2ym1 = (2^^(y-rndint(y))-1). + * If 2rndy < 0.5*real.epsilon, result is -1. + * Implementation is otherwise the same as for exp2() + */ + naked; + fld real ptr [ESP+4] ; // x + mov AX, [ESP+4+8]; // AX = exponent and sign + sub ESP, 12+8; // Create scratch space on the stack + // [ESP,ESP+2] = scratchint + // [ESP+4..+6, +8..+10, +10] = scratchreal + // set scratchreal mantissa = 1.0 + mov dword ptr [ESP+8], 0; + mov dword ptr [ESP+8+4], 0x80000000; + and AX, 0x7FFF; // drop sign bit + cmp AX, 0x401D; // avoid InvalidException in fist + jae L_extreme; + fldl2e; + fmulp ST(1), ST; // y = x*log2(e) + fist dword ptr [ESP]; // scratchint = rndint(y) + fisub dword ptr [ESP]; // y - rndint(y) + // and now set scratchreal exponent + mov EAX, [ESP]; + add EAX, 0x3fff; + jle short L_largenegative; + cmp EAX,0x8000; + jge short L_largepositive; + mov [ESP+8+8],AX; + f2xm1; // 2ym1 = 2^^(y-rndint(y)) -1 + fld real ptr [ESP+8] ; // 2rndy = 2^^rndint(y) + fmul ST(1), ST; // ST=2rndy, ST(1)=2rndy*2ym1 + fld1; + fsubp ST(1), ST; // ST = 2rndy-1, ST(1) = 2rndy * 2ym1 - 1 + faddp ST(1), ST; // ST = 2rndy * 2ym1 + 2rndy - 1 + add ESP,12+8; + ret PARAMSIZE; + +L_extreme: // Extreme exponent. X is very large positive, very + // large negative, infinity, or NaN. + fxam; + fstsw AX; + test AX, 0x0400; // NaN_or_zero, but we already know x != 0 + jz L_was_nan; // if x is NaN, returns x + test AX, 0x0200; + jnz L_largenegative; +L_largepositive: + // Set scratchreal = real.max. + // squaring it will create infinity, and set overflow flag. + mov word ptr [ESP+8+8], 0x7FFE; + fstp ST(0); + fld real ptr [ESP+8]; // load scratchreal + fmul ST(0), ST; // square it, to create havoc! +L_was_nan: + add ESP,12+8; + ret PARAMSIZE; +L_largenegative: + fstp ST(0); + fld1; + fchs; // return -1. Underflow flag is not set. + add ESP,12+8; + ret PARAMSIZE; + } + } + else version (D_InlineAsm_X86_64) + { + asm pure nothrow @nogc + { + naked; + } + version (Win64) + { + asm pure nothrow @nogc + { + fld real ptr [RCX]; // x + mov AX,[RCX+8]; // AX = exponent and sign + } + } + else + { + asm pure nothrow @nogc + { + fld real ptr [RSP+8]; // x + mov AX,[RSP+8+8]; // AX = exponent and sign + } + } + asm pure nothrow @nogc + { + /* expm1() for x87 80-bit reals, IEEE754-2008 conformant. + * Author: Don Clugston. + * + * expm1(x) = 2^(rndint(y))* 2^(y-rndint(y)) - 1 where y = LN2*x. + * = 2rndy * 2ym1 + 2rndy - 1, where 2rndy = 2^(rndint(y)) + * and 2ym1 = (2^(y-rndint(y))-1). + * If 2rndy < 0.5*real.epsilon, result is -1. + * Implementation is otherwise the same as for exp2() + */ + sub RSP, 24; // Create scratch space on the stack + // [RSP,RSP+2] = scratchint + // [RSP+4..+6, +8..+10, +10] = scratchreal + // set scratchreal mantissa = 1.0 + mov dword ptr [RSP+8], 0; + mov dword ptr [RSP+8+4], 0x80000000; + and AX, 0x7FFF; // drop sign bit + cmp AX, 0x401D; // avoid InvalidException in fist + jae L_extreme; + fldl2e; + fmul ; // y = x*log2(e) + fist dword ptr [RSP]; // scratchint = rndint(y) + fisub dword ptr [RSP]; // y - rndint(y) + // and now set scratchreal exponent + mov EAX, [RSP]; + add EAX, 0x3fff; + jle short L_largenegative; + cmp EAX,0x8000; + jge short L_largepositive; + mov [RSP+8+8],AX; + f2xm1; // 2^(y-rndint(y)) -1 + fld real ptr [RSP+8] ; // 2^rndint(y) + fmul ST(1), ST; + fld1; + fsubp ST(1), ST; + fadd; + add RSP,24; + ret; + +L_extreme: // Extreme exponent. X is very large positive, very + // large negative, infinity, or NaN. + fxam; + fstsw AX; + test AX, 0x0400; // NaN_or_zero, but we already know x != 0 + jz L_was_nan; // if x is NaN, returns x + test AX, 0x0200; + jnz L_largenegative; +L_largepositive: + // Set scratchreal = real.max. + // squaring it will create infinity, and set overflow flag. + mov word ptr [RSP+8+8], 0x7FFE; + fstp ST(0); + fld real ptr [RSP+8]; // load scratchreal + fmul ST(0), ST; // square it, to create havoc! +L_was_nan: + add RSP,24; + ret; + +L_largenegative: + fstp ST(0); + fld1; + fchs; // return -1. Underflow flag is not set. + add RSP,24; + ret; + } + } + else + { + // Coefficients for exp(x) - 1 and overflow/underflow limits. + static if (floatTraits!real.realFormat == RealFormat.ieeeQuadruple) + { + static immutable real[8] P = [ + 2.943520915569954073888921213330863757240E8L, + -5.722847283900608941516165725053359168840E7L, + 8.944630806357575461578107295909719817253E6L, + -7.212432713558031519943281748462837065308E5L, + 4.578962475841642634225390068461943438441E4L, + -1.716772506388927649032068540558788106762E3L, + 4.401308817383362136048032038528753151144E1L, + -4.888737542888633647784737721812546636240E-1L + ]; + + static immutable real[9] Q = [ + 1.766112549341972444333352727998584753865E9L, + -7.848989743695296475743081255027098295771E8L, + 1.615869009634292424463780387327037251069E8L, + -2.019684072836541751428967854947019415698E7L, + 1.682912729190313538934190635536631941751E6L, + -9.615511549171441430850103489315371768998E4L, + 3.697714952261803935521187272204485251835E3L, + -8.802340681794263968892934703309274564037E1L, + 1.0 + ]; + + enum real OF = 1.1356523406294143949491931077970764891253E4L; + enum real UF = -1.143276959615573793352782661133116431383730e4L; + } + else + { + static immutable real[5] P = [ + -1.586135578666346600772998894928250240826E4L, + 2.642771505685952966904660652518429479531E3L, + -3.423199068835684263987132888286791620673E2L, + 1.800826371455042224581246202420972737840E1L, + -5.238523121205561042771939008061958820811E-1L, + ]; + static immutable real[6] Q = [ + -9.516813471998079611319047060563358064497E4L, + 3.964866271411091674556850458227710004570E4L, + -7.207678383830091850230366618190187434796E3L, + 7.206038318724600171970199625081491823079E2L, + -4.002027679107076077238836622982900945173E1L, + 1.0 + ]; + + enum real OF = 1.1356523406294143949492E4L; + enum real UF = -4.5054566736396445112120088E1L; + } + + + // C1 + C2 = LN2. + enum real C1 = 6.9314575195312500000000E-1L; + enum real C2 = 1.428606820309417232121458176568075500134E-6L; + + // Special cases. Raises an overflow flag, except in the case + // for CTFE, where there are no hardware controls. + if (x > OF) + { + if (__ctfe) + return real.infinity; + else + return real.max * copysign(real.max, real.infinity); + } + if (x == 0.0) + return x; + if (x < UF) + return -1.0; + + // Express x = LN2 (n + remainder), remainder not exceeding 1/2. + int n = cast(int) floor(0.5 + x / LN2); + x -= n * C1; + x -= n * C2; + + // Rational approximation: + // exp(x) - 1 = x + 0.5 x^^2 + x^^3 P(x) / Q(x) + real px = x * poly(x, P); + real qx = poly(x, Q); + const real xx = x * x; + qx = x + (0.5 * xx + xx * px / qx); + + // We have qx = exp(remainder LN2) - 1, so: + // exp(x) - 1 = 2^^n (qx + 1) - 1 = 2^^n qx + 2^^n - 1. + px = ldexp(1.0, n); + x = px * qx + (px - 1.0); + + return x; + } +} + + + +/** + * Calculates 2$(SUPERSCRIPT x). + * + * $(TABLE_SV + * $(TR $(TH x) $(TH exp2(x)) ) + * $(TR $(TD +$(INFIN)) $(TD +$(INFIN)) ) + * $(TR $(TD -$(INFIN)) $(TD +0.0) ) + * $(TR $(TD $(NAN)) $(TD $(NAN)) ) + * ) + */ +pragma(inline, true) +real exp2(real x) @nogc @trusted pure nothrow +{ + version (InlineAsm_X86_Any) + { + if (!__ctfe) + return exp2Asm(x); + else + return exp2Impl(x); + } + else + { + return exp2Impl(x); + } +} + +version (InlineAsm_X86_Any) +private real exp2Asm(real x) @nogc @trusted pure nothrow +{ + version (D_InlineAsm_X86) + { + enum PARAMSIZE = (real.sizeof+3)&(0xFFFF_FFFC); // always a multiple of 4 + + asm pure nothrow @nogc + { + /* exp2() for x87 80-bit reals, IEEE754-2008 conformant. + * Author: Don Clugston. + * + * exp2(x) = 2^^(rndint(x))* 2^^(y-rndint(x)) + * The trick for high performance is to avoid the fscale(28cycles on core2), + * frndint(19 cycles), leaving f2xm1(19 cycles) as the only slow instruction. + * + * We can do frndint by using fist. BUT we can't use it for huge numbers, + * because it will set the Invalid Operation flag if overflow or NaN occurs. + * Fortunately, whenever this happens the result would be zero or infinity. + * + * We can perform fscale by directly poking into the exponent. BUT this doesn't + * work for the (very rare) cases where the result is subnormal. So we fall back + * to the slow method in that case. + */ + naked; + fld real ptr [ESP+4] ; // x + mov AX, [ESP+4+8]; // AX = exponent and sign + sub ESP, 12+8; // Create scratch space on the stack + // [ESP,ESP+2] = scratchint + // [ESP+4..+6, +8..+10, +10] = scratchreal + // set scratchreal mantissa = 1.0 + mov dword ptr [ESP+8], 0; + mov dword ptr [ESP+8+4], 0x80000000; + and AX, 0x7FFF; // drop sign bit + cmp AX, 0x401D; // avoid InvalidException in fist + jae L_extreme; + fist dword ptr [ESP]; // scratchint = rndint(x) + fisub dword ptr [ESP]; // x - rndint(x) + // and now set scratchreal exponent + mov EAX, [ESP]; + add EAX, 0x3fff; + jle short L_subnormal; + cmp EAX,0x8000; + jge short L_overflow; + mov [ESP+8+8],AX; +L_normal: + f2xm1; + fld1; + faddp ST(1), ST; // 2^^(x-rndint(x)) + fld real ptr [ESP+8] ; // 2^^rndint(x) + add ESP,12+8; + fmulp ST(1), ST; + ret PARAMSIZE; + +L_subnormal: + // Result will be subnormal. + // In this rare case, the simple poking method doesn't work. + // The speed doesn't matter, so use the slow fscale method. + fild dword ptr [ESP]; // scratchint + fld1; + fscale; + fstp real ptr [ESP+8]; // scratchreal = 2^^scratchint + fstp ST(0); // drop scratchint + jmp L_normal; + +L_extreme: // Extreme exponent. X is very large positive, very + // large negative, infinity, or NaN. + fxam; + fstsw AX; + test AX, 0x0400; // NaN_or_zero, but we already know x != 0 + jz L_was_nan; // if x is NaN, returns x + // set scratchreal = real.min_normal + // squaring it will return 0, setting underflow flag + mov word ptr [ESP+8+8], 1; + test AX, 0x0200; + jnz L_waslargenegative; +L_overflow: + // Set scratchreal = real.max. + // squaring it will create infinity, and set overflow flag. + mov word ptr [ESP+8+8], 0x7FFE; +L_waslargenegative: + fstp ST(0); + fld real ptr [ESP+8]; // load scratchreal + fmul ST(0), ST; // square it, to create havoc! +L_was_nan: + add ESP,12+8; + ret PARAMSIZE; + } + } + else version (D_InlineAsm_X86_64) + { + asm pure nothrow @nogc + { + naked; + } + version (Win64) + { + asm pure nothrow @nogc + { + fld real ptr [RCX]; // x + mov AX,[RCX+8]; // AX = exponent and sign + } + } + else + { + asm pure nothrow @nogc + { + fld real ptr [RSP+8]; // x + mov AX,[RSP+8+8]; // AX = exponent and sign + } + } + asm pure nothrow @nogc + { + /* exp2() for x87 80-bit reals, IEEE754-2008 conformant. + * Author: Don Clugston. + * + * exp2(x) = 2^(rndint(x))* 2^(y-rndint(x)) + * The trick for high performance is to avoid the fscale(28cycles on core2), + * frndint(19 cycles), leaving f2xm1(19 cycles) as the only slow instruction. + * + * We can do frndint by using fist. BUT we can't use it for huge numbers, + * because it will set the Invalid Operation flag is overflow or NaN occurs. + * Fortunately, whenever this happens the result would be zero or infinity. + * + * We can perform fscale by directly poking into the exponent. BUT this doesn't + * work for the (very rare) cases where the result is subnormal. So we fall back + * to the slow method in that case. + */ + sub RSP, 24; // Create scratch space on the stack + // [RSP,RSP+2] = scratchint + // [RSP+4..+6, +8..+10, +10] = scratchreal + // set scratchreal mantissa = 1.0 + mov dword ptr [RSP+8], 0; + mov dword ptr [RSP+8+4], 0x80000000; + and AX, 0x7FFF; // drop sign bit + cmp AX, 0x401D; // avoid InvalidException in fist + jae L_extreme; + fist dword ptr [RSP]; // scratchint = rndint(x) + fisub dword ptr [RSP]; // x - rndint(x) + // and now set scratchreal exponent + mov EAX, [RSP]; + add EAX, 0x3fff; + jle short L_subnormal; + cmp EAX,0x8000; + jge short L_overflow; + mov [RSP+8+8],AX; +L_normal: + f2xm1; + fld1; + fadd; // 2^(x-rndint(x)) + fld real ptr [RSP+8] ; // 2^rndint(x) + add RSP,24; + fmulp ST(1), ST; + ret; + +L_subnormal: + // Result will be subnormal. + // In this rare case, the simple poking method doesn't work. + // The speed doesn't matter, so use the slow fscale method. + fild dword ptr [RSP]; // scratchint + fld1; + fscale; + fstp real ptr [RSP+8]; // scratchreal = 2^scratchint + fstp ST(0); // drop scratchint + jmp L_normal; + +L_extreme: // Extreme exponent. X is very large positive, very + // large negative, infinity, or NaN. + fxam; + fstsw AX; + test AX, 0x0400; // NaN_or_zero, but we already know x != 0 + jz L_was_nan; // if x is NaN, returns x + // set scratchreal = real.min + // squaring it will return 0, setting underflow flag + mov word ptr [RSP+8+8], 1; + test AX, 0x0200; + jnz L_waslargenegative; +L_overflow: + // Set scratchreal = real.max. + // squaring it will create infinity, and set overflow flag. + mov word ptr [RSP+8+8], 0x7FFE; +L_waslargenegative: + fstp ST(0); + fld real ptr [RSP+8]; // load scratchreal + fmul ST(0), ST; // square it, to create havoc! +L_was_nan: + add RSP,24; + ret; + } + } + else + static assert(0); +} + +private real exp2Impl(real x) @nogc @trusted pure nothrow +{ + // Coefficients for exp2(x) + static if (floatTraits!real.realFormat == RealFormat.ieeeQuadruple) + { + static immutable real[5] P = [ + 9.079594442980146270952372234833529694788E12L, + 1.530625323728429161131811299626419117557E11L, + 5.677513871931844661829755443994214173883E8L, + 6.185032670011643762127954396427045467506E5L, + 1.587171580015525194694938306936721666031E2L + ]; + + static immutable real[6] Q = [ + 2.619817175234089411411070339065679229869E13L, + 1.490560994263653042761789432690793026977E12L, + 1.092141473886177435056423606755843616331E10L, + 2.186249607051644894762167991800811827835E7L, + 1.236602014442099053716561665053645270207E4L, + 1.0 + ]; + } + else + { + static immutable real[3] P = [ + 2.0803843631901852422887E6L, + 3.0286971917562792508623E4L, + 6.0614853552242266094567E1L, + ]; + static immutable real[4] Q = [ + 6.0027204078348487957118E6L, + 3.2772515434906797273099E5L, + 1.7492876999891839021063E3L, + 1.0000000000000000000000E0L, + ]; + } + + // Overflow and Underflow limits. + enum real OF = 16_384.0L; + enum real UF = -16_382.0L; + + // Special cases. Raises an overflow or underflow flag accordingly, + // except in the case for CTFE, where there are no hardware controls. + if (isNaN(x)) + return x; + if (x > OF) + { + if (__ctfe) + return real.infinity; + else + return real.max * copysign(real.max, real.infinity); + } + if (x < UF) + { + if (__ctfe) + return 0.0; + else + return real.min_normal * copysign(real.min_normal, 0.0); + } + + // Separate into integer and fractional parts. + int n = cast(int) floor(x + 0.5); + x -= n; + + // Rational approximation: + // exp2(x) = 1.0 + 2x P(x^^2) / (Q(x^^2) - P(x^^2)) + const real xx = x * x; + const real px = x * poly(xx, P); + x = px / (poly(xx, Q) - px); + x = 1.0 + ldexp(x, 1); + + // Scale by power of 2. + x = ldexp(x, n); + + return x; +} + +/// +@safe unittest +{ + assert(feqrel(exp2(0.5L), SQRT2) >= real.mant_dig -1); + assert(exp2(8.0L) == 256.0); + assert(exp2(-9.0L)== 1.0L/512.0); +} + +@safe unittest +{ + version (CRuntime_Microsoft) {} else // aexp2/exp2f/exp2l not implemented + { + assert( core.stdc.math.exp2f(0.0f) == 1 ); + assert( core.stdc.math.exp2 (0.0) == 1 ); + assert( core.stdc.math.exp2l(0.0L) == 1 ); + } +} + +@system unittest +{ + FloatingPointControl ctrl; + if (FloatingPointControl.hasExceptionTraps) + ctrl.disableExceptions(FloatingPointControl.allExceptions); + ctrl.rounding = FloatingPointControl.roundToNearest; + + static if (real.mant_dig == 113) + { + static immutable real[2][] exptestpoints = + [ // x exp(x) + [ 1.0L, E ], + [ 0.5L, 0x1.a61298e1e069bc972dfefab6df34p+0L ], + [ 3.0L, E*E*E ], + [ 0x1.6p+13L, 0x1.6e509d45728655cdb4840542acb5p+16250L ], // near overflow + [ 0x1.7p+13L, real.infinity ], // close overflow + [ 0x1p+80L, real.infinity ], // far overflow + [ real.infinity, real.infinity ], + [-0x1.18p+13L, 0x1.5e4bf54b4807034ea97fef0059a6p-12927L ], // near underflow + [-0x1.625p+13L, 0x1.a6bd68a39d11fec3a250cd97f524p-16358L ], // ditto + [-0x1.62dafp+13L, 0x0.cb629e9813b80ed4d639e875be6cp-16382L ], // near underflow - subnormal + [-0x1.6549p+13L, 0x0.0000000000000000000000000001p-16382L ], // ditto + [-0x1.655p+13L, 0 ], // close underflow + [-0x1p+30L, 0 ], // far underflow + ]; + } + else static if (real.mant_dig == 64) // 80-bit reals + { + static immutable real[2][] exptestpoints = + [ // x exp(x) + [ 1.0L, E ], + [ 0.5L, 0x1.a61298e1e069bc97p+0L ], + [ 3.0L, E*E*E ], + [ 0x1.1p+13L, 0x1.29aeffefc8ec645p+12557L ], // near overflow + [ 0x1.7p+13L, real.infinity ], // close overflow + [ 0x1p+80L, real.infinity ], // far overflow + [ real.infinity, real.infinity ], + [-0x1.18p+13L, 0x1.5e4bf54b4806db9p-12927L ], // near underflow + [-0x1.625p+13L, 0x1.a6bd68a39d11f35cp-16358L ], // ditto + [-0x1.62dafp+13L, 0x1.96c53d30277021dp-16383L ], // near underflow - subnormal + [-0x1.643p+13L, 0x1p-16444L ], // ditto + [-0x1.645p+13L, 0 ], // close underflow + [-0x1p+30L, 0 ], // far underflow + ]; + } + else static if (real.mant_dig == 53) // 64-bit reals + { + static immutable real[2][] exptestpoints = + [ // x, exp(x) + [ 1.0L, E ], + [ 0.5L, 0x1.a61298e1e069cp+0L ], + [ 3.0L, E*E*E ], + [ 0x1.6p+9L, 0x1.93bf4ec282efbp+1015L ], // near overflow + [ 0x1.7p+9L, real.infinity ], // close overflow + [ 0x1p+80L, real.infinity ], // far overflow + [ real.infinity, real.infinity ], + [-0x1.6p+9L, 0x1.44a3824e5285fp-1016L ], // near underflow + [-0x1.64p+9L, 0x0.06f84920bb2d3p-1022L ], // near underflow - subnormal + [-0x1.743p+9L, 0x0.0000000000001p-1022L ], // ditto + [-0x1.8p+9L, 0 ], // close underflow + [-0x1p30L, 0 ], // far underflow + ]; + } + else + static assert(0, "No exp() tests for real type!"); + + const minEqualDecimalDigits = real.dig - 3; + real x; + IeeeFlags f; + foreach (ref pair; exptestpoints) + { + resetIeeeFlags(); + x = exp(pair[0]); + f = ieeeFlags; + assert(equalsDigit(x, pair[1], minEqualDecimalDigits)); + + version (IeeeFlagsSupport) + { + // Check the overflow bit + if (x == real.infinity) + { + // don't care about the overflow bit if input was inf + // (e.g., the LLVM intrinsic doesn't set it on Linux x86_64) + assert(pair[0] == real.infinity || f.overflow); + } + else + assert(!f.overflow); + // Check the underflow bit + assert(f.underflow == (fabs(x) < real.min_normal)); + // Invalid and div by zero shouldn't be affected. + assert(!f.invalid); + assert(!f.divByZero); + } + } + // Ideally, exp(0) would not set the inexact flag. + // Unfortunately, fldl2e sets it! + // So it's not realistic to avoid setting it. + assert(exp(0.0L) == 1.0); + + // NaN propagation. Doesn't set flags, bcos was already NaN. + resetIeeeFlags(); + x = exp(real.nan); + f = ieeeFlags; + assert(isIdentical(abs(x), real.nan)); + assert(f.flags == 0); + + resetIeeeFlags(); + x = exp(-real.nan); + f = ieeeFlags; + assert(isIdentical(abs(x), real.nan)); + assert(f.flags == 0); + + x = exp(NaN(0x123)); + assert(isIdentical(x, NaN(0x123))); + + // High resolution test (verified against GNU MPFR/Mathematica). + assert(exp(0.5L) == 0x1.A612_98E1_E069_BC97_2DFE_FAB6_DF34p+0L); +} + + +/** + * Calculate cos(y) + i sin(y). + * + * On many CPUs (such as x86), this is a very efficient operation; + * almost twice as fast as calculating sin(y) and cos(y) separately, + * and is the preferred method when both are required. + */ +creal expi(real y) @trusted pure nothrow @nogc +{ + version (InlineAsm_X86_Any) + { + version (Win64) + { + asm pure nothrow @nogc + { + naked; + fld real ptr [ECX]; + fsincos; + fxch ST(1), ST(0); + ret; + } + } + else + { + asm pure nothrow @nogc + { + fld y; + fsincos; + fxch ST(1), ST(0); + } + } + } + else + { + return cos(y) + sin(y)*1i; + } +} + +/// +@safe pure nothrow @nogc unittest +{ + assert(expi(1.3e5L) == cos(1.3e5L) + sin(1.3e5L) * 1i); + assert(expi(0.0L) == 1L + 0.0Li); +} + +/********************************************************************* + * Separate floating point value into significand and exponent. + * + * Returns: + * Calculate and return $(I x) and $(I exp) such that + * value =$(I x)*2$(SUPERSCRIPT exp) and + * .5 $(LT)= |$(I x)| $(LT) 1.0 + * + * $(I x) has same sign as value. + * + * $(TABLE_SV + * $(TR $(TH value) $(TH returns) $(TH exp)) + * $(TR $(TD $(PLUSMN)0.0) $(TD $(PLUSMN)0.0) $(TD 0)) + * $(TR $(TD +$(INFIN)) $(TD +$(INFIN)) $(TD int.max)) + * $(TR $(TD -$(INFIN)) $(TD -$(INFIN)) $(TD int.min)) + * $(TR $(TD $(PLUSMN)$(NAN)) $(TD $(PLUSMN)$(NAN)) $(TD int.min)) + * ) + */ +T frexp(T)(const T value, out int exp) @trusted pure nothrow @nogc +if (isFloatingPoint!T) +{ + Unqual!T vf = value; + ushort* vu = cast(ushort*)&vf; + static if (is(Unqual!T == float)) + int* vi = cast(int*)&vf; + else + long* vl = cast(long*)&vf; + int ex; + alias F = floatTraits!T; + + ex = vu[F.EXPPOS_SHORT] & F.EXPMASK; + static if (F.realFormat == RealFormat.ieeeExtended) + { + if (ex) + { // If exponent is non-zero + if (ex == F.EXPMASK) // infinity or NaN + { + if (*vl & 0x7FFF_FFFF_FFFF_FFFF) // NaN + { + *vl |= 0xC000_0000_0000_0000; // convert NaNS to NaNQ + exp = int.min; + } + else if (vu[F.EXPPOS_SHORT] & 0x8000) // negative infinity + exp = int.min; + else // positive infinity + exp = int.max; + + } + else + { + exp = ex - F.EXPBIAS; + vu[F.EXPPOS_SHORT] = (0x8000 & vu[F.EXPPOS_SHORT]) | 0x3FFE; + } + } + else if (!*vl) + { + // vf is +-0.0 + exp = 0; + } + else + { + // subnormal + + vf *= F.RECIP_EPSILON; + ex = vu[F.EXPPOS_SHORT] & F.EXPMASK; + exp = ex - F.EXPBIAS - T.mant_dig + 1; + vu[F.EXPPOS_SHORT] = ((-1 - F.EXPMASK) & vu[F.EXPPOS_SHORT]) | 0x3FFE; + } + return vf; + } + else static if (F.realFormat == RealFormat.ieeeQuadruple) + { + if (ex) // If exponent is non-zero + { + if (ex == F.EXPMASK) + { + // infinity or NaN + if (vl[MANTISSA_LSB] | + (vl[MANTISSA_MSB] & 0x0000_FFFF_FFFF_FFFF)) // NaN + { + // convert NaNS to NaNQ + vl[MANTISSA_MSB] |= 0x0000_8000_0000_0000; + exp = int.min; + } + else if (vu[F.EXPPOS_SHORT] & 0x8000) // negative infinity + exp = int.min; + else // positive infinity + exp = int.max; + } + else + { + exp = ex - F.EXPBIAS; + vu[F.EXPPOS_SHORT] = F.EXPBIAS | (0x8000 & vu[F.EXPPOS_SHORT]); + } + } + else if ((vl[MANTISSA_LSB] | + (vl[MANTISSA_MSB] & 0x0000_FFFF_FFFF_FFFF)) == 0) + { + // vf is +-0.0 + exp = 0; + } + else + { + // subnormal + vf *= F.RECIP_EPSILON; + ex = vu[F.EXPPOS_SHORT] & F.EXPMASK; + exp = ex - F.EXPBIAS - T.mant_dig + 1; + vu[F.EXPPOS_SHORT] = F.EXPBIAS | (0x8000 & vu[F.EXPPOS_SHORT]); + } + return vf; + } + else static if (F.realFormat == RealFormat.ieeeDouble) + { + if (ex) // If exponent is non-zero + { + if (ex == F.EXPMASK) // infinity or NaN + { + if (*vl == 0x7FF0_0000_0000_0000) // positive infinity + { + exp = int.max; + } + else if (*vl == 0xFFF0_0000_0000_0000) // negative infinity + exp = int.min; + else + { // NaN + *vl |= 0x0008_0000_0000_0000; // convert NaNS to NaNQ + exp = int.min; + } + } + else + { + exp = (ex - F.EXPBIAS) >> 4; + vu[F.EXPPOS_SHORT] = cast(ushort)((0x800F & vu[F.EXPPOS_SHORT]) | 0x3FE0); + } + } + else if (!(*vl & 0x7FFF_FFFF_FFFF_FFFF)) + { + // vf is +-0.0 + exp = 0; + } + else + { + // subnormal + vf *= F.RECIP_EPSILON; + ex = vu[F.EXPPOS_SHORT] & F.EXPMASK; + exp = ((ex - F.EXPBIAS) >> 4) - T.mant_dig + 1; + vu[F.EXPPOS_SHORT] = + cast(ushort)(((-1 - F.EXPMASK) & vu[F.EXPPOS_SHORT]) | 0x3FE0); + } + return vf; + } + else static if (F.realFormat == RealFormat.ieeeSingle) + { + if (ex) // If exponent is non-zero + { + if (ex == F.EXPMASK) // infinity or NaN + { + if (*vi == 0x7F80_0000) // positive infinity + { + exp = int.max; + } + else if (*vi == 0xFF80_0000) // negative infinity + exp = int.min; + else + { // NaN + *vi |= 0x0040_0000; // convert NaNS to NaNQ + exp = int.min; + } + } + else + { + exp = (ex - F.EXPBIAS) >> 7; + vu[F.EXPPOS_SHORT] = cast(ushort)((0x807F & vu[F.EXPPOS_SHORT]) | 0x3F00); + } + } + else if (!(*vi & 0x7FFF_FFFF)) + { + // vf is +-0.0 + exp = 0; + } + else + { + // subnormal + vf *= F.RECIP_EPSILON; + ex = vu[F.EXPPOS_SHORT] & F.EXPMASK; + exp = ((ex - F.EXPBIAS) >> 7) - T.mant_dig + 1; + vu[F.EXPPOS_SHORT] = + cast(ushort)(((-1 - F.EXPMASK) & vu[F.EXPPOS_SHORT]) | 0x3F00); + } + return vf; + } + else // static if (F.realFormat == RealFormat.ibmExtended) + { + assert(0, "frexp not implemented"); + } +} + +/// +@system unittest +{ + int exp; + real mantissa = frexp(123.456L, exp); + + // check if values are equal to 19 decimal digits of precision + assert(equalsDigit(mantissa * pow(2.0L, cast(real) exp), 123.456L, 19)); + + assert(frexp(-real.nan, exp) && exp == int.min); + assert(frexp(real.nan, exp) && exp == int.min); + assert(frexp(-real.infinity, exp) == -real.infinity && exp == int.min); + assert(frexp(real.infinity, exp) == real.infinity && exp == int.max); + assert(frexp(-0.0, exp) == -0.0 && exp == 0); + assert(frexp(0.0, exp) == 0.0 && exp == 0); +} + +@safe unittest +{ + import std.meta : AliasSeq; + import std.typecons : tuple, Tuple; + + foreach (T; AliasSeq!(real, double, float)) + { + Tuple!(T, T, int)[] vals = // x,frexp,exp + [ + tuple(T(0.0), T( 0.0 ), 0), + tuple(T(-0.0), T( -0.0), 0), + tuple(T(1.0), T( .5 ), 1), + tuple(T(-1.0), T( -.5 ), 1), + tuple(T(2.0), T( .5 ), 2), + tuple(T(float.min_normal/2.0f), T(.5), -126), + tuple(T.infinity, T.infinity, int.max), + tuple(-T.infinity, -T.infinity, int.min), + tuple(T.nan, T.nan, int.min), + tuple(-T.nan, -T.nan, int.min), + + // Phobos issue #16026: + tuple(3 * (T.min_normal * T.epsilon), T( .75), (T.min_exp - T.mant_dig) + 2) + ]; + + foreach (elem; vals) + { + T x = elem[0]; + T e = elem[1]; + int exp = elem[2]; + int eptr; + T v = frexp(x, eptr); + assert(isIdentical(e, v)); + assert(exp == eptr); + + } + + static if (floatTraits!(T).realFormat == RealFormat.ieeeExtended) + { + static T[3][] extendedvals = [ // x,frexp,exp + [0x1.a5f1c2eb3fe4efp+73L, 0x1.A5F1C2EB3FE4EFp-1L, 74], // normal + [0x1.fa01712e8f0471ap-1064L, 0x1.fa01712e8f0471ap-1L, -1063], + [T.min_normal, .5, -16381], + [T.min_normal/2.0L, .5, -16382] // subnormal + ]; + foreach (elem; extendedvals) + { + T x = elem[0]; + T e = elem[1]; + int exp = cast(int) elem[2]; + int eptr; + T v = frexp(x, eptr); + assert(isIdentical(e, v)); + assert(exp == eptr); + + } + } + } +} + +@safe unittest +{ + import std.meta : AliasSeq; + void foo() { + foreach (T; AliasSeq!(real, double, float)) + { + int exp; + const T a = 1; + immutable T b = 2; + auto c = frexp(a, exp); + auto d = frexp(b, exp); + } + } +} + +/****************************************** + * Extracts the exponent of x as a signed integral value. + * + * If x is not a special value, the result is the same as + * $(D cast(int) logb(x)). + * + * $(TABLE_SV + * $(TR $(TH x) $(TH ilogb(x)) $(TH Range error?)) + * $(TR $(TD 0) $(TD FP_ILOGB0) $(TD yes)) + * $(TR $(TD $(PLUSMN)$(INFIN)) $(TD int.max) $(TD no)) + * $(TR $(TD $(NAN)) $(TD FP_ILOGBNAN) $(TD no)) + * ) + */ +int ilogb(T)(const T x) @trusted pure nothrow @nogc +if (isFloatingPoint!T) +{ + import core.bitop : bsr; + alias F = floatTraits!T; + + union floatBits + { + T rv; + ushort[T.sizeof/2] vu; + uint[T.sizeof/4] vui; + static if (T.sizeof >= 8) + ulong[T.sizeof/8] vul; + } + floatBits y = void; + y.rv = x; + + int ex = y.vu[F.EXPPOS_SHORT] & F.EXPMASK; + static if (F.realFormat == RealFormat.ieeeExtended) + { + if (ex) + { + // If exponent is non-zero + if (ex == F.EXPMASK) // infinity or NaN + { + if (y.vul[0] & 0x7FFF_FFFF_FFFF_FFFF) // NaN + return FP_ILOGBNAN; + else // +-infinity + return int.max; + } + else + { + return ex - F.EXPBIAS - 1; + } + } + else if (!y.vul[0]) + { + // vf is +-0.0 + return FP_ILOGB0; + } + else + { + // subnormal + return ex - F.EXPBIAS - T.mant_dig + 1 + bsr(y.vul[0]); + } + } + else static if (F.realFormat == RealFormat.ieeeQuadruple) + { + if (ex) // If exponent is non-zero + { + if (ex == F.EXPMASK) + { + // infinity or NaN + if (y.vul[MANTISSA_LSB] | ( y.vul[MANTISSA_MSB] & 0x0000_FFFF_FFFF_FFFF)) // NaN + return FP_ILOGBNAN; + else // +- infinity + return int.max; + } + else + { + return ex - F.EXPBIAS - 1; + } + } + else if ((y.vul[MANTISSA_LSB] | (y.vul[MANTISSA_MSB] & 0x0000_FFFF_FFFF_FFFF)) == 0) + { + // vf is +-0.0 + return FP_ILOGB0; + } + else + { + // subnormal + const ulong msb = y.vul[MANTISSA_MSB] & 0x0000_FFFF_FFFF_FFFF; + const ulong lsb = y.vul[MANTISSA_LSB]; + if (msb) + return ex - F.EXPBIAS - T.mant_dig + 1 + bsr(msb) + 64; + else + return ex - F.EXPBIAS - T.mant_dig + 1 + bsr(lsb); + } + } + else static if (F.realFormat == RealFormat.ieeeDouble) + { + if (ex) // If exponent is non-zero + { + if (ex == F.EXPMASK) // infinity or NaN + { + if ((y.vul[0] & 0x7FFF_FFFF_FFFF_FFFF) == 0x7FF0_0000_0000_0000) // +- infinity + return int.max; + else // NaN + return FP_ILOGBNAN; + } + else + { + return ((ex - F.EXPBIAS) >> 4) - 1; + } + } + else if (!(y.vul[0] & 0x7FFF_FFFF_FFFF_FFFF)) + { + // vf is +-0.0 + return FP_ILOGB0; + } + else + { + // subnormal + enum MANTISSAMASK_64 = ((cast(ulong) F.MANTISSAMASK_INT) << 32) | 0xFFFF_FFFF; + return ((ex - F.EXPBIAS) >> 4) - T.mant_dig + 1 + bsr(y.vul[0] & MANTISSAMASK_64); + } + } + else static if (F.realFormat == RealFormat.ieeeSingle) + { + if (ex) // If exponent is non-zero + { + if (ex == F.EXPMASK) // infinity or NaN + { + if ((y.vui[0] & 0x7FFF_FFFF) == 0x7F80_0000) // +- infinity + return int.max; + else // NaN + return FP_ILOGBNAN; + } + else + { + return ((ex - F.EXPBIAS) >> 7) - 1; + } + } + else if (!(y.vui[0] & 0x7FFF_FFFF)) + { + // vf is +-0.0 + return FP_ILOGB0; + } + else + { + // subnormal + const uint mantissa = y.vui[0] & F.MANTISSAMASK_INT; + return ((ex - F.EXPBIAS) >> 7) - T.mant_dig + 1 + bsr(mantissa); + } + } + else // static if (F.realFormat == RealFormat.ibmExtended) + { + core.stdc.math.ilogbl(x); + } +} +/// ditto +int ilogb(T)(const T x) @safe pure nothrow @nogc +if (isIntegral!T && isUnsigned!T) +{ + import core.bitop : bsr; + if (x == 0) + return FP_ILOGB0; + else + { + static assert(T.sizeof <= ulong.sizeof, "integer size too large for the current ilogb implementation"); + return bsr(x); + } +} +/// ditto +int ilogb(T)(const T x) @safe pure nothrow @nogc +if (isIntegral!T && isSigned!T) +{ + import std.traits : Unsigned; + // Note: abs(x) can not be used because the return type is not Unsigned and + // the return value would be wrong for x == int.min + Unsigned!T absx = x >= 0 ? x : -x; + return ilogb(absx); +} + +alias FP_ILOGB0 = core.stdc.math.FP_ILOGB0; +alias FP_ILOGBNAN = core.stdc.math.FP_ILOGBNAN; + +@system nothrow @nogc unittest +{ + import std.meta : AliasSeq; + import std.typecons : Tuple; + foreach (F; AliasSeq!(float, double, real)) + { + alias T = Tuple!(F, int); + T[13] vals = // x, ilogb(x) + [ + T( F.nan , FP_ILOGBNAN ), + T( -F.nan , FP_ILOGBNAN ), + T( F.infinity, int.max ), + T( -F.infinity, int.max ), + T( 0.0 , FP_ILOGB0 ), + T( -0.0 , FP_ILOGB0 ), + T( 2.0 , 1 ), + T( 2.0001 , 1 ), + T( 1.9999 , 0 ), + T( 0.5 , -1 ), + T( 123.123 , 6 ), + T( -123.123 , 6 ), + T( 0.123 , -4 ), + ]; + + foreach (elem; vals) + { + assert(ilogb(elem[0]) == elem[1]); + } + } + + // min_normal and subnormals + assert(ilogb(-float.min_normal) == -126); + assert(ilogb(nextUp(-float.min_normal)) == -127); + assert(ilogb(nextUp(-float(0.0))) == -149); + assert(ilogb(-double.min_normal) == -1022); + assert(ilogb(nextUp(-double.min_normal)) == -1023); + assert(ilogb(nextUp(-double(0.0))) == -1074); + static if (floatTraits!(real).realFormat == RealFormat.ieeeExtended) + { + assert(ilogb(-real.min_normal) == -16382); + assert(ilogb(nextUp(-real.min_normal)) == -16383); + assert(ilogb(nextUp(-real(0.0))) == -16445); + } + else static if (floatTraits!(real).realFormat == RealFormat.ieeeDouble) + { + assert(ilogb(-real.min_normal) == -1022); + assert(ilogb(nextUp(-real.min_normal)) == -1023); + assert(ilogb(nextUp(-real(0.0))) == -1074); + } + + // test integer types + assert(ilogb(0) == FP_ILOGB0); + assert(ilogb(int.max) == 30); + assert(ilogb(int.min) == 31); + assert(ilogb(uint.max) == 31); + assert(ilogb(long.max) == 62); + assert(ilogb(long.min) == 63); + assert(ilogb(ulong.max) == 63); +} + +/******************************************* + * Compute n * 2$(SUPERSCRIPT exp) + * References: frexp + */ + +real ldexp(real n, int exp) @nogc @safe pure nothrow { pragma(inline, true); return core.math.ldexp(n, exp); } +//FIXME +///ditto +double ldexp(double n, int exp) @safe pure nothrow @nogc { return ldexp(cast(real) n, exp); } +//FIXME +///ditto +float ldexp(float n, int exp) @safe pure nothrow @nogc { return ldexp(cast(real) n, exp); } + +/// +@nogc @safe pure nothrow unittest +{ + import std.meta : AliasSeq; + foreach (T; AliasSeq!(float, double, real)) + { + T r; + + r = ldexp(3.0L, 3); + assert(r == 24); + + r = ldexp(cast(T) 3.0, cast(int) 3); + assert(r == 24); + + T n = 3.0; + int exp = 3; + r = ldexp(n, exp); + assert(r == 24); + } +} + +@safe pure nothrow @nogc unittest +{ + static if (floatTraits!(real).realFormat == RealFormat.ieeeExtended) + { + assert(ldexp(1.0L, -16384) == 0x1p-16384L); + assert(ldexp(1.0L, -16382) == 0x1p-16382L); + int x; + real n = frexp(0x1p-16384L, x); + assert(n == 0.5L); + assert(x==-16383); + assert(ldexp(n, x)==0x1p-16384L); + } + else static if (floatTraits!(real).realFormat == RealFormat.ieeeDouble) + { + assert(ldexp(1.0L, -1024) == 0x1p-1024L); + assert(ldexp(1.0L, -1022) == 0x1p-1022L); + int x; + real n = frexp(0x1p-1024L, x); + assert(n == 0.5L); + assert(x==-1023); + assert(ldexp(n, x)==0x1p-1024L); + } + else static assert(false, "Floating point type real not supported"); +} + +/* workaround Issue 14718, float parsing depends on platform strtold +@safe pure nothrow @nogc unittest +{ + assert(ldexp(1.0, -1024) == 0x1p-1024); + assert(ldexp(1.0, -1022) == 0x1p-1022); + int x; + double n = frexp(0x1p-1024, x); + assert(n == 0.5); + assert(x==-1023); + assert(ldexp(n, x)==0x1p-1024); +} + +@safe pure nothrow @nogc unittest +{ + assert(ldexp(1.0f, -128) == 0x1p-128f); + assert(ldexp(1.0f, -126) == 0x1p-126f); + int x; + float n = frexp(0x1p-128f, x); + assert(n == 0.5f); + assert(x==-127); + assert(ldexp(n, x)==0x1p-128f); +} +*/ + +@system unittest +{ + static real[3][] vals = // value,exp,ldexp + [ + [ 0, 0, 0], + [ 1, 0, 1], + [ -1, 0, -1], + [ 1, 1, 2], + [ 123, 10, 125952], + [ real.max, int.max, real.infinity], + [ real.max, -int.max, 0], + [ real.min_normal, -int.max, 0], + ]; + int i; + + for (i = 0; i < vals.length; i++) + { + real x = vals[i][0]; + int exp = cast(int) vals[i][1]; + real z = vals[i][2]; + real l = ldexp(x, exp); + + assert(equalsDigit(z, l, 7)); + } + + real function(real, int) pldexp = &ldexp; + assert(pldexp != null); +} + +private +{ + version (INLINE_YL2X) {} else + { + static if (floatTraits!real.realFormat == RealFormat.ieeeQuadruple) + { + // Coefficients for log(1 + x) = x - x**2/2 + x**3 P(x)/Q(x) + static immutable real[13] logCoeffsP = [ + 1.313572404063446165910279910527789794488E4L, + 7.771154681358524243729929227226708890930E4L, + 2.014652742082537582487669938141683759923E5L, + 3.007007295140399532324943111654767187848E5L, + 2.854829159639697837788887080758954924001E5L, + 1.797628303815655343403735250238293741397E5L, + 7.594356839258970405033155585486712125861E4L, + 2.128857716871515081352991964243375186031E4L, + 3.824952356185897735160588078446136783779E3L, + 4.114517881637811823002128927449878962058E2L, + 2.321125933898420063925789532045674660756E1L, + 4.998469661968096229986658302195402690910E-1L, + 1.538612243596254322971797716843006400388E-6L + ]; + static immutable real[13] logCoeffsQ = [ + 3.940717212190338497730839731583397586124E4L, + 2.626900195321832660448791748036714883242E5L, + 7.777690340007566932935753241556479363645E5L, + 1.347518538384329112529391120390701166528E6L, + 1.514882452993549494932585972882995548426E6L, + 1.158019977462989115839826904108208787040E6L, + 6.132189329546557743179177159925690841200E5L, + 2.248234257620569139969141618556349415120E5L, + 5.605842085972455027590989944010492125825E4L, + 9.147150349299596453976674231612674085381E3L, + 9.104928120962988414618126155557301584078E2L, + 4.839208193348159620282142911143429644326E1L, + 1.0 + ]; + + // Coefficients for log(x) = z + z^3 P(z^2)/Q(z^2) + // where z = 2(x-1)/(x+1) + static immutable real[6] logCoeffsR = [ + -8.828896441624934385266096344596648080902E-1L, + 8.057002716646055371965756206836056074715E1L, + -2.024301798136027039250415126250455056397E3L, + 2.048819892795278657810231591630928516206E4L, + -8.977257995689735303686582344659576526998E4L, + 1.418134209872192732479751274970992665513E5L + ]; + static immutable real[6] logCoeffsS = [ + 1.701761051846631278975701529965589676574E6L + -1.332535117259762928288745111081235577029E6L, + 4.001557694070773974936904547424676279307E5L, + -5.748542087379434595104154610899551484314E4L, + 3.998526750980007367835804959888064681098E3L, + -1.186359407982897997337150403816839480438E2L, + 1.0 + ]; + } + else + { + // Coefficients for log(1 + x) = x - x**2/2 + x**3 P(x)/Q(x) + static immutable real[7] logCoeffsP = [ + 2.0039553499201281259648E1L, + 5.7112963590585538103336E1L, + 6.0949667980987787057556E1L, + 2.9911919328553073277375E1L, + 6.5787325942061044846969E0L, + 4.9854102823193375972212E-1L, + 4.5270000862445199635215E-5L, + ]; + static immutable real[7] logCoeffsQ = [ + 6.0118660497603843919306E1L, + 2.1642788614495947685003E2L, + 3.0909872225312059774938E2L, + 2.2176239823732856465394E2L, + 8.3047565967967209469434E1L, + 1.5062909083469192043167E1L, + 1.0000000000000000000000E0L, + ]; + + // Coefficients for log(x) = z + z^3 P(z^2)/Q(z^2) + // where z = 2(x-1)/(x+1) + static immutable real[4] logCoeffsR = [ + -3.5717684488096787370998E1L, + 1.0777257190312272158094E1L, + -7.1990767473014147232598E-1L, + 1.9757429581415468984296E-3L, + ]; + static immutable real[4] logCoeffsS = [ + -4.2861221385716144629696E2L, + 1.9361891836232102174846E2L, + -2.6201045551331104417768E1L, + 1.0000000000000000000000E0L, + ]; + } + } +} + +/************************************** + * Calculate the natural logarithm of x. + * + * $(TABLE_SV + * $(TR $(TH x) $(TH log(x)) $(TH divide by 0?) $(TH invalid?)) + * $(TR $(TD $(PLUSMN)0.0) $(TD -$(INFIN)) $(TD yes) $(TD no)) + * $(TR $(TD $(LT)0.0) $(TD $(NAN)) $(TD no) $(TD yes)) + * $(TR $(TD +$(INFIN)) $(TD +$(INFIN)) $(TD no) $(TD no)) + * ) + */ +real log(real x) @safe pure nothrow @nogc +{ + version (INLINE_YL2X) + return core.math.yl2x(x, LN2); + else + { + // C1 + C2 = LN2. + enum real C1 = 6.93145751953125E-1L; + enum real C2 = 1.428606820309417232121458176568075500134E-6L; + + // Special cases. + if (isNaN(x)) + return x; + if (isInfinity(x) && !signbit(x)) + return x; + if (x == 0.0) + return -real.infinity; + if (x < 0.0) + return real.nan; + + // Separate mantissa from exponent. + // Note, frexp is used so that denormal numbers will be handled properly. + real y, z; + int exp; + + x = frexp(x, exp); + + // Logarithm using log(x) = z + z^^3 R(z) / S(z), + // where z = 2(x - 1)/(x + 1) + if ((exp > 2) || (exp < -2)) + { + if (x < SQRT1_2) + { // 2(2x - 1)/(2x + 1) + exp -= 1; + z = x - 0.5; + y = 0.5 * z + 0.5; + } + else + { // 2(x - 1)/(x + 1) + z = x - 0.5; + z -= 0.5; + y = 0.5 * x + 0.5; + } + x = z / y; + z = x * x; + z = x * (z * poly(z, logCoeffsR) / poly(z, logCoeffsS)); + z += exp * C2; + z += x; + z += exp * C1; + + return z; + } + + // Logarithm using log(1 + x) = x - .5x^^2 + x^^3 P(x) / Q(x) + if (x < SQRT1_2) + { // 2x - 1 + exp -= 1; + x = ldexp(x, 1) - 1.0; + } + else + { + x = x - 1.0; + } + z = x * x; + y = x * (z * poly(x, logCoeffsP) / poly(x, logCoeffsQ)); + y += exp * C2; + z = y - ldexp(z, -1); + + // Note, the sum of above terms does not exceed x/4, + // so it contributes at most about 1/4 lsb to the error. + z += x; + z += exp * C1; + + return z; + } +} + +/// +@safe pure nothrow @nogc unittest +{ + assert(log(E) == 1); +} + +/************************************** + * Calculate the base-10 logarithm of x. + * + * $(TABLE_SV + * $(TR $(TH x) $(TH log10(x)) $(TH divide by 0?) $(TH invalid?)) + * $(TR $(TD $(PLUSMN)0.0) $(TD -$(INFIN)) $(TD yes) $(TD no)) + * $(TR $(TD $(LT)0.0) $(TD $(NAN)) $(TD no) $(TD yes)) + * $(TR $(TD +$(INFIN)) $(TD +$(INFIN)) $(TD no) $(TD no)) + * ) + */ +real log10(real x) @safe pure nothrow @nogc +{ + version (INLINE_YL2X) + return core.math.yl2x(x, LOG2); + else + { + // log10(2) split into two parts. + enum real L102A = 0.3125L; + enum real L102B = -1.14700043360188047862611052755069732318101185E-2L; + + // log10(e) split into two parts. + enum real L10EA = 0.5L; + enum real L10EB = -6.570551809674817234887108108339491770560299E-2L; + + // Special cases are the same as for log. + if (isNaN(x)) + return x; + if (isInfinity(x) && !signbit(x)) + return x; + if (x == 0.0) + return -real.infinity; + if (x < 0.0) + return real.nan; + + // Separate mantissa from exponent. + // Note, frexp is used so that denormal numbers will be handled properly. + real y, z; + int exp; + + x = frexp(x, exp); + + // Logarithm using log(x) = z + z^^3 R(z) / S(z), + // where z = 2(x - 1)/(x + 1) + if ((exp > 2) || (exp < -2)) + { + if (x < SQRT1_2) + { // 2(2x - 1)/(2x + 1) + exp -= 1; + z = x - 0.5; + y = 0.5 * z + 0.5; + } + else + { // 2(x - 1)/(x + 1) + z = x - 0.5; + z -= 0.5; + y = 0.5 * x + 0.5; + } + x = z / y; + z = x * x; + y = x * (z * poly(z, logCoeffsR) / poly(z, logCoeffsS)); + goto Ldone; + } + + // Logarithm using log(1 + x) = x - .5x^^2 + x^^3 P(x) / Q(x) + if (x < SQRT1_2) + { // 2x - 1 + exp -= 1; + x = ldexp(x, 1) - 1.0; + } + else + x = x - 1.0; + + z = x * x; + y = x * (z * poly(x, logCoeffsP) / poly(x, logCoeffsQ)); + y = y - ldexp(z, -1); + + // Multiply log of fraction by log10(e) and base 2 exponent by log10(2). + // This sequence of operations is critical and it may be horribly + // defeated by some compiler optimizers. + Ldone: + z = y * L10EB; + z += x * L10EB; + z += exp * L102B; + z += y * L10EA; + z += x * L10EA; + z += exp * L102A; + + return z; + } +} + +/// +@safe pure nothrow @nogc unittest +{ + assert(fabs(log10(1000) - 3) < .000001); +} + +/****************************************** + * Calculates the natural logarithm of 1 + x. + * + * For very small x, log1p(x) will be more accurate than + * log(1 + x). + * + * $(TABLE_SV + * $(TR $(TH x) $(TH log1p(x)) $(TH divide by 0?) $(TH invalid?)) + * $(TR $(TD $(PLUSMN)0.0) $(TD $(PLUSMN)0.0) $(TD no) $(TD no)) + * $(TR $(TD -1.0) $(TD -$(INFIN)) $(TD yes) $(TD no)) + * $(TR $(TD $(LT)-1.0) $(TD $(NAN)) $(TD no) $(TD yes)) + * $(TR $(TD +$(INFIN)) $(TD -$(INFIN)) $(TD no) $(TD no)) + * ) + */ +real log1p(real x) @safe pure nothrow @nogc +{ + version (INLINE_YL2X) + { + // On x87, yl2xp1 is valid if and only if -0.5 <= lg(x) <= 0.5, + // ie if -0.29 <= x <= 0.414 + return (fabs(x) <= 0.25) ? core.math.yl2xp1(x, LN2) : core.math.yl2x(x+1, LN2); + } + else + { + // Special cases. + if (isNaN(x) || x == 0.0) + return x; + if (isInfinity(x) && !signbit(x)) + return x; + if (x == -1.0) + return -real.infinity; + if (x < -1.0) + return real.nan; + + return log(x + 1.0); + } +} + +/*************************************** + * Calculates the base-2 logarithm of x: + * $(SUB log, 2)x + * + * $(TABLE_SV + * $(TR $(TH x) $(TH log2(x)) $(TH divide by 0?) $(TH invalid?)) + * $(TR $(TD $(PLUSMN)0.0) $(TD -$(INFIN)) $(TD yes) $(TD no) ) + * $(TR $(TD $(LT)0.0) $(TD $(NAN)) $(TD no) $(TD yes) ) + * $(TR $(TD +$(INFIN)) $(TD +$(INFIN)) $(TD no) $(TD no) ) + * ) + */ +real log2(real x) @safe pure nothrow @nogc +{ + version (INLINE_YL2X) + return core.math.yl2x(x, 1); + else + { + // Special cases are the same as for log. + if (isNaN(x)) + return x; + if (isInfinity(x) && !signbit(x)) + return x; + if (x == 0.0) + return -real.infinity; + if (x < 0.0) + return real.nan; + + // Separate mantissa from exponent. + // Note, frexp is used so that denormal numbers will be handled properly. + real y, z; + int exp; + + x = frexp(x, exp); + + // Logarithm using log(x) = z + z^^3 R(z) / S(z), + // where z = 2(x - 1)/(x + 1) + if ((exp > 2) || (exp < -2)) + { + if (x < SQRT1_2) + { // 2(2x - 1)/(2x + 1) + exp -= 1; + z = x - 0.5; + y = 0.5 * z + 0.5; + } + else + { // 2(x - 1)/(x + 1) + z = x - 0.5; + z -= 0.5; + y = 0.5 * x + 0.5; + } + x = z / y; + z = x * x; + y = x * (z * poly(z, logCoeffsR) / poly(z, logCoeffsS)); + goto Ldone; + } + + // Logarithm using log(1 + x) = x - .5x^^2 + x^^3 P(x) / Q(x) + if (x < SQRT1_2) + { // 2x - 1 + exp -= 1; + x = ldexp(x, 1) - 1.0; + } + else + x = x - 1.0; + + z = x * x; + y = x * (z * poly(x, logCoeffsP) / poly(x, logCoeffsQ)); + y = y - ldexp(z, -1); + + // Multiply log of fraction by log10(e) and base 2 exponent by log10(2). + // This sequence of operations is critical and it may be horribly + // defeated by some compiler optimizers. + Ldone: + z = y * (LOG2E - 1.0); + z += x * (LOG2E - 1.0); + z += y; + z += x; + z += exp; + + return z; + } +} + +/// +@system unittest +{ + // check if values are equal to 19 decimal digits of precision + assert(equalsDigit(log2(1024.0L), 10, 19)); +} + +/***************************************** + * Extracts the exponent of x as a signed integral value. + * + * If x is subnormal, it is treated as if it were normalized. + * For a positive, finite x: + * + * 1 $(LT)= $(I x) * FLT_RADIX$(SUPERSCRIPT -logb(x)) $(LT) FLT_RADIX + * + * $(TABLE_SV + * $(TR $(TH x) $(TH logb(x)) $(TH divide by 0?) ) + * $(TR $(TD $(PLUSMN)$(INFIN)) $(TD +$(INFIN)) $(TD no)) + * $(TR $(TD $(PLUSMN)0.0) $(TD -$(INFIN)) $(TD yes) ) + * ) + */ +real logb(real x) @trusted nothrow @nogc +{ + version (Win64_DMD_InlineAsm) + { + asm pure nothrow @nogc + { + naked ; + fld real ptr [RCX] ; + fxtract ; + fstp ST(0) ; + ret ; + } + } + else version (CRuntime_Microsoft) + { + asm pure nothrow @nogc + { + fld x ; + fxtract ; + fstp ST(0) ; + } + } + else + return core.stdc.math.logbl(x); +} + +/************************************ + * Calculates the remainder from the calculation x/y. + * Returns: + * The value of x - i * y, where i is the number of times that y can + * be completely subtracted from x. The result has the same sign as x. + * + * $(TABLE_SV + * $(TR $(TH x) $(TH y) $(TH fmod(x, y)) $(TH invalid?)) + * $(TR $(TD $(PLUSMN)0.0) $(TD not 0.0) $(TD $(PLUSMN)0.0) $(TD no)) + * $(TR $(TD $(PLUSMNINF)) $(TD anything) $(TD $(NAN)) $(TD yes)) + * $(TR $(TD anything) $(TD $(PLUSMN)0.0) $(TD $(NAN)) $(TD yes)) + * $(TR $(TD !=$(PLUSMNINF)) $(TD $(PLUSMNINF)) $(TD x) $(TD no)) + * ) + */ +real fmod(real x, real y) @trusted nothrow @nogc +{ + version (CRuntime_Microsoft) + { + return x % y; + } + else + return core.stdc.math.fmodl(x, y); +} + +/************************************ + * Breaks x into an integral part and a fractional part, each of which has + * the same sign as x. The integral part is stored in i. + * Returns: + * The fractional part of x. + * + * $(TABLE_SV + * $(TR $(TH x) $(TH i (on input)) $(TH modf(x, i)) $(TH i (on return))) + * $(TR $(TD $(PLUSMNINF)) $(TD anything) $(TD $(PLUSMN)0.0) $(TD $(PLUSMNINF))) + * ) + */ +real modf(real x, ref real i) @trusted nothrow @nogc +{ + version (CRuntime_Microsoft) + { + i = trunc(x); + return copysign(isInfinity(x) ? 0.0 : x - i, x); + } + else + return core.stdc.math.modfl(x,&i); +} + +/************************************* + * Efficiently calculates x * 2$(SUPERSCRIPT n). + * + * scalbn handles underflow and overflow in + * the same fashion as the basic arithmetic operators. + * + * $(TABLE_SV + * $(TR $(TH x) $(TH scalb(x))) + * $(TR $(TD $(PLUSMNINF)) $(TD $(PLUSMNINF)) ) + * $(TR $(TD $(PLUSMN)0.0) $(TD $(PLUSMN)0.0) ) + * ) + */ +real scalbn(real x, int n) @trusted nothrow @nogc +{ + version (InlineAsm_X86_Any) + { + // scalbnl is not supported on DMD-Windows, so use asm pure nothrow @nogc. + version (Win64) + { + asm pure nothrow @nogc { + naked ; + mov 16[RSP],RCX ; + fild word ptr 16[RSP] ; + fld real ptr [RDX] ; + fscale ; + fstp ST(1) ; + ret ; + } + } + else + { + asm pure nothrow @nogc { + fild n; + fld x; + fscale; + fstp ST(1); + } + } + } + else + { + return core.stdc.math.scalbnl(x, n); + } +} + +/// +@safe nothrow @nogc unittest +{ + assert(scalbn(-real.infinity, 5) == -real.infinity); +} + +/*************** + * Calculates the cube root of x. + * + * $(TABLE_SV + * $(TR $(TH $(I x)) $(TH cbrt(x)) $(TH invalid?)) + * $(TR $(TD $(PLUSMN)0.0) $(TD $(PLUSMN)0.0) $(TD no) ) + * $(TR $(TD $(NAN)) $(TD $(NAN)) $(TD yes) ) + * $(TR $(TD $(PLUSMN)$(INFIN)) $(TD $(PLUSMN)$(INFIN)) $(TD no) ) + * ) + */ +real cbrt(real x) @trusted nothrow @nogc +{ + version (CRuntime_Microsoft) + { + version (INLINE_YL2X) + return copysign(exp2(core.math.yl2x(fabs(x), 1.0L/3.0L)), x); + else + return core.stdc.math.cbrtl(x); + } + else + return core.stdc.math.cbrtl(x); +} + + +/******************************* + * Returns |x| + * + * $(TABLE_SV + * $(TR $(TH x) $(TH fabs(x))) + * $(TR $(TD $(PLUSMN)0.0) $(TD +0.0) ) + * $(TR $(TD $(PLUSMN)$(INFIN)) $(TD +$(INFIN)) ) + * ) + */ +real fabs(real x) @safe pure nothrow @nogc { pragma(inline, true); return core.math.fabs(x); } +//FIXME +///ditto +double fabs(double x) @safe pure nothrow @nogc { return fabs(cast(real) x); } +//FIXME +///ditto +float fabs(float x) @safe pure nothrow @nogc { return fabs(cast(real) x); } + +@safe unittest +{ + real function(real) pfabs = &fabs; + assert(pfabs != null); +} + +/*********************************************************************** + * Calculates the length of the + * hypotenuse of a right-angled triangle with sides of length x and y. + * The hypotenuse is the value of the square root of + * the sums of the squares of x and y: + * + * sqrt($(POWER x, 2) + $(POWER y, 2)) + * + * Note that hypot(x, y), hypot(y, x) and + * hypot(x, -y) are equivalent. + * + * $(TABLE_SV + * $(TR $(TH x) $(TH y) $(TH hypot(x, y)) $(TH invalid?)) + * $(TR $(TD x) $(TD $(PLUSMN)0.0) $(TD |x|) $(TD no)) + * $(TR $(TD $(PLUSMNINF)) $(TD y) $(TD +$(INFIN)) $(TD no)) + * $(TR $(TD $(PLUSMNINF)) $(TD $(NAN)) $(TD +$(INFIN)) $(TD no)) + * ) + */ + +real hypot(real x, real y) @safe pure nothrow @nogc +{ + // Scale x and y to avoid underflow and overflow. + // If one is huge and the other tiny, return the larger. + // If both are huge, avoid overflow by scaling by 1/sqrt(real.max/2). + // If both are tiny, avoid underflow by scaling by sqrt(real.min_normal*real.epsilon). + + enum real SQRTMIN = 0.5 * sqrt(real.min_normal); // This is a power of 2. + enum real SQRTMAX = 1.0L / SQRTMIN; // 2^^((max_exp)/2) = nextUp(sqrt(real.max)) + + static assert(2*(SQRTMAX/2)*(SQRTMAX/2) <= real.max); + + // Proves that sqrt(real.max) ~~ 0.5/sqrt(real.min_normal) + static assert(real.min_normal*real.max > 2 && real.min_normal*real.max <= 4); + + real u = fabs(x); + real v = fabs(y); + if (!(u >= v)) // check for NaN as well. + { + v = u; + u = fabs(y); + if (u == real.infinity) return u; // hypot(inf, nan) == inf + if (v == real.infinity) return v; // hypot(nan, inf) == inf + } + + // Now u >= v, or else one is NaN. + if (v >= SQRTMAX*0.5) + { + // hypot(huge, huge) -- avoid overflow + u *= SQRTMIN*0.5; + v *= SQRTMIN*0.5; + return sqrt(u*u + v*v) * SQRTMAX * 2.0; + } + + if (u <= SQRTMIN) + { + // hypot (tiny, tiny) -- avoid underflow + // This is only necessary to avoid setting the underflow + // flag. + u *= SQRTMAX / real.epsilon; + v *= SQRTMAX / real.epsilon; + return sqrt(u*u + v*v) * SQRTMIN * real.epsilon; + } + + if (u * real.epsilon > v) + { + // hypot (huge, tiny) = huge + return u; + } + + // both are in the normal range + return sqrt(u*u + v*v); +} + +@safe unittest +{ + static real[3][] vals = // x,y,hypot + [ + [ 0.0, 0.0, 0.0], + [ 0.0, -0.0, 0.0], + [ -0.0, -0.0, 0.0], + [ 3.0, 4.0, 5.0], + [ -300, -400, 500], + [0.0, 7.0, 7.0], + [9.0, 9*real.epsilon, 9.0], + [88/(64*sqrt(real.min_normal)), 105/(64*sqrt(real.min_normal)), 137/(64*sqrt(real.min_normal))], + [88/(128*sqrt(real.min_normal)), 105/(128*sqrt(real.min_normal)), 137/(128*sqrt(real.min_normal))], + [3*real.min_normal*real.epsilon, 4*real.min_normal*real.epsilon, 5*real.min_normal*real.epsilon], + [ real.min_normal, real.min_normal, sqrt(2.0L)*real.min_normal], + [ real.max/sqrt(2.0L), real.max/sqrt(2.0L), real.max], + [ real.infinity, real.nan, real.infinity], + [ real.nan, real.infinity, real.infinity], + [ real.nan, real.nan, real.nan], + [ real.nan, real.max, real.nan], + [ real.max, real.nan, real.nan], + ]; + for (int i = 0; i < vals.length; i++) + { + real x = vals[i][0]; + real y = vals[i][1]; + real z = vals[i][2]; + real h = hypot(x, y); + assert(isIdentical(z,h) || feqrel(z, h) >= real.mant_dig - 1); + } +} + +/************************************** + * Returns the value of x rounded upward to the next integer + * (toward positive infinity). + */ +real ceil(real x) @trusted pure nothrow @nogc +{ + version (Win64_DMD_InlineAsm) + { + asm pure nothrow @nogc + { + naked ; + fld real ptr [RCX] ; + fstcw 8[RSP] ; + mov AL,9[RSP] ; + mov DL,AL ; + and AL,0xC3 ; + or AL,0x08 ; // round to +infinity + mov 9[RSP],AL ; + fldcw 8[RSP] ; + frndint ; + mov 9[RSP],DL ; + fldcw 8[RSP] ; + ret ; + } + } + else version (CRuntime_Microsoft) + { + short cw; + asm pure nothrow @nogc + { + fld x ; + fstcw cw ; + mov AL,byte ptr cw+1 ; + mov DL,AL ; + and AL,0xC3 ; + or AL,0x08 ; // round to +infinity + mov byte ptr cw+1,AL ; + fldcw cw ; + frndint ; + mov byte ptr cw+1,DL ; + fldcw cw ; + } + } + else + { + // Special cases. + if (isNaN(x) || isInfinity(x)) + return x; + + real y = floorImpl(x); + if (y < x) + y += 1.0; + + return y; + } +} + +/// +@safe pure nothrow @nogc unittest +{ + assert(ceil(+123.456L) == +124); + assert(ceil(-123.456L) == -123); + assert(ceil(-1.234L) == -1); + assert(ceil(-0.123L) == 0); + assert(ceil(0.0L) == 0); + assert(ceil(+0.123L) == 1); + assert(ceil(+1.234L) == 2); + assert(ceil(real.infinity) == real.infinity); + assert(isNaN(ceil(real.nan))); + assert(isNaN(ceil(real.init))); +} + +// ditto +double ceil(double x) @trusted pure nothrow @nogc +{ + // Special cases. + if (isNaN(x) || isInfinity(x)) + return x; + + double y = floorImpl(x); + if (y < x) + y += 1.0; + + return y; +} + +@safe pure nothrow @nogc unittest +{ + assert(ceil(+123.456) == +124); + assert(ceil(-123.456) == -123); + assert(ceil(-1.234) == -1); + assert(ceil(-0.123) == 0); + assert(ceil(0.0) == 0); + assert(ceil(+0.123) == 1); + assert(ceil(+1.234) == 2); + assert(ceil(double.infinity) == double.infinity); + assert(isNaN(ceil(double.nan))); + assert(isNaN(ceil(double.init))); +} + +// ditto +float ceil(float x) @trusted pure nothrow @nogc +{ + // Special cases. + if (isNaN(x) || isInfinity(x)) + return x; + + float y = floorImpl(x); + if (y < x) + y += 1.0; + + return y; +} + +@safe pure nothrow @nogc unittest +{ + assert(ceil(+123.456f) == +124); + assert(ceil(-123.456f) == -123); + assert(ceil(-1.234f) == -1); + assert(ceil(-0.123f) == 0); + assert(ceil(0.0f) == 0); + assert(ceil(+0.123f) == 1); + assert(ceil(+1.234f) == 2); + assert(ceil(float.infinity) == float.infinity); + assert(isNaN(ceil(float.nan))); + assert(isNaN(ceil(float.init))); +} + +/************************************** + * Returns the value of x rounded downward to the next integer + * (toward negative infinity). + */ +real floor(real x) @trusted pure nothrow @nogc +{ + version (Win64_DMD_InlineAsm) + { + asm pure nothrow @nogc + { + naked ; + fld real ptr [RCX] ; + fstcw 8[RSP] ; + mov AL,9[RSP] ; + mov DL,AL ; + and AL,0xC3 ; + or AL,0x04 ; // round to -infinity + mov 9[RSP],AL ; + fldcw 8[RSP] ; + frndint ; + mov 9[RSP],DL ; + fldcw 8[RSP] ; + ret ; + } + } + else version (CRuntime_Microsoft) + { + short cw; + asm pure nothrow @nogc + { + fld x ; + fstcw cw ; + mov AL,byte ptr cw+1 ; + mov DL,AL ; + and AL,0xC3 ; + or AL,0x04 ; // round to -infinity + mov byte ptr cw+1,AL ; + fldcw cw ; + frndint ; + mov byte ptr cw+1,DL ; + fldcw cw ; + } + } + else + { + // Special cases. + if (isNaN(x) || isInfinity(x) || x == 0.0) + return x; + + return floorImpl(x); + } +} + +/// +@safe pure nothrow @nogc unittest +{ + assert(floor(+123.456L) == +123); + assert(floor(-123.456L) == -124); + assert(floor(-1.234L) == -2); + assert(floor(-0.123L) == -1); + assert(floor(0.0L) == 0); + assert(floor(+0.123L) == 0); + assert(floor(+1.234L) == 1); + assert(floor(real.infinity) == real.infinity); + assert(isNaN(floor(real.nan))); + assert(isNaN(floor(real.init))); +} + +// ditto +double floor(double x) @trusted pure nothrow @nogc +{ + // Special cases. + if (isNaN(x) || isInfinity(x) || x == 0.0) + return x; + + return floorImpl(x); +} + +@safe pure nothrow @nogc unittest +{ + assert(floor(+123.456) == +123); + assert(floor(-123.456) == -124); + assert(floor(-1.234) == -2); + assert(floor(-0.123) == -1); + assert(floor(0.0) == 0); + assert(floor(+0.123) == 0); + assert(floor(+1.234) == 1); + assert(floor(double.infinity) == double.infinity); + assert(isNaN(floor(double.nan))); + assert(isNaN(floor(double.init))); +} + +// ditto +float floor(float x) @trusted pure nothrow @nogc +{ + // Special cases. + if (isNaN(x) || isInfinity(x) || x == 0.0) + return x; + + return floorImpl(x); +} + +@safe pure nothrow @nogc unittest +{ + assert(floor(+123.456f) == +123); + assert(floor(-123.456f) == -124); + assert(floor(-1.234f) == -2); + assert(floor(-0.123f) == -1); + assert(floor(0.0f) == 0); + assert(floor(+0.123f) == 0); + assert(floor(+1.234f) == 1); + assert(floor(float.infinity) == float.infinity); + assert(isNaN(floor(float.nan))); + assert(isNaN(floor(float.init))); +} + +/** + * Round `val` to a multiple of `unit`. `rfunc` specifies the rounding + * function to use; by default this is `rint`, which uses the current + * rounding mode. + */ +Unqual!F quantize(alias rfunc = rint, F)(const F val, const F unit) +if (is(typeof(rfunc(F.init)) : F) && isFloatingPoint!F) +{ + typeof(return) ret = val; + if (unit != 0) + { + const scaled = val / unit; + if (!scaled.isInfinity) + ret = rfunc(scaled) * unit; + } + return ret; +} + +/// +@safe pure nothrow @nogc unittest +{ + assert(12345.6789L.quantize(0.01L) == 12345.68L); + assert(12345.6789L.quantize!floor(0.01L) == 12345.67L); + assert(12345.6789L.quantize(22.0L) == 12342.0L); +} + +/// +@safe pure nothrow @nogc unittest +{ + assert(12345.6789L.quantize(0) == 12345.6789L); + assert(12345.6789L.quantize(real.infinity).isNaN); + assert(12345.6789L.quantize(real.nan).isNaN); + assert(real.infinity.quantize(0.01L) == real.infinity); + assert(real.infinity.quantize(real.nan).isNaN); + assert(real.nan.quantize(0.01L).isNaN); + assert(real.nan.quantize(real.infinity).isNaN); + assert(real.nan.quantize(real.nan).isNaN); +} + +/** + * Round `val` to a multiple of `pow(base, exp)`. `rfunc` specifies the + * rounding function to use; by default this is `rint`, which uses the + * current rounding mode. + */ +Unqual!F quantize(real base, alias rfunc = rint, F, E)(const F val, const E exp) +if (is(typeof(rfunc(F.init)) : F) && isFloatingPoint!F && isIntegral!E) +{ + // TODO: Compile-time optimization for power-of-two bases? + return quantize!rfunc(val, pow(cast(F) base, exp)); +} + +/// ditto +Unqual!F quantize(real base, long exp = 1, alias rfunc = rint, F)(const F val) +if (is(typeof(rfunc(F.init)) : F) && isFloatingPoint!F) +{ + enum unit = cast(F) pow(base, exp); + return quantize!rfunc(val, unit); +} + +/// +@safe pure nothrow @nogc unittest +{ + assert(12345.6789L.quantize!10(-2) == 12345.68L); + assert(12345.6789L.quantize!(10, -2) == 12345.68L); + assert(12345.6789L.quantize!(10, floor)(-2) == 12345.67L); + assert(12345.6789L.quantize!(10, -2, floor) == 12345.67L); + + assert(12345.6789L.quantize!22(1) == 12342.0L); + assert(12345.6789L.quantize!22 == 12342.0L); +} + +@safe pure nothrow @nogc unittest +{ + import std.meta : AliasSeq; + + foreach (F; AliasSeq!(real, double, float)) + { + const maxL10 = cast(int) F.max.log10.floor; + const maxR10 = pow(cast(F) 10, maxL10); + assert((cast(F) 0.9L * maxR10).quantize!10(maxL10) == maxR10); + assert((cast(F)-0.9L * maxR10).quantize!10(maxL10) == -maxR10); + + assert(F.max.quantize(F.min_normal) == F.max); + assert((-F.max).quantize(F.min_normal) == -F.max); + assert(F.min_normal.quantize(F.max) == 0); + assert((-F.min_normal).quantize(F.max) == 0); + assert(F.min_normal.quantize(F.min_normal) == F.min_normal); + assert((-F.min_normal).quantize(F.min_normal) == -F.min_normal); + } +} + +/****************************************** + * Rounds x to the nearest integer value, using the current rounding + * mode. + * + * Unlike the rint functions, nearbyint does not raise the + * FE_INEXACT exception. + */ +real nearbyint(real x) @trusted nothrow @nogc +{ + version (CRuntime_Microsoft) + { + assert(0); // not implemented in C library + } + else + return core.stdc.math.nearbyintl(x); +} + +/********************************** + * Rounds x to the nearest integer value, using the current rounding + * mode. + * If the return value is not equal to x, the FE_INEXACT + * exception is raised. + * $(B nearbyint) performs + * the same operation, but does not set the FE_INEXACT exception. + */ +real rint(real x) @safe pure nothrow @nogc { pragma(inline, true); return core.math.rint(x); } +//FIXME +///ditto +double rint(double x) @safe pure nothrow @nogc { return rint(cast(real) x); } +//FIXME +///ditto +float rint(float x) @safe pure nothrow @nogc { return rint(cast(real) x); } + +@safe unittest +{ + real function(real) print = &rint; + assert(print != null); +} + +/*************************************** + * Rounds x to the nearest integer value, using the current rounding + * mode. + * + * This is generally the fastest method to convert a floating-point number + * to an integer. Note that the results from this function + * depend on the rounding mode, if the fractional part of x is exactly 0.5. + * If using the default rounding mode (ties round to even integers) + * lrint(4.5) == 4, lrint(5.5)==6. + */ +long lrint(real x) @trusted pure nothrow @nogc +{ + version (InlineAsm_X86_Any) + { + version (Win64) + { + asm pure nothrow @nogc + { + naked; + fld real ptr [RCX]; + fistp qword ptr 8[RSP]; + mov RAX,8[RSP]; + ret; + } + } + else + { + long n; + asm pure nothrow @nogc + { + fld x; + fistp n; + } + return n; + } + } + else + { + alias F = floatTraits!(real); + static if (F.realFormat == RealFormat.ieeeDouble) + { + long result; + + // Rounding limit when casting from real(double) to ulong. + enum real OF = 4.50359962737049600000E15L; + + uint* vi = cast(uint*)(&x); + + // Find the exponent and sign + uint msb = vi[MANTISSA_MSB]; + uint lsb = vi[MANTISSA_LSB]; + int exp = ((msb >> 20) & 0x7ff) - 0x3ff; + const int sign = msb >> 31; + msb &= 0xfffff; + msb |= 0x100000; + + if (exp < 63) + { + if (exp >= 52) + result = (cast(long) msb << (exp - 20)) | (lsb << (exp - 52)); + else + { + // Adjust x and check result. + const real j = sign ? -OF : OF; + x = (j + x) - j; + msb = vi[MANTISSA_MSB]; + lsb = vi[MANTISSA_LSB]; + exp = ((msb >> 20) & 0x7ff) - 0x3ff; + msb &= 0xfffff; + msb |= 0x100000; + + if (exp < 0) + result = 0; + else if (exp < 20) + result = cast(long) msb >> (20 - exp); + else if (exp == 20) + result = cast(long) msb; + else + result = (cast(long) msb << (exp - 20)) | (lsb >> (52 - exp)); + } + } + else + { + // It is left implementation defined when the number is too large. + return cast(long) x; + } + + return sign ? -result : result; + } + else static if (F.realFormat == RealFormat.ieeeExtended) + { + long result; + + // Rounding limit when casting from real(80-bit) to ulong. + enum real OF = 9.22337203685477580800E18L; + + ushort* vu = cast(ushort*)(&x); + uint* vi = cast(uint*)(&x); + + // Find the exponent and sign + int exp = (vu[F.EXPPOS_SHORT] & 0x7fff) - 0x3fff; + const int sign = (vu[F.EXPPOS_SHORT] >> 15) & 1; + + if (exp < 63) + { + // Adjust x and check result. + const real j = sign ? -OF : OF; + x = (j + x) - j; + exp = (vu[F.EXPPOS_SHORT] & 0x7fff) - 0x3fff; + + version (LittleEndian) + { + if (exp < 0) + result = 0; + else if (exp <= 31) + result = vi[1] >> (31 - exp); + else + result = (cast(long) vi[1] << (exp - 31)) | (vi[0] >> (63 - exp)); + } + else + { + if (exp < 0) + result = 0; + else if (exp <= 31) + result = vi[1] >> (31 - exp); + else + result = (cast(long) vi[1] << (exp - 31)) | (vi[2] >> (63 - exp)); + } + } + else + { + // It is left implementation defined when the number is too large + // to fit in a 64bit long. + return cast(long) x; + } + + return sign ? -result : result; + } + else static if (F.realFormat == RealFormat.ieeeQuadruple) + { + const vu = cast(ushort*)(&x); + + // Find the exponent and sign + const sign = (vu[F.EXPPOS_SHORT] >> 15) & 1; + if ((vu[F.EXPPOS_SHORT] & F.EXPMASK) - (F.EXPBIAS + 1) > 63) + { + // The result is left implementation defined when the number is + // too large to fit in a 64 bit long. + return cast(long) x; + } + + // Force rounding of lower bits according to current rounding + // mode by adding ±2^-112 and subtracting it again. + enum OF = 5.19229685853482762853049632922009600E33L; + const j = sign ? -OF : OF; + x = (j + x) - j; + + const implicitOne = 1UL << 48; + auto vl = cast(ulong*)(&x); + vl[MANTISSA_MSB] &= implicitOne - 1; + vl[MANTISSA_MSB] |= implicitOne; + + long result; + + const exp = (vu[F.EXPPOS_SHORT] & F.EXPMASK) - (F.EXPBIAS + 1); + if (exp < 0) + result = 0; + else if (exp <= 48) + result = vl[MANTISSA_MSB] >> (48 - exp); + else + result = (vl[MANTISSA_MSB] << (exp - 48)) | (vl[MANTISSA_LSB] >> (112 - exp)); + + return sign ? -result : result; + } + else + { + static assert(false, "real type not supported by lrint()"); + } + } +} + +/// +@safe pure nothrow @nogc unittest +{ + assert(lrint(4.5) == 4); + assert(lrint(5.5) == 6); + assert(lrint(-4.5) == -4); + assert(lrint(-5.5) == -6); + + assert(lrint(int.max - 0.5) == 2147483646L); + assert(lrint(int.max + 0.5) == 2147483648L); + assert(lrint(int.min - 0.5) == -2147483648L); + assert(lrint(int.min + 0.5) == -2147483648L); +} + +static if (real.mant_dig >= long.sizeof * 8) +{ + @safe pure nothrow @nogc unittest + { + assert(lrint(long.max - 1.5L) == long.max - 1); + assert(lrint(long.max - 0.5L) == long.max - 1); + assert(lrint(long.min + 0.5L) == long.min); + assert(lrint(long.min + 1.5L) == long.min + 2); + } +} + +/******************************************* + * Return the value of x rounded to the nearest integer. + * If the fractional part of x is exactly 0.5, the return value is + * rounded away from zero. + */ +real round(real x) @trusted nothrow @nogc +{ + version (CRuntime_Microsoft) + { + auto old = FloatingPointControl.getControlState(); + FloatingPointControl.setControlState( + (old & ~FloatingPointControl.roundingMask) | FloatingPointControl.roundToZero + ); + x = rint((x >= 0) ? x + 0.5 : x - 0.5); + FloatingPointControl.setControlState(old); + return x; + } + else + return core.stdc.math.roundl(x); +} + +/********************************************** + * Return the value of x rounded to the nearest integer. + * + * If the fractional part of x is exactly 0.5, the return value is rounded + * away from zero. + * + * $(BLUE This function is Posix-Only.) + */ +long lround(real x) @trusted nothrow @nogc +{ + version (Posix) + return core.stdc.math.llroundl(x); + else + assert(0, "lround not implemented"); +} + +version (Posix) +{ + @safe nothrow @nogc unittest + { + assert(lround(0.49) == 0); + assert(lround(0.5) == 1); + assert(lround(1.5) == 2); + } +} + +/**************************************************** + * Returns the integer portion of x, dropping the fractional portion. + * + * This is also known as "chop" rounding. + */ +real trunc(real x) @trusted nothrow @nogc +{ + version (Win64_DMD_InlineAsm) + { + asm pure nothrow @nogc + { + naked ; + fld real ptr [RCX] ; + fstcw 8[RSP] ; + mov AL,9[RSP] ; + mov DL,AL ; + and AL,0xC3 ; + or AL,0x0C ; // round to 0 + mov 9[RSP],AL ; + fldcw 8[RSP] ; + frndint ; + mov 9[RSP],DL ; + fldcw 8[RSP] ; + ret ; + } + } + else version (CRuntime_Microsoft) + { + short cw; + asm pure nothrow @nogc + { + fld x ; + fstcw cw ; + mov AL,byte ptr cw+1 ; + mov DL,AL ; + and AL,0xC3 ; + or AL,0x0C ; // round to 0 + mov byte ptr cw+1,AL ; + fldcw cw ; + frndint ; + mov byte ptr cw+1,DL ; + fldcw cw ; + } + } + else + return core.stdc.math.truncl(x); +} + +/**************************************************** + * Calculate the remainder x REM y, following IEC 60559. + * + * REM is the value of x - y * n, where n is the integer nearest the exact + * value of x / y. + * If |n - x / y| == 0.5, n is even. + * If the result is zero, it has the same sign as x. + * Otherwise, the sign of the result is the sign of x / y. + * Precision mode has no effect on the remainder functions. + * + * remquo returns n in the parameter n. + * + * $(TABLE_SV + * $(TR $(TH x) $(TH y) $(TH remainder(x, y)) $(TH n) $(TH invalid?)) + * $(TR $(TD $(PLUSMN)0.0) $(TD not 0.0) $(TD $(PLUSMN)0.0) $(TD 0.0) $(TD no)) + * $(TR $(TD $(PLUSMNINF)) $(TD anything) $(TD $(NAN)) $(TD ?) $(TD yes)) + * $(TR $(TD anything) $(TD $(PLUSMN)0.0) $(TD $(NAN)) $(TD ?) $(TD yes)) + * $(TR $(TD != $(PLUSMNINF)) $(TD $(PLUSMNINF)) $(TD x) $(TD ?) $(TD no)) + * ) + * + * $(BLUE `remquo` and `remainder` not supported on Windows.) + */ +real remainder(real x, real y) @trusted nothrow @nogc +{ + version (CRuntime_Microsoft) + { + int n; + return remquo(x, y, n); + } + else + return core.stdc.math.remainderl(x, y); +} + +real remquo(real x, real y, out int n) @trusted nothrow @nogc /// ditto +{ + version (Posix) + return core.stdc.math.remquol(x, y, &n); + else + assert(0, "remquo not implemented"); +} + +/** IEEE exception status flags ('sticky bits') + + These flags indicate that an exceptional floating-point condition has occurred. + They indicate that a NaN or an infinity has been generated, that a result + is inexact, or that a signalling NaN has been encountered. If floating-point + exceptions are enabled (unmasked), a hardware exception will be generated + instead of setting these flags. + */ +struct IeeeFlags +{ +private: + // The x87 FPU status register is 16 bits. + // The Pentium SSE2 status register is 32 bits. + // The ARM and PowerPC FPSCR is a 32-bit register. + // The SPARC FSR is a 32bit register (64 bits for SPARC 7 & 8, but high bits are uninteresting). + uint flags; + + version (CRuntime_Microsoft) + { + // Microsoft uses hardware-incompatible custom constants in fenv.h (core.stdc.fenv). + // Applies to both x87 status word (16 bits) and SSE2 status word(32 bits). + enum : int + { + INEXACT_MASK = 0x20, + UNDERFLOW_MASK = 0x10, + OVERFLOW_MASK = 0x08, + DIVBYZERO_MASK = 0x04, + INVALID_MASK = 0x01, + + EXCEPTIONS_MASK = 0b11_1111 + } + // Don't bother about subnormals, they are not supported on most CPUs. + // SUBNORMAL_MASK = 0x02; + } + else + { + enum : int + { + INEXACT_MASK = core.stdc.fenv.FE_INEXACT, + UNDERFLOW_MASK = core.stdc.fenv.FE_UNDERFLOW, + OVERFLOW_MASK = core.stdc.fenv.FE_OVERFLOW, + DIVBYZERO_MASK = core.stdc.fenv.FE_DIVBYZERO, + INVALID_MASK = core.stdc.fenv.FE_INVALID, + EXCEPTIONS_MASK = core.stdc.fenv.FE_ALL_EXCEPT, + } + } + +private: + static uint getIeeeFlags() + { + version (GNU) + { + version (X86_Any) + { + ushort sw; + asm pure nothrow @nogc + { + "fstsw %0" : "=a" (sw); + } + // OR the result with the SSE2 status register (MXCSR). + if (haveSSE) + { + uint mxcsr; + asm pure nothrow @nogc + { + "stmxcsr %0" : "=m" (mxcsr); + } + return (sw | mxcsr) & EXCEPTIONS_MASK; + } + else + return sw & EXCEPTIONS_MASK; + } + else version (ARM) + { + version (ARM_SoftFloat) + return 0; + else + { + uint result = void; + asm pure nothrow @nogc + { + "vmrs %0, FPSCR; and %0, %0, #0x1F;" : "=r" result; + } + return result; + } + } + else + assert(0, "Not yet supported"); + } + else + version (InlineAsm_X86_Any) + { + ushort sw; + asm pure nothrow @nogc { fstsw sw; } + + // OR the result with the SSE2 status register (MXCSR). + if (haveSSE) + { + uint mxcsr; + asm pure nothrow @nogc { stmxcsr mxcsr; } + return (sw | mxcsr) & EXCEPTIONS_MASK; + } + else return sw & EXCEPTIONS_MASK; + } + else version (SPARC) + { + /* + int retval; + asm pure nothrow @nogc { st %fsr, retval; } + return retval; + */ + assert(0, "Not yet supported"); + } + else version (ARM) + { + assert(false, "Not yet supported."); + } + else + assert(0, "Not yet supported"); + } + static void resetIeeeFlags() @nogc + { + version (GNU) + { + version (X86_Any) + { + asm pure nothrow @nogc + { + "fnclex"; + } + + // Also clear exception flags in MXCSR, SSE's control register. + if (haveSSE) + { + uint mxcsr; + asm pure nothrow @nogc + { + "stmxcsr %0" : "=m" (mxcsr); + } + mxcsr &= ~EXCEPTIONS_MASK; + asm pure nothrow @nogc + { + "ldmxcsr %0" : : "m" (mxcsr); + } + } + } + else version (ARM) + { + version (ARM_SoftFloat) + return; + else + { + uint old = FloatingPointControl.getControlState(); + old &= ~0b11111; // http://infocenter.arm.com/help/topic/com.arm.doc.ddi0408i/Chdfifdc.html + asm pure nothrow @nogc + { + "vmsr FPSCR, %0" : : "r" (old); + } + } + } + else + assert(0, "Not yet supported"); + } + else + version (InlineAsm_X86_Any) + { + asm pure nothrow @nogc + { + fnclex; + } + + // Also clear exception flags in MXCSR, SSE's control register. + if (haveSSE) + { + uint mxcsr; + asm nothrow @nogc { stmxcsr mxcsr; } + mxcsr &= ~EXCEPTIONS_MASK; + asm nothrow @nogc { ldmxcsr mxcsr; } + } + } + else + { + /* SPARC: + int tmpval; + asm pure nothrow @nogc { st %fsr, tmpval; } + tmpval &=0xFFFF_FC00; + asm pure nothrow @nogc { ld tmpval, %fsr; } + */ + assert(0, "Not yet supported"); + } + } +public: + version (IeeeFlagsSupport) + { + + /** + * The result cannot be represented exactly, so rounding occurred. + * Example: `x = sin(0.1);` + */ + @property bool inexact() const { return (flags & INEXACT_MASK) != 0; } + + /** + * A zero was generated by underflow + * Example: `x = real.min*real.epsilon/2;` + */ + @property bool underflow() const { return (flags & UNDERFLOW_MASK) != 0; } + + /** + * An infinity was generated by overflow + * Example: `x = real.max*2;` + */ + @property bool overflow() const { return (flags & OVERFLOW_MASK) != 0; } + + /** + * An infinity was generated by division by zero + * Example: `x = 3/0.0;` + */ + @property bool divByZero() const { return (flags & DIVBYZERO_MASK) != 0; } + + /** + * A machine NaN was generated. + * Example: `x = real.infinity * 0.0;` + */ + @property bool invalid() const { return (flags & INVALID_MASK) != 0; } + + } +} + +/// +version (GNU) +{ + unittest + { + pragma(msg, "ieeeFlags test disabled, see LDC Issue #888"); + } +} +else +@system unittest +{ + static void func() { + int a = 10 * 10; + } + + real a=3.5; + // Set all the flags to zero + resetIeeeFlags(); + assert(!ieeeFlags.divByZero); + // Perform a division by zero. + a/=0.0L; + assert(a == real.infinity); + assert(ieeeFlags.divByZero); + // Create a NaN + a*=0.0L; + assert(ieeeFlags.invalid); + assert(isNaN(a)); + + // Check that calling func() has no effect on the + // status flags. + IeeeFlags f = ieeeFlags; + func(); + assert(ieeeFlags == f); +} + +version (GNU) +{ + unittest + { + pragma(msg, "ieeeFlags test disabled, see LDC Issue #888"); + } +} +else +@system unittest +{ + import std.meta : AliasSeq; + + static struct Test + { + void delegate() action; + bool function() ieeeCheck; + } + + foreach (T; AliasSeq!(float, double, real)) + { + T x; /* Needs to be here to trick -O. It would optimize away the + calculations if x were local to the function literals. */ + auto tests = [ + Test( + () { x = 1; x += 0.1; }, + () => ieeeFlags.inexact + ), + Test( + () { x = T.min_normal; x /= T.max; }, + () => ieeeFlags.underflow + ), + Test( + () { x = T.max; x += T.max; }, + () => ieeeFlags.overflow + ), + Test( + () { x = 1; x /= 0; }, + () => ieeeFlags.divByZero + ), + Test( + () { x = 0; x /= 0; }, + () => ieeeFlags.invalid + ) + ]; + foreach (test; tests) + { + resetIeeeFlags(); + assert(!test.ieeeCheck()); + test.action(); + assert(test.ieeeCheck()); + } + } +} + +version (X86_Any) +{ + version = IeeeFlagsSupport; +} +version (X86_Any) +{ + version = IeeeFlagsSupport; +} +else version (PPC_Any) +{ + version = IeeeFlagsSupport; +} +else version (MIPS_Any) +{ + version = IeeeFlagsSupport; +} +else version (ARM_Any) +{ + version = IeeeFlagsSupport; +} + +/// Set all of the floating-point status flags to false. +void resetIeeeFlags() @nogc { IeeeFlags.resetIeeeFlags(); } + +/// Returns: snapshot of the current state of the floating-point status flags +@property IeeeFlags ieeeFlags() +{ + return IeeeFlags(IeeeFlags.getIeeeFlags()); +} + +/** Control the Floating point hardware + + Change the IEEE754 floating-point rounding mode and the floating-point + hardware exceptions. + + By default, the rounding mode is roundToNearest and all hardware exceptions + are disabled. For most applications, debugging is easier if the $(I division + by zero), $(I overflow), and $(I invalid operation) exceptions are enabled. + These three are combined into a $(I severeExceptions) value for convenience. + Note in particular that if $(I invalidException) is enabled, a hardware trap + will be generated whenever an uninitialized floating-point variable is used. + + All changes are temporary. The previous state is restored at the + end of the scope. + + +Example: +---- +{ + FloatingPointControl fpctrl; + + // Enable hardware exceptions for division by zero, overflow to infinity, + // invalid operations, and uninitialized floating-point variables. + fpctrl.enableExceptions(FloatingPointControl.severeExceptions); + + // This will generate a hardware exception, if x is a + // default-initialized floating point variable: + real x; // Add `= 0` or even `= real.nan` to not throw the exception. + real y = x * 3.0; + + // The exception is only thrown for default-uninitialized NaN-s. + // NaN-s with other payload are valid: + real z = y * real.nan; // ok + + // Changing the rounding mode: + fpctrl.rounding = FloatingPointControl.roundUp; + assert(rint(1.1) == 2); + + // The set hardware exceptions will be disabled when leaving this scope. + // The original rounding mode will also be restored. +} + +// Ensure previous values are returned: +assert(!FloatingPointControl.enabledExceptions); +assert(FloatingPointControl.rounding == FloatingPointControl.roundToNearest); +assert(rint(1.1) == 1); +---- + + */ +struct FloatingPointControl +{ + alias RoundingMode = uint; /// + + version (StdDdoc) + { + enum : RoundingMode + { + /** IEEE rounding modes. + * The default mode is roundToNearest. + * + * roundingMask = A mask of all rounding modes. + */ + roundToNearest, + roundDown, /// ditto + roundUp, /// ditto + roundToZero, /// ditto + roundingMask, /// ditto + } + } + else version (CRuntime_Microsoft) + { + // Microsoft uses hardware-incompatible custom constants in fenv.h (core.stdc.fenv). + enum : RoundingMode + { + roundToNearest = 0x0000, + roundDown = 0x0400, + roundUp = 0x0800, + roundToZero = 0x0C00, + roundingMask = roundToNearest | roundDown + | roundUp | roundToZero, + } + } + else + { + enum : RoundingMode + { + roundToNearest = core.stdc.fenv.FE_TONEAREST, + roundDown = core.stdc.fenv.FE_DOWNWARD, + roundUp = core.stdc.fenv.FE_UPWARD, + roundToZero = core.stdc.fenv.FE_TOWARDZERO, + roundingMask = roundToNearest | roundDown + | roundUp | roundToZero, + } + } + + //// Change the floating-point hardware rounding mode + @property void rounding(RoundingMode newMode) @nogc + { + initialize(); + setControlState(cast(ushort)((getControlState() & (-1 - roundingMask)) | (newMode & roundingMask))); + } + + /// Returns: the currently active rounding mode + @property static RoundingMode rounding() @nogc + { + return cast(RoundingMode)(getControlState() & roundingMask); + } + + alias ExceptionMask = uint; /// + + version (StdDdoc) + { + enum : ExceptionMask + { + /** IEEE hardware exceptions. + * By default, all exceptions are masked (disabled). + * + * severeExceptions = The overflow, division by zero, and invalid + * exceptions. + */ + subnormalException, + inexactException, /// ditto + underflowException, /// ditto + overflowException, /// ditto + divByZeroException, /// ditto + invalidException, /// ditto + severeExceptions, /// ditto + allExceptions, /// ditto + } + } + else version (ARM_Any) + { + enum : ExceptionMask + { + subnormalException = 0x8000, + inexactException = 0x1000, + underflowException = 0x0800, + overflowException = 0x0400, + divByZeroException = 0x0200, + invalidException = 0x0100, + severeExceptions = overflowException | divByZeroException + | invalidException, + allExceptions = severeExceptions | underflowException + | inexactException | subnormalException, + } + } + else version (MIPS_Any) + { + enum : ExceptionMask + { + inexactException = 0x0080, + underflowException = 0x0100, + overflowException = 0x0200, + divByZeroException = 0x0400, + invalidException = 0x0800, + severeExceptions = overflowException | divByZeroException + | invalidException, + allExceptions = severeExceptions | underflowException + | inexactException, + } + } + else version (PPC_Any) + { + enum : ExceptionMask + { + inexactException = 0x08, + divByZeroException = 0x10, + underflowException = 0x20, + overflowException = 0x40, + invalidException = 0x80, + severeExceptions = overflowException | divByZeroException + | invalidException, + allExceptions = severeExceptions | underflowException + | inexactException, + } + } + else version (SPARC64) + { + enum : ExceptionMask + { + inexactException = 0x0800000, + divByZeroException = 0x1000000, + overflowException = 0x4000000, + underflowException = 0x2000000, + invalidException = 0x8000000, + severeExceptions = overflowException | divByZeroException + | invalidException, + allExceptions = severeExceptions | underflowException + | inexactException, + } + } + else version (SystemZ) + { + enum : ExceptionMask + { + inexactException = 0x08000000, + divByZeroException = 0x40000000, + overflowException = 0x20000000, + underflowException = 0x10000000, + invalidException = 0x80000000, + severeExceptions = overflowException | divByZeroException + | invalidException, + allExceptions = severeExceptions | underflowException + | inexactException, + } + } + else version (X86_Any) + { + enum : ExceptionMask + { + inexactException = 0x20, + underflowException = 0x10, + overflowException = 0x08, + divByZeroException = 0x04, + subnormalException = 0x02, + invalidException = 0x01, + severeExceptions = overflowException | divByZeroException + | invalidException, + allExceptions = severeExceptions | underflowException + | inexactException | subnormalException, + } + } + else + static assert(false, "Not implemented for this architecture"); + +public: + /// Returns: true if the current FPU supports exception trapping + @property static bool hasExceptionTraps() @safe nothrow @nogc + { + version (X86_Any) + return true; + else version (PPC_Any) + return true; + else version (MIPS_Any) + return true; + else version (ARM_Any) + { + auto oldState = getControlState(); + // If exceptions are not supported, we set the bit but read it back as zero + // https://sourceware.org/ml/libc-ports/2012-06/msg00091.html + setControlState(oldState | divByZeroException); + immutable result = (getControlState() & allExceptions) != 0; + setControlState(oldState); + return result; + } + else + assert(0, "Not yet supported"); + } + + /// Enable (unmask) specific hardware exceptions. Multiple exceptions may be ORed together. + void enableExceptions(ExceptionMask exceptions) @nogc + { + assert(hasExceptionTraps); + initialize(); + version (X86_Any) + setControlState(getControlState() & ~(exceptions & allExceptions)); + else + setControlState(getControlState() | (exceptions & allExceptions)); + } + + /// Disable (mask) specific hardware exceptions. Multiple exceptions may be ORed together. + void disableExceptions(ExceptionMask exceptions) @nogc + { + assert(hasExceptionTraps); + initialize(); + version (X86_Any) + setControlState(getControlState() | (exceptions & allExceptions)); + else + setControlState(getControlState() & ~(exceptions & allExceptions)); + } + + /// Returns: the exceptions which are currently enabled (unmasked) + @property static ExceptionMask enabledExceptions() @nogc + { + assert(hasExceptionTraps); + version (X86_Any) + return (getControlState() & allExceptions) ^ allExceptions; + else + return (getControlState() & allExceptions); + } + + /// Clear all pending exceptions, then restore the original exception state and rounding mode. + ~this() @nogc + { + clearExceptions(); + if (initialized) + setControlState(savedState); + } + +private: + ControlState savedState; + + bool initialized = false; + + version (ARM_Any) + { + alias ControlState = uint; + } + else version (PPC_Any) + { + alias ControlState = uint; + } + else version (MIPS_Any) + { + alias ControlState = uint; + } + else version (SPARC64) + { + alias ControlState = ulong; + } + else version (SystemZ) + { + alias ControlState = uint; + } + else version (X86_Any) + { + alias ControlState = ushort; + } + else + static assert(false, "Not implemented for this architecture"); + + void initialize() @nogc + { + // BUG: This works around the absence of this() constructors. + if (initialized) return; + clearExceptions(); + savedState = getControlState(); + initialized = true; + } + + // Clear all pending exceptions + static void clearExceptions() @nogc + { + resetIeeeFlags(); + } + + // Read from the control register + static ControlState getControlState() @trusted nothrow @nogc + { + version (GNU) + { + version (X86_Any) + { + ControlState cont; + asm pure nothrow @nogc + { + "fstcw %0" : "=m" cont; + } + return cont; + } + else version (AArch64) + { + asm pure nothrow @nogc + { + "mrs %0, FPCR;" : "=r" cont; + } + return cont; + } + else version (ARM) + { + ControlState cont; + version (ARM_SoftFloat) + cont = 0; + else + { + asm pure nothrow @nogc + { + "vmrs %0, FPSCR" : "=r" cont; + } + } + return cont; + } + else + assert(0, "Not yet supported"); + } + else + version (D_InlineAsm_X86) + { + short cont; + asm nothrow @nogc + { + xor EAX, EAX; + fstcw cont; + } + return cont; + } + else + version (D_InlineAsm_X86_64) + { + short cont; + asm nothrow @nogc + { + xor RAX, RAX; + fstcw cont; + } + return cont; + } + else + assert(0, "Not yet supported"); + } + + // Set the control register + static void setControlState(ControlState newState) @trusted nothrow @nogc + { + version (GNU) + { + version (X86_Any) + { + asm pure nothrow @nogc + { + "fclex; fldcw %0" : : "m" newState; + } + + // Also update MXCSR, SSE's control register. + if (haveSSE) + { + uint mxcsr; + asm pure nothrow @nogc + { + "stmxcsr %0" : "=m" mxcsr; + } + + /* In the FPU control register, rounding mode is in bits 10 and + 11. In MXCSR it's in bits 13 and 14. */ + mxcsr &= ~(roundingMask << 3); // delete old rounding mode + mxcsr |= (newState & roundingMask) << 3; // write new rounding mode + + /* In the FPU control register, masks are bits 0 through 5. + In MXCSR they're 7 through 12. */ + mxcsr &= ~(allExceptions << 7); // delete old masks + mxcsr |= (newState & allExceptions) << 7; // write new exception masks + + asm pure nothrow @nogc + { + "ldmxcsr %0" : : "m" mxcsr; + } + } + } + else version (AArch64) + { + asm pure nothrow @nogc + { + "msr FPCR, %0;" : : "r" (newState); + } + } + else version (ARM) + { + version (ARM_SoftFloat) + return; + else + { + asm pure nothrow @nogc + { + "vmsr FPSCR, %0" : : "r" (newState); + } + } + } + else + assert(0, "Not yet supported"); + } + else + version (InlineAsm_X86_Any) + { + asm nothrow @nogc + { + fclex; + fldcw newState; + } + + // Also update MXCSR, SSE's control register. + if (haveSSE) + { + uint mxcsr; + asm nothrow @nogc { stmxcsr mxcsr; } + + /* In the FPU control register, rounding mode is in bits 10 and + 11. In MXCSR it's in bits 13 and 14. */ + mxcsr &= ~(roundingMask << 3); // delete old rounding mode + mxcsr |= (newState & roundingMask) << 3; // write new rounding mode + + /* In the FPU control register, masks are bits 0 through 5. + In MXCSR they're 7 through 12. */ + mxcsr &= ~(allExceptions << 7); // delete old masks + mxcsr |= (newState & allExceptions) << 7; // write new exception masks + + asm nothrow @nogc { ldmxcsr mxcsr; } + } + } + else + assert(0, "Not yet supported"); + } +} + +@system unittest +{ + // GCC floating point emulation doesn't allow changing + // rounding modes, getting error bits etc + version (GNU) version (D_SoftFloat) + return; + + void ensureDefaults() + { + assert(FloatingPointControl.rounding + == FloatingPointControl.roundToNearest); + if (FloatingPointControl.hasExceptionTraps) + assert(FloatingPointControl.enabledExceptions == 0); + } + + { + FloatingPointControl ctrl; + } + ensureDefaults(); + + version (D_HardFloat) + { + { + FloatingPointControl ctrl; + ctrl.rounding = FloatingPointControl.roundDown; + assert(FloatingPointControl.rounding == FloatingPointControl.roundDown); + } + ensureDefaults(); + } + + if (FloatingPointControl.hasExceptionTraps) + { + FloatingPointControl ctrl; + ctrl.enableExceptions(FloatingPointControl.divByZeroException + | FloatingPointControl.overflowException); + assert(ctrl.enabledExceptions == + (FloatingPointControl.divByZeroException + | FloatingPointControl.overflowException)); + + ctrl.rounding = FloatingPointControl.roundUp; + assert(FloatingPointControl.rounding == FloatingPointControl.roundUp); + } + ensureDefaults(); +} + +@system unittest // rounding +{ + import std.meta : AliasSeq; + + foreach (T; AliasSeq!(float, double, real)) + { + FloatingPointControl fpctrl; + + fpctrl.rounding = FloatingPointControl.roundUp; + T u = 1; + u += 0.1; + + fpctrl.rounding = FloatingPointControl.roundDown; + T d = 1; + d += 0.1; + + fpctrl.rounding = FloatingPointControl.roundToZero; + T z = 1; + z += 0.1; + + assert(u > d); + assert(z == d); + + fpctrl.rounding = FloatingPointControl.roundUp; + u = -1; + u -= 0.1; + + fpctrl.rounding = FloatingPointControl.roundDown; + d = -1; + d -= 0.1; + + fpctrl.rounding = FloatingPointControl.roundToZero; + z = -1; + z -= 0.1; + + assert(u > d); + assert(z == u); + } +} + + +/********************************* + * Determines if $(D_PARAM x) is NaN. + * Params: + * x = a floating point number. + * Returns: + * $(D true) if $(D_PARAM x) is Nan. + */ +bool isNaN(X)(X x) @nogc @trusted pure nothrow +if (isFloatingPoint!(X)) +{ + alias F = floatTraits!(X); + static if (F.realFormat == RealFormat.ieeeSingle) + { + const uint p = *cast(uint *)&x; + return ((p & 0x7F80_0000) == 0x7F80_0000) + && p & 0x007F_FFFF; // not infinity + } + else static if (F.realFormat == RealFormat.ieeeDouble) + { + const ulong p = *cast(ulong *)&x; + return ((p & 0x7FF0_0000_0000_0000) == 0x7FF0_0000_0000_0000) + && p & 0x000F_FFFF_FFFF_FFFF; // not infinity + } + else static if (F.realFormat == RealFormat.ieeeExtended) + { + const ushort e = F.EXPMASK & (cast(ushort *)&x)[F.EXPPOS_SHORT]; + const ulong ps = *cast(ulong *)&x; + return e == F.EXPMASK && + ps & 0x7FFF_FFFF_FFFF_FFFF; // not infinity + } + else static if (F.realFormat == RealFormat.ieeeQuadruple) + { + const ushort e = F.EXPMASK & (cast(ushort *)&x)[F.EXPPOS_SHORT]; + const ulong psLsb = (cast(ulong *)&x)[MANTISSA_LSB]; + const ulong psMsb = (cast(ulong *)&x)[MANTISSA_MSB]; + return e == F.EXPMASK && + (psLsb | (psMsb& 0x0000_FFFF_FFFF_FFFF)) != 0; + } + else + { + return x != x; + } +} + +/// +@safe pure nothrow @nogc unittest +{ + assert( isNaN(float.init)); + assert( isNaN(-double.init)); + assert( isNaN(real.nan)); + assert( isNaN(-real.nan)); + assert(!isNaN(cast(float) 53.6)); + assert(!isNaN(cast(real)-53.6)); +} + +@safe pure nothrow @nogc unittest +{ + import std.meta : AliasSeq; + + foreach (T; AliasSeq!(float, double, real)) + { + // CTFE-able tests + assert(isNaN(T.init)); + assert(isNaN(-T.init)); + assert(isNaN(T.nan)); + assert(isNaN(-T.nan)); + assert(!isNaN(T.infinity)); + assert(!isNaN(-T.infinity)); + assert(!isNaN(cast(T) 53.6)); + assert(!isNaN(cast(T)-53.6)); + + // Runtime tests + shared T f; + f = T.init; + assert(isNaN(f)); + assert(isNaN(-f)); + f = T.nan; + assert(isNaN(f)); + assert(isNaN(-f)); + f = T.infinity; + assert(!isNaN(f)); + assert(!isNaN(-f)); + f = cast(T) 53.6; + assert(!isNaN(f)); + assert(!isNaN(-f)); + } +} + +/********************************* + * Determines if $(D_PARAM x) is finite. + * Params: + * x = a floating point number. + * Returns: + * $(D true) if $(D_PARAM x) is finite. + */ +bool isFinite(X)(X x) @trusted pure nothrow @nogc +{ + alias F = floatTraits!(X); + ushort* pe = cast(ushort *)&x; + return (pe[F.EXPPOS_SHORT] & F.EXPMASK) != F.EXPMASK; +} + +/// +@safe pure nothrow @nogc unittest +{ + assert( isFinite(1.23f)); + assert( isFinite(float.max)); + assert( isFinite(float.min_normal)); + assert(!isFinite(float.nan)); + assert(!isFinite(float.infinity)); +} + +@safe pure nothrow @nogc unittest +{ + assert(isFinite(1.23)); + assert(isFinite(double.max)); + assert(isFinite(double.min_normal)); + assert(!isFinite(double.nan)); + assert(!isFinite(double.infinity)); + + assert(isFinite(1.23L)); + assert(isFinite(real.max)); + assert(isFinite(real.min_normal)); + assert(!isFinite(real.nan)); + assert(!isFinite(real.infinity)); +} + + +/********************************* + * Determines if $(D_PARAM x) is normalized. + * + * A normalized number must not be zero, subnormal, infinite nor $(NAN). + * + * Params: + * x = a floating point number. + * Returns: + * $(D true) if $(D_PARAM x) is normalized. + */ + +/* Need one for each format because subnormal floats might + * be converted to normal reals. + */ +bool isNormal(X)(X x) @trusted pure nothrow @nogc +{ + alias F = floatTraits!(X); + static if (F.realFormat == RealFormat.ibmExtended) + { + // doubledouble is normal if the least significant part is normal. + return isNormal((cast(double*)&x)[MANTISSA_LSB]); + } + else + { + ushort e = F.EXPMASK & (cast(ushort *)&x)[F.EXPPOS_SHORT]; + return (e != F.EXPMASK && e != 0); + } +} + +/// +@safe pure nothrow @nogc unittest +{ + float f = 3; + double d = 500; + real e = 10e+48; + + assert(isNormal(f)); + assert(isNormal(d)); + assert(isNormal(e)); + f = d = e = 0; + assert(!isNormal(f)); + assert(!isNormal(d)); + assert(!isNormal(e)); + assert(!isNormal(real.infinity)); + assert(isNormal(-real.max)); + assert(!isNormal(real.min_normal/4)); + +} + +/********************************* + * Determines if $(D_PARAM x) is subnormal. + * + * Subnormals (also known as "denormal number"), have a 0 exponent + * and a 0 most significant mantissa bit. + * + * Params: + * x = a floating point number. + * Returns: + * $(D true) if $(D_PARAM x) is a denormal number. + */ +bool isSubnormal(X)(X x) @trusted pure nothrow @nogc +{ + /* + Need one for each format because subnormal floats might + be converted to normal reals. + */ + alias F = floatTraits!(X); + static if (F.realFormat == RealFormat.ieeeSingle) + { + uint *p = cast(uint *)&x; + return (*p & F.EXPMASK_INT) == 0 && *p & F.MANTISSAMASK_INT; + } + else static if (F.realFormat == RealFormat.ieeeDouble) + { + uint *p = cast(uint *)&x; + return (p[MANTISSA_MSB] & F.EXPMASK_INT) == 0 + && (p[MANTISSA_LSB] || p[MANTISSA_MSB] & F.MANTISSAMASK_INT); + } + else static if (F.realFormat == RealFormat.ieeeQuadruple) + { + ushort e = F.EXPMASK & (cast(ushort *)&x)[F.EXPPOS_SHORT]; + long* ps = cast(long *)&x; + return (e == 0 && + ((ps[MANTISSA_LSB]|(ps[MANTISSA_MSB]& 0x0000_FFFF_FFFF_FFFF)) != 0)); + } + else static if (F.realFormat == RealFormat.ieeeExtended) + { + ushort* pe = cast(ushort *)&x; + long* ps = cast(long *)&x; + + return (pe[F.EXPPOS_SHORT] & F.EXPMASK) == 0 && *ps > 0; + } + else static if (F.realFormat == RealFormat.ibmExtended) + { + return isSubnormal((cast(double*)&x)[MANTISSA_MSB]); + } + else + { + static assert(false, "Not implemented for this architecture"); + } +} + +/// +@safe pure nothrow @nogc unittest +{ + import std.meta : AliasSeq; + + foreach (T; AliasSeq!(float, double, real)) + { + T f; + for (f = 1.0; !isSubnormal(f); f /= 2) + assert(f != 0); + } +} + +/********************************* + * Determines if $(D_PARAM x) is $(PLUSMN)$(INFIN). + * Params: + * x = a floating point number. + * Returns: + * $(D true) if $(D_PARAM x) is $(PLUSMN)$(INFIN). + */ +bool isInfinity(X)(X x) @nogc @trusted pure nothrow +if (isFloatingPoint!(X)) +{ + alias F = floatTraits!(X); + static if (F.realFormat == RealFormat.ieeeSingle) + { + return ((*cast(uint *)&x) & 0x7FFF_FFFF) == 0x7F80_0000; + } + else static if (F.realFormat == RealFormat.ieeeDouble) + { + return ((*cast(ulong *)&x) & 0x7FFF_FFFF_FFFF_FFFF) + == 0x7FF0_0000_0000_0000; + } + else static if (F.realFormat == RealFormat.ieeeExtended) + { + const ushort e = cast(ushort)(F.EXPMASK & (cast(ushort *)&x)[F.EXPPOS_SHORT]); + const ulong ps = *cast(ulong *)&x; + + // On Motorola 68K, infinity can have hidden bit = 1 or 0. On x86, it is always 1. + return e == F.EXPMASK && (ps & 0x7FFF_FFFF_FFFF_FFFF) == 0; + } + else static if (F.realFormat == RealFormat.ibmExtended) + { + return (((cast(ulong *)&x)[MANTISSA_MSB]) & 0x7FFF_FFFF_FFFF_FFFF) + == 0x7FF8_0000_0000_0000; + } + else static if (F.realFormat == RealFormat.ieeeQuadruple) + { + const long psLsb = (cast(long *)&x)[MANTISSA_LSB]; + const long psMsb = (cast(long *)&x)[MANTISSA_MSB]; + return (psLsb == 0) + && (psMsb & 0x7FFF_FFFF_FFFF_FFFF) == 0x7FFF_0000_0000_0000; + } + else + { + return (x < -X.max) || (X.max < x); + } +} + +/// +@nogc @safe pure nothrow unittest +{ + assert(!isInfinity(float.init)); + assert(!isInfinity(-float.init)); + assert(!isInfinity(float.nan)); + assert(!isInfinity(-float.nan)); + assert(isInfinity(float.infinity)); + assert(isInfinity(-float.infinity)); + assert(isInfinity(-1.0f / 0.0f)); +} + +@safe pure nothrow @nogc unittest +{ + // CTFE-able tests + assert(!isInfinity(double.init)); + assert(!isInfinity(-double.init)); + assert(!isInfinity(double.nan)); + assert(!isInfinity(-double.nan)); + assert(isInfinity(double.infinity)); + assert(isInfinity(-double.infinity)); + assert(isInfinity(-1.0 / 0.0)); + + assert(!isInfinity(real.init)); + assert(!isInfinity(-real.init)); + assert(!isInfinity(real.nan)); + assert(!isInfinity(-real.nan)); + assert(isInfinity(real.infinity)); + assert(isInfinity(-real.infinity)); + assert(isInfinity(-1.0L / 0.0L)); + + // Runtime tests + shared float f; + f = float.init; + assert(!isInfinity(f)); + assert(!isInfinity(-f)); + f = float.nan; + assert(!isInfinity(f)); + assert(!isInfinity(-f)); + f = float.infinity; + assert(isInfinity(f)); + assert(isInfinity(-f)); + f = (-1.0f / 0.0f); + assert(isInfinity(f)); + + shared double d; + d = double.init; + assert(!isInfinity(d)); + assert(!isInfinity(-d)); + d = double.nan; + assert(!isInfinity(d)); + assert(!isInfinity(-d)); + d = double.infinity; + assert(isInfinity(d)); + assert(isInfinity(-d)); + d = (-1.0 / 0.0); + assert(isInfinity(d)); + + shared real e; + e = real.init; + assert(!isInfinity(e)); + assert(!isInfinity(-e)); + e = real.nan; + assert(!isInfinity(e)); + assert(!isInfinity(-e)); + e = real.infinity; + assert(isInfinity(e)); + assert(isInfinity(-e)); + e = (-1.0L / 0.0L); + assert(isInfinity(e)); +} + +/********************************* + * Is the binary representation of x identical to y? + * + * Same as ==, except that positive and negative zero are not identical, + * and two $(NAN)s are identical if they have the same 'payload'. + */ +bool isIdentical(real x, real y) @trusted pure nothrow @nogc +{ + // We're doing a bitwise comparison so the endianness is irrelevant. + long* pxs = cast(long *)&x; + long* pys = cast(long *)&y; + alias F = floatTraits!(real); + static if (F.realFormat == RealFormat.ieeeDouble) + { + return pxs[0] == pys[0]; + } + else static if (F.realFormat == RealFormat.ieeeQuadruple + || F.realFormat == RealFormat.ibmExtended) + { + return pxs[0] == pys[0] && pxs[1] == pys[1]; + } + else + { + ushort* pxe = cast(ushort *)&x; + ushort* pye = cast(ushort *)&y; + return pxe[4] == pye[4] && pxs[0] == pys[0]; + } +} + +/********************************* + * Return 1 if sign bit of e is set, 0 if not. + */ +int signbit(X)(X x) @nogc @trusted pure nothrow +{ + alias F = floatTraits!(X); + return ((cast(ubyte *)&x)[F.SIGNPOS_BYTE] & 0x80) != 0; +} + +/// +@nogc @safe pure nothrow unittest +{ + assert(!signbit(float.nan)); + assert(signbit(-float.nan)); + assert(!signbit(168.1234f)); + assert(signbit(-168.1234f)); + assert(!signbit(0.0f)); + assert(signbit(-0.0f)); + assert(signbit(-float.max)); + assert(!signbit(float.max)); + + assert(!signbit(double.nan)); + assert(signbit(-double.nan)); + assert(!signbit(168.1234)); + assert(signbit(-168.1234)); + assert(!signbit(0.0)); + assert(signbit(-0.0)); + assert(signbit(-double.max)); + assert(!signbit(double.max)); + + assert(!signbit(real.nan)); + assert(signbit(-real.nan)); + assert(!signbit(168.1234L)); + assert(signbit(-168.1234L)); + assert(!signbit(0.0L)); + assert(signbit(-0.0L)); + assert(signbit(-real.max)); + assert(!signbit(real.max)); +} + + +/********************************* + * Return a value composed of to with from's sign bit. + */ +R copysign(R, X)(R to, X from) @trusted pure nothrow @nogc +if (isFloatingPoint!(R) && isFloatingPoint!(X)) +{ + ubyte* pto = cast(ubyte *)&to; + const ubyte* pfrom = cast(ubyte *)&from; + + alias T = floatTraits!(R); + alias F = floatTraits!(X); + pto[T.SIGNPOS_BYTE] &= 0x7F; + pto[T.SIGNPOS_BYTE] |= pfrom[F.SIGNPOS_BYTE] & 0x80; + return to; +} + +// ditto +R copysign(R, X)(X to, R from) @trusted pure nothrow @nogc +if (isIntegral!(X) && isFloatingPoint!(R)) +{ + return copysign(cast(R) to, from); +} + +@safe pure nothrow @nogc unittest +{ + import std.meta : AliasSeq; + + foreach (X; AliasSeq!(float, double, real, int, long)) + { + foreach (Y; AliasSeq!(float, double, real)) + (){ // avoid slow optimizations for large functions @@@BUG@@@ 2396 + X x = 21; + Y y = 23.8; + Y e = void; + + e = copysign(x, y); + assert(e == 21.0); + + e = copysign(-x, y); + assert(e == 21.0); + + e = copysign(x, -y); + assert(e == -21.0); + + e = copysign(-x, -y); + assert(e == -21.0); + + static if (isFloatingPoint!X) + { + e = copysign(X.nan, y); + assert(isNaN(e) && !signbit(e)); + + e = copysign(X.nan, -y); + assert(isNaN(e) && signbit(e)); + } + }(); + } +} + +/********************************* +Returns $(D -1) if $(D x < 0), $(D x) if $(D x == 0), $(D 1) if +$(D x > 0), and $(NAN) if x==$(NAN). + */ +F sgn(F)(F x) @safe pure nothrow @nogc +{ + // @@@TODO@@@: make this faster + return x > 0 ? 1 : x < 0 ? -1 : x; +} + +/// +@safe pure nothrow @nogc unittest +{ + assert(sgn(168.1234) == 1); + assert(sgn(-168.1234) == -1); + assert(sgn(0.0) == 0); + assert(sgn(-0.0) == 0); +} + +// Functions for NaN payloads +/* + * A 'payload' can be stored in the significand of a $(NAN). One bit is required + * to distinguish between a quiet and a signalling $(NAN). This leaves 22 bits + * of payload for a float; 51 bits for a double; 62 bits for an 80-bit real; + * and 111 bits for a 128-bit quad. +*/ +/** + * Create a quiet $(NAN), storing an integer inside the payload. + * + * For floats, the largest possible payload is 0x3F_FFFF. + * For doubles, it is 0x3_FFFF_FFFF_FFFF. + * For 80-bit or 128-bit reals, it is 0x3FFF_FFFF_FFFF_FFFF. + */ +real NaN(ulong payload) @trusted pure nothrow @nogc +{ + alias F = floatTraits!(real); + static if (F.realFormat == RealFormat.ieeeExtended) + { + // real80 (in x86 real format, the implied bit is actually + // not implied but a real bit which is stored in the real) + ulong v = 3; // implied bit = 1, quiet bit = 1 + } + else + { + ulong v = 1; // no implied bit. quiet bit = 1 + } + + ulong a = payload; + + // 22 Float bits + ulong w = a & 0x3F_FFFF; + a -= w; + + v <<=22; + v |= w; + a >>=22; + + // 29 Double bits + v <<=29; + w = a & 0xFFF_FFFF; + v |= w; + a -= w; + a >>=29; + + static if (F.realFormat == RealFormat.ieeeDouble) + { + v |= 0x7FF0_0000_0000_0000; + real x; + * cast(ulong *)(&x) = v; + return x; + } + else + { + v <<=11; + a &= 0x7FF; + v |= a; + real x = real.nan; + + // Extended real bits + static if (F.realFormat == RealFormat.ieeeQuadruple) + { + v <<= 1; // there's no implicit bit + + version (LittleEndian) + { + *cast(ulong*)(6+cast(ubyte*)(&x)) = v; + } + else + { + *cast(ulong*)(2+cast(ubyte*)(&x)) = v; + } + } + else + { + *cast(ulong *)(&x) = v; + } + return x; + } +} + +@system pure nothrow @nogc unittest // not @safe because taking address of local. +{ + static if (floatTraits!(real).realFormat == RealFormat.ieeeDouble) + { + auto x = NaN(1); + auto xl = *cast(ulong*)&x; + assert(xl & 0x8_0000_0000_0000UL); //non-signaling bit, bit 52 + assert((xl & 0x7FF0_0000_0000_0000UL) == 0x7FF0_0000_0000_0000UL); //all exp bits set + } +} + +/** + * Extract an integral payload from a $(NAN). + * + * Returns: + * the integer payload as a ulong. + * + * For floats, the largest possible payload is 0x3F_FFFF. + * For doubles, it is 0x3_FFFF_FFFF_FFFF. + * For 80-bit or 128-bit reals, it is 0x3FFF_FFFF_FFFF_FFFF. + */ +ulong getNaNPayload(real x) @trusted pure nothrow @nogc +{ + // assert(isNaN(x)); + alias F = floatTraits!(real); + static if (F.realFormat == RealFormat.ieeeDouble) + { + ulong m = *cast(ulong *)(&x); + // Make it look like an 80-bit significand. + // Skip exponent, and quiet bit + m &= 0x0007_FFFF_FFFF_FFFF; + m <<= 11; + } + else static if (F.realFormat == RealFormat.ieeeQuadruple) + { + version (LittleEndian) + { + ulong m = *cast(ulong*)(6+cast(ubyte*)(&x)); + } + else + { + ulong m = *cast(ulong*)(2+cast(ubyte*)(&x)); + } + + m >>= 1; // there's no implicit bit + } + else + { + ulong m = *cast(ulong *)(&x); + } + + // ignore implicit bit and quiet bit + + const ulong f = m & 0x3FFF_FF00_0000_0000L; + + ulong w = f >>> 40; + w |= (m & 0x00FF_FFFF_F800L) << (22 - 11); + w |= (m & 0x7FF) << 51; + return w; +} + +debug(UnitTest) +{ + @safe pure nothrow @nogc unittest + { + real nan4 = NaN(0x789_ABCD_EF12_3456); + static if (floatTraits!(real).realFormat == RealFormat.ieeeExtended + || floatTraits!(real).realFormat == RealFormat.ieeeQuadruple) + { + assert(getNaNPayload(nan4) == 0x789_ABCD_EF12_3456); + } + else + { + assert(getNaNPayload(nan4) == 0x1_ABCD_EF12_3456); + } + double nan5 = nan4; + assert(getNaNPayload(nan5) == 0x1_ABCD_EF12_3456); + float nan6 = nan4; + assert(getNaNPayload(nan6) == 0x12_3456); + nan4 = NaN(0xFABCD); + assert(getNaNPayload(nan4) == 0xFABCD); + nan6 = nan4; + assert(getNaNPayload(nan6) == 0xFABCD); + nan5 = NaN(0x100_0000_0000_3456); + assert(getNaNPayload(nan5) == 0x0000_0000_3456); + } +} + +/** + * Calculate the next largest floating point value after x. + * + * Return the least number greater than x that is representable as a real; + * thus, it gives the next point on the IEEE number line. + * + * $(TABLE_SV + * $(SVH x, nextUp(x) ) + * $(SV -$(INFIN), -real.max ) + * $(SV $(PLUSMN)0.0, real.min_normal*real.epsilon ) + * $(SV real.max, $(INFIN) ) + * $(SV $(INFIN), $(INFIN) ) + * $(SV $(NAN), $(NAN) ) + * ) + */ +real nextUp(real x) @trusted pure nothrow @nogc +{ + alias F = floatTraits!(real); + static if (F.realFormat == RealFormat.ieeeDouble) + { + return nextUp(cast(double) x); + } + else static if (F.realFormat == RealFormat.ieeeQuadruple) + { + ushort e = F.EXPMASK & (cast(ushort *)&x)[F.EXPPOS_SHORT]; + if (e == F.EXPMASK) + { + // NaN or Infinity + if (x == -real.infinity) return -real.max; + return x; // +Inf and NaN are unchanged. + } + + auto ps = cast(ulong *)&x; + if (ps[MANTISSA_MSB] & 0x8000_0000_0000_0000) + { + // Negative number + if (ps[MANTISSA_LSB] == 0 && ps[MANTISSA_MSB] == 0x8000_0000_0000_0000) + { + // it was negative zero, change to smallest subnormal + ps[MANTISSA_LSB] = 1; + ps[MANTISSA_MSB] = 0; + return x; + } + if (ps[MANTISSA_LSB] == 0) --ps[MANTISSA_MSB]; + --ps[MANTISSA_LSB]; + } + else + { + // Positive number + ++ps[MANTISSA_LSB]; + if (ps[MANTISSA_LSB] == 0) ++ps[MANTISSA_MSB]; + } + return x; + } + else static if (F.realFormat == RealFormat.ieeeExtended) + { + // For 80-bit reals, the "implied bit" is a nuisance... + ushort *pe = cast(ushort *)&x; + ulong *ps = cast(ulong *)&x; + + if ((pe[F.EXPPOS_SHORT] & F.EXPMASK) == F.EXPMASK) + { + // First, deal with NANs and infinity + if (x == -real.infinity) return -real.max; + return x; // +Inf and NaN are unchanged. + } + if (pe[F.EXPPOS_SHORT] & 0x8000) + { + // Negative number -- need to decrease the significand + --*ps; + // Need to mask with 0x7FFF... so subnormals are treated correctly. + if ((*ps & 0x7FFF_FFFF_FFFF_FFFF) == 0x7FFF_FFFF_FFFF_FFFF) + { + if (pe[F.EXPPOS_SHORT] == 0x8000) // it was negative zero + { + *ps = 1; + pe[F.EXPPOS_SHORT] = 0; // smallest subnormal. + return x; + } + + --pe[F.EXPPOS_SHORT]; + + if (pe[F.EXPPOS_SHORT] == 0x8000) + return x; // it's become a subnormal, implied bit stays low. + + *ps = 0xFFFF_FFFF_FFFF_FFFF; // set the implied bit + return x; + } + return x; + } + else + { + // Positive number -- need to increase the significand. + // Works automatically for positive zero. + ++*ps; + if ((*ps & 0x7FFF_FFFF_FFFF_FFFF) == 0) + { + // change in exponent + ++pe[F.EXPPOS_SHORT]; + *ps = 0x8000_0000_0000_0000; // set the high bit + } + } + return x; + } + else // static if (F.realFormat == RealFormat.ibmExtended) + { + assert(0, "nextUp not implemented"); + } +} + +/** ditto */ +double nextUp(double x) @trusted pure nothrow @nogc +{ + ulong *ps = cast(ulong *)&x; + + if ((*ps & 0x7FF0_0000_0000_0000) == 0x7FF0_0000_0000_0000) + { + // First, deal with NANs and infinity + if (x == -x.infinity) return -x.max; + return x; // +INF and NAN are unchanged. + } + if (*ps & 0x8000_0000_0000_0000) // Negative number + { + if (*ps == 0x8000_0000_0000_0000) // it was negative zero + { + *ps = 0x0000_0000_0000_0001; // change to smallest subnormal + return x; + } + --*ps; + } + else + { // Positive number + ++*ps; + } + return x; +} + +/** ditto */ +float nextUp(float x) @trusted pure nothrow @nogc +{ + uint *ps = cast(uint *)&x; + + if ((*ps & 0x7F80_0000) == 0x7F80_0000) + { + // First, deal with NANs and infinity + if (x == -x.infinity) return -x.max; + + return x; // +INF and NAN are unchanged. + } + if (*ps & 0x8000_0000) // Negative number + { + if (*ps == 0x8000_0000) // it was negative zero + { + *ps = 0x0000_0001; // change to smallest subnormal + return x; + } + + --*ps; + } + else + { + // Positive number + ++*ps; + } + return x; +} + +/** + * Calculate the next smallest floating point value before x. + * + * Return the greatest number less than x that is representable as a real; + * thus, it gives the previous point on the IEEE number line. + * + * $(TABLE_SV + * $(SVH x, nextDown(x) ) + * $(SV $(INFIN), real.max ) + * $(SV $(PLUSMN)0.0, -real.min_normal*real.epsilon ) + * $(SV -real.max, -$(INFIN) ) + * $(SV -$(INFIN), -$(INFIN) ) + * $(SV $(NAN), $(NAN) ) + * ) + */ +real nextDown(real x) @safe pure nothrow @nogc +{ + return -nextUp(-x); +} + +/** ditto */ +double nextDown(double x) @safe pure nothrow @nogc +{ + return -nextUp(-x); +} + +/** ditto */ +float nextDown(float x) @safe pure nothrow @nogc +{ + return -nextUp(-x); +} + +/// +@safe pure nothrow @nogc unittest +{ + assert( nextDown(1.0 + real.epsilon) == 1.0); +} + +@safe pure nothrow @nogc unittest +{ + static if (floatTraits!(real).realFormat == RealFormat.ieeeExtended) + { + + // Tests for 80-bit reals + assert(isIdentical(nextUp(NaN(0xABC)), NaN(0xABC))); + // negative numbers + assert( nextUp(-real.infinity) == -real.max ); + assert( nextUp(-1.0L-real.epsilon) == -1.0 ); + assert( nextUp(-2.0L) == -2.0 + real.epsilon); + // subnormals and zero + assert( nextUp(-real.min_normal) == -real.min_normal*(1-real.epsilon) ); + assert( nextUp(-real.min_normal*(1-real.epsilon)) == -real.min_normal*(1-2*real.epsilon) ); + assert( isIdentical(-0.0L, nextUp(-real.min_normal*real.epsilon)) ); + assert( nextUp(-0.0L) == real.min_normal*real.epsilon ); + assert( nextUp(0.0L) == real.min_normal*real.epsilon ); + assert( nextUp(real.min_normal*(1-real.epsilon)) == real.min_normal ); + assert( nextUp(real.min_normal) == real.min_normal*(1+real.epsilon) ); + // positive numbers + assert( nextUp(1.0L) == 1.0 + real.epsilon ); + assert( nextUp(2.0L-real.epsilon) == 2.0 ); + assert( nextUp(real.max) == real.infinity ); + assert( nextUp(real.infinity)==real.infinity ); + } + + double n = NaN(0xABC); + assert(isIdentical(nextUp(n), n)); + // negative numbers + assert( nextUp(-double.infinity) == -double.max ); + assert( nextUp(-1-double.epsilon) == -1.0 ); + assert( nextUp(-2.0) == -2.0 + double.epsilon); + // subnormals and zero + + assert( nextUp(-double.min_normal) == -double.min_normal*(1-double.epsilon) ); + assert( nextUp(-double.min_normal*(1-double.epsilon)) == -double.min_normal*(1-2*double.epsilon) ); + assert( isIdentical(-0.0, nextUp(-double.min_normal*double.epsilon)) ); + assert( nextUp(0.0) == double.min_normal*double.epsilon ); + assert( nextUp(-0.0) == double.min_normal*double.epsilon ); + assert( nextUp(double.min_normal*(1-double.epsilon)) == double.min_normal ); + assert( nextUp(double.min_normal) == double.min_normal*(1+double.epsilon) ); + // positive numbers + assert( nextUp(1.0) == 1.0 + double.epsilon ); + assert( nextUp(2.0-double.epsilon) == 2.0 ); + assert( nextUp(double.max) == double.infinity ); + + float fn = NaN(0xABC); + assert(isIdentical(nextUp(fn), fn)); + float f = -float.min_normal*(1-float.epsilon); + float f1 = -float.min_normal; + assert( nextUp(f1) == f); + f = 1.0f+float.epsilon; + f1 = 1.0f; + assert( nextUp(f1) == f ); + f1 = -0.0f; + assert( nextUp(f1) == float.min_normal*float.epsilon); + assert( nextUp(float.infinity)==float.infinity ); + + assert(nextDown(1.0L+real.epsilon)==1.0); + assert(nextDown(1.0+double.epsilon)==1.0); + f = 1.0f+float.epsilon; + assert(nextDown(f)==1.0); + assert(nextafter(1.0+real.epsilon, -real.infinity)==1.0); +} + + + +/****************************************** + * Calculates the next representable value after x in the direction of y. + * + * If y > x, the result will be the next largest floating-point value; + * if y < x, the result will be the next smallest value. + * If x == y, the result is y. + * + * Remarks: + * This function is not generally very useful; it's almost always better to use + * the faster functions nextUp() or nextDown() instead. + * + * The FE_INEXACT and FE_OVERFLOW exceptions will be raised if x is finite and + * the function result is infinite. The FE_INEXACT and FE_UNDERFLOW + * exceptions will be raised if the function value is subnormal, and x is + * not equal to y. + */ +T nextafter(T)(const T x, const T y) @safe pure nothrow @nogc +{ + if (x == y) return y; + return ((y>x) ? nextUp(x) : nextDown(x)); +} + +/// +@safe pure nothrow @nogc unittest +{ + float a = 1; + assert(is(typeof(nextafter(a, a)) == float)); + assert(nextafter(a, a.infinity) > a); + + double b = 2; + assert(is(typeof(nextafter(b, b)) == double)); + assert(nextafter(b, b.infinity) > b); + + real c = 3; + assert(is(typeof(nextafter(c, c)) == real)); + assert(nextafter(c, c.infinity) > c); +} + +//real nexttoward(real x, real y) { return core.stdc.math.nexttowardl(x, y); } + +/******************************************* + * Returns the positive difference between x and y. + * Returns: + * $(TABLE_SV + * $(TR $(TH x, y) $(TH fdim(x, y))) + * $(TR $(TD x $(GT) y) $(TD x - y)) + * $(TR $(TD x $(LT)= y) $(TD +0.0)) + * ) + */ +real fdim(real x, real y) @safe pure nothrow @nogc { return (x > y) ? x - y : +0.0; } + +/**************************************** + * Returns the larger of x and y. + */ +real fmax(real x, real y) @safe pure nothrow @nogc { return x > y ? x : y; } + +/**************************************** + * Returns the smaller of x and y. + */ +real fmin(real x, real y) @safe pure nothrow @nogc { return x < y ? x : y; } + +/************************************** + * Returns (x * y) + z, rounding only once according to the + * current rounding mode. + * + * BUGS: Not currently implemented - rounds twice. + */ +real fma(real x, real y, real z) @safe pure nothrow @nogc { return (x * y) + z; } + +/******************************************************************* + * Compute the value of x $(SUPERSCRIPT n), where n is an integer + */ +Unqual!F pow(F, G)(F x, G n) @nogc @trusted pure nothrow +if (isFloatingPoint!(F) && isIntegral!(G)) +{ + import std.traits : Unsigned; + real p = 1.0, v = void; + Unsigned!(Unqual!G) m = n; + if (n < 0) + { + switch (n) + { + case -1: + return 1 / x; + case -2: + return 1 / (x * x); + default: + } + + m = cast(typeof(m))(0 - n); + v = p / x; + } + else + { + switch (n) + { + case 0: + return 1.0; + case 1: + return x; + case 2: + return x * x; + default: + } + + v = x; + } + + while (1) + { + if (m & 1) + p *= v; + m >>= 1; + if (!m) + break; + v *= v; + } + return p; +} + +@safe pure nothrow @nogc unittest +{ + // Make sure it instantiates and works properly on immutable values and + // with various integer and float types. + immutable real x = 46; + immutable float xf = x; + immutable double xd = x; + immutable uint one = 1; + immutable ushort two = 2; + immutable ubyte three = 3; + immutable ulong eight = 8; + + immutable int neg1 = -1; + immutable short neg2 = -2; + immutable byte neg3 = -3; + immutable long neg8 = -8; + + + assert(pow(x,0) == 1.0); + assert(pow(xd,one) == x); + assert(pow(xf,two) == x * x); + assert(pow(x,three) == x * x * x); + assert(pow(x,eight) == (x * x) * (x * x) * (x * x) * (x * x)); + + assert(pow(x, neg1) == 1 / x); + + version (X86_64) + { + pragma(msg, "test disabled on x86_64, see bug 5628"); + } + else version (ARM) + { + pragma(msg, "test disabled on ARM, see bug 5628"); + } + else + { + assert(pow(xd, neg2) == 1 / (x * x)); + assert(pow(xf, neg8) == 1 / ((x * x) * (x * x) * (x * x) * (x * x))); + } + + assert(feqrel(pow(x, neg3), 1 / (x * x * x)) >= real.mant_dig - 1); +} + +@system unittest +{ + assert(equalsDigit(pow(2.0L, 10.0L), 1024, 19)); +} + +/** Compute the value of an integer x, raised to the power of a positive + * integer n. + * + * If both x and n are 0, the result is 1. + * If n is negative, an integer divide error will occur at runtime, + * regardless of the value of x. + */ +typeof(Unqual!(F).init * Unqual!(G).init) pow(F, G)(F x, G n) @nogc @trusted pure nothrow +if (isIntegral!(F) && isIntegral!(G)) +{ + if (n<0) return x/0; // Only support positive powers + typeof(return) p, v = void; + Unqual!G m = n; + + switch (m) + { + case 0: + p = 1; + break; + + case 1: + p = x; + break; + + case 2: + p = x * x; + break; + + default: + v = x; + p = 1; + while (1) + { + if (m & 1) + p *= v; + m >>= 1; + if (!m) + break; + v *= v; + } + break; + } + return p; +} + +/// +@safe pure nothrow @nogc unittest +{ + immutable int one = 1; + immutable byte two = 2; + immutable ubyte three = 3; + immutable short four = 4; + immutable long ten = 10; + + assert(pow(two, three) == 8); + assert(pow(two, ten) == 1024); + assert(pow(one, ten) == 1); + assert(pow(ten, four) == 10_000); + assert(pow(four, 10) == 1_048_576); + assert(pow(three, four) == 81); + +} + +/**Computes integer to floating point powers.*/ +real pow(I, F)(I x, F y) @nogc @trusted pure nothrow +if (isIntegral!I && isFloatingPoint!F) +{ + return pow(cast(real) x, cast(Unqual!F) y); +} + +/********************************************* + * Calculates x$(SUPERSCRIPT y). + * + * $(TABLE_SV + * $(TR $(TH x) $(TH y) $(TH pow(x, y)) + * $(TH div 0) $(TH invalid?)) + * $(TR $(TD anything) $(TD $(PLUSMN)0.0) $(TD 1.0) + * $(TD no) $(TD no) ) + * $(TR $(TD |x| $(GT) 1) $(TD +$(INFIN)) $(TD +$(INFIN)) + * $(TD no) $(TD no) ) + * $(TR $(TD |x| $(LT) 1) $(TD +$(INFIN)) $(TD +0.0) + * $(TD no) $(TD no) ) + * $(TR $(TD |x| $(GT) 1) $(TD -$(INFIN)) $(TD +0.0) + * $(TD no) $(TD no) ) + * $(TR $(TD |x| $(LT) 1) $(TD -$(INFIN)) $(TD +$(INFIN)) + * $(TD no) $(TD no) ) + * $(TR $(TD +$(INFIN)) $(TD $(GT) 0.0) $(TD +$(INFIN)) + * $(TD no) $(TD no) ) + * $(TR $(TD +$(INFIN)) $(TD $(LT) 0.0) $(TD +0.0) + * $(TD no) $(TD no) ) + * $(TR $(TD -$(INFIN)) $(TD odd integer $(GT) 0.0) $(TD -$(INFIN)) + * $(TD no) $(TD no) ) + * $(TR $(TD -$(INFIN)) $(TD $(GT) 0.0, not odd integer) $(TD +$(INFIN)) + * $(TD no) $(TD no)) + * $(TR $(TD -$(INFIN)) $(TD odd integer $(LT) 0.0) $(TD -0.0) + * $(TD no) $(TD no) ) + * $(TR $(TD -$(INFIN)) $(TD $(LT) 0.0, not odd integer) $(TD +0.0) + * $(TD no) $(TD no) ) + * $(TR $(TD $(PLUSMN)1.0) $(TD $(PLUSMN)$(INFIN)) $(TD $(NAN)) + * $(TD no) $(TD yes) ) + * $(TR $(TD $(LT) 0.0) $(TD finite, nonintegral) $(TD $(NAN)) + * $(TD no) $(TD yes)) + * $(TR $(TD $(PLUSMN)0.0) $(TD odd integer $(LT) 0.0) $(TD $(PLUSMNINF)) + * $(TD yes) $(TD no) ) + * $(TR $(TD $(PLUSMN)0.0) $(TD $(LT) 0.0, not odd integer) $(TD +$(INFIN)) + * $(TD yes) $(TD no)) + * $(TR $(TD $(PLUSMN)0.0) $(TD odd integer $(GT) 0.0) $(TD $(PLUSMN)0.0) + * $(TD no) $(TD no) ) + * $(TR $(TD $(PLUSMN)0.0) $(TD $(GT) 0.0, not odd integer) $(TD +0.0) + * $(TD no) $(TD no) ) + * ) + */ +Unqual!(Largest!(F, G)) pow(F, G)(F x, G y) @nogc @trusted pure nothrow +if (isFloatingPoint!(F) && isFloatingPoint!(G)) +{ + alias Float = typeof(return); + + static real impl(real x, real y) @nogc pure nothrow + { + // Special cases. + if (isNaN(y)) + return y; + if (isNaN(x) && y != 0.0) + return x; + + // Even if x is NaN. + if (y == 0.0) + return 1.0; + if (y == 1.0) + return x; + + if (isInfinity(y)) + { + if (fabs(x) > 1) + { + if (signbit(y)) + return +0.0; + else + return F.infinity; + } + else if (fabs(x) == 1) + { + return y * 0; // generate NaN. + } + else // < 1 + { + if (signbit(y)) + return F.infinity; + else + return +0.0; + } + } + if (isInfinity(x)) + { + if (signbit(x)) + { + long i = cast(long) y; + if (y > 0.0) + { + if (i == y && i & 1) + return -F.infinity; + else + return F.infinity; + } + else if (y < 0.0) + { + if (i == y && i & 1) + return -0.0; + else + return +0.0; + } + } + else + { + if (y > 0.0) + return F.infinity; + else if (y < 0.0) + return +0.0; + } + } + + if (x == 0.0) + { + if (signbit(x)) + { + long i = cast(long) y; + if (y > 0.0) + { + if (i == y && i & 1) + return -0.0; + else + return +0.0; + } + else if (y < 0.0) + { + if (i == y && i & 1) + return -F.infinity; + else + return F.infinity; + } + } + else + { + if (y > 0.0) + return +0.0; + else if (y < 0.0) + return F.infinity; + } + } + if (x == 1.0) + return 1.0; + + if (y >= F.max) + { + if ((x > 0.0 && x < 1.0) || (x > -1.0 && x < 0.0)) + return 0.0; + if (x > 1.0 || x < -1.0) + return F.infinity; + } + if (y <= -F.max) + { + if ((x > 0.0 && x < 1.0) || (x > -1.0 && x < 0.0)) + return F.infinity; + if (x > 1.0 || x < -1.0) + return 0.0; + } + + if (x >= F.max) + { + if (y > 0.0) + return F.infinity; + else + return 0.0; + } + if (x <= -F.max) + { + long i = cast(long) y; + if (y > 0.0) + { + if (i == y && i & 1) + return -F.infinity; + else + return F.infinity; + } + else if (y < 0.0) + { + if (i == y && i & 1) + return -0.0; + else + return +0.0; + } + } + + // Integer power of x. + long iy = cast(long) y; + if (iy == y && fabs(y) < 32_768.0) + return pow(x, iy); + + real sign = 1.0; + if (x < 0) + { + // Result is real only if y is an integer + // Check for a non-zero fractional part + enum maxOdd = pow(2.0L, real.mant_dig) - 1.0L; + static if (maxOdd > ulong.max) + { + // Generic method, for any FP type + if (floor(y) != y) + return sqrt(x); // Complex result -- create a NaN + + const hy = ldexp(y, -1); + if (floor(hy) != hy) + sign = -1.0; + } + else + { + // Much faster, if ulong has enough precision + const absY = fabs(y); + if (absY <= maxOdd) + { + const uy = cast(ulong) absY; + if (uy != absY) + return sqrt(x); // Complex result -- create a NaN + + if (uy & 1) + sign = -1.0; + } + } + x = -x; + } + version (INLINE_YL2X) + { + // If x > 0, x ^^ y == 2 ^^ ( y * log2(x) ) + // TODO: This is not accurate in practice. A fast and accurate + // (though complicated) method is described in: + // "An efficient rounding boundary test for pow(x, y) + // in double precision", C.Q. Lauter and V. Lefèvre, INRIA (2007). + return sign * exp2( core.math.yl2x(x, y) ); + } + else + { + // If x > 0, x ^^ y == 2 ^^ ( y * log2(x) ) + // TODO: This is not accurate in practice. A fast and accurate + // (though complicated) method is described in: + // "An efficient rounding boundary test for pow(x, y) + // in double precision", C.Q. Lauter and V. Lefèvre, INRIA (2007). + Float w = exp2(y * log2(x)); + return sign * w; + } + } + return impl(x, y); +} + +@safe pure nothrow @nogc unittest +{ + // Test all the special values. These unittests can be run on Windows + // by temporarily changing the version (linux) to version (all). + immutable float zero = 0; + immutable real one = 1; + immutable double two = 2; + immutable float three = 3; + immutable float fnan = float.nan; + immutable double dnan = double.nan; + immutable real rnan = real.nan; + immutable dinf = double.infinity; + immutable rninf = -real.infinity; + + assert(pow(fnan, zero) == 1); + assert(pow(dnan, zero) == 1); + assert(pow(rnan, zero) == 1); + + assert(pow(two, dinf) == double.infinity); + assert(isIdentical(pow(0.2f, dinf), +0.0)); + assert(pow(0.99999999L, rninf) == real.infinity); + assert(isIdentical(pow(1.000000001, rninf), +0.0)); + assert(pow(dinf, 0.001) == dinf); + assert(isIdentical(pow(dinf, -0.001), +0.0)); + assert(pow(rninf, 3.0L) == rninf); + assert(pow(rninf, 2.0L) == real.infinity); + assert(isIdentical(pow(rninf, -3.0), -0.0)); + assert(isIdentical(pow(rninf, -2.0), +0.0)); + + // @@@BUG@@@ somewhere + version (OSX) {} else assert(isNaN(pow(one, dinf))); + version (OSX) {} else assert(isNaN(pow(-one, dinf))); + assert(isNaN(pow(-0.2, PI))); + // boundary cases. Note that epsilon == 2^^-n for some n, + // so 1/epsilon == 2^^n is always even. + assert(pow(-1.0L, 1/real.epsilon - 1.0L) == -1.0L); + assert(pow(-1.0L, 1/real.epsilon) == 1.0L); + assert(isNaN(pow(-1.0L, 1/real.epsilon-0.5L))); + assert(isNaN(pow(-1.0L, -1/real.epsilon+0.5L))); + + assert(pow(0.0, -3.0) == double.infinity); + assert(pow(-0.0, -3.0) == -double.infinity); + assert(pow(0.0, -PI) == double.infinity); + assert(pow(-0.0, -PI) == double.infinity); + assert(isIdentical(pow(0.0, 5.0), 0.0)); + assert(isIdentical(pow(-0.0, 5.0), -0.0)); + assert(isIdentical(pow(0.0, 6.0), 0.0)); + assert(isIdentical(pow(-0.0, 6.0), 0.0)); + + // Issue #14786 fixed + immutable real maxOdd = pow(2.0L, real.mant_dig) - 1.0L; + assert(pow(-1.0L, maxOdd) == -1.0L); + assert(pow(-1.0L, -maxOdd) == -1.0L); + assert(pow(-1.0L, maxOdd + 1.0L) == 1.0L); + assert(pow(-1.0L, -maxOdd + 1.0L) == 1.0L); + assert(pow(-1.0L, maxOdd - 1.0L) == 1.0L); + assert(pow(-1.0L, -maxOdd - 1.0L) == 1.0L); + + // Now, actual numbers. + assert(approxEqual(pow(two, three), 8.0)); + assert(approxEqual(pow(two, -2.5), 0.1767767)); + + // Test integer to float power. + immutable uint twoI = 2; + assert(approxEqual(pow(twoI, three), 8.0)); +} + +/************************************** + * To what precision is x equal to y? + * + * Returns: the number of mantissa bits which are equal in x and y. + * eg, 0x1.F8p+60 and 0x1.F1p+60 are equal to 5 bits of precision. + * + * $(TABLE_SV + * $(TR $(TH x) $(TH y) $(TH feqrel(x, y))) + * $(TR $(TD x) $(TD x) $(TD real.mant_dig)) + * $(TR $(TD x) $(TD $(GT)= 2*x) $(TD 0)) + * $(TR $(TD x) $(TD $(LT)= x/2) $(TD 0)) + * $(TR $(TD $(NAN)) $(TD any) $(TD 0)) + * $(TR $(TD any) $(TD $(NAN)) $(TD 0)) + * ) + */ +int feqrel(X)(const X x, const X y) @trusted pure nothrow @nogc +if (isFloatingPoint!(X)) +{ + /* Public Domain. Author: Don Clugston, 18 Aug 2005. + */ + alias F = floatTraits!(X); + static if (F.realFormat == RealFormat.ibmExtended) + { + if (cast(double*)(&x)[MANTISSA_MSB] == cast(double*)(&y)[MANTISSA_MSB]) + { + return double.mant_dig + + feqrel(cast(double*)(&x)[MANTISSA_LSB], + cast(double*)(&y)[MANTISSA_LSB]); + } + else + { + return feqrel(cast(double*)(&x)[MANTISSA_MSB], + cast(double*)(&y)[MANTISSA_MSB]); + } + } + else + { + static assert(F.realFormat == RealFormat.ieeeSingle + || F.realFormat == RealFormat.ieeeDouble + || F.realFormat == RealFormat.ieeeExtended + || F.realFormat == RealFormat.ieeeQuadruple); + + if (x == y) + return X.mant_dig; // ensure diff != 0, cope with INF. + + Unqual!X diff = fabs(x - y); + + ushort *pa = cast(ushort *)(&x); + ushort *pb = cast(ushort *)(&y); + ushort *pd = cast(ushort *)(&diff); + + + // The difference in abs(exponent) between x or y and abs(x-y) + // is equal to the number of significand bits of x which are + // equal to y. If negative, x and y have different exponents. + // If positive, x and y are equal to 'bitsdiff' bits. + // AND with 0x7FFF to form the absolute value. + // To avoid out-by-1 errors, we subtract 1 so it rounds down + // if the exponents were different. This means 'bitsdiff' is + // always 1 lower than we want, except that if bitsdiff == 0, + // they could have 0 or 1 bits in common. + + int bitsdiff = ((( (pa[F.EXPPOS_SHORT] & F.EXPMASK) + + (pb[F.EXPPOS_SHORT] & F.EXPMASK) + - (1 << F.EXPSHIFT)) >> 1) + - (pd[F.EXPPOS_SHORT] & F.EXPMASK)) >> F.EXPSHIFT; + if ( (pd[F.EXPPOS_SHORT] & F.EXPMASK) == 0) + { // Difference is subnormal + // For subnormals, we need to add the number of zeros that + // lie at the start of diff's significand. + // We do this by multiplying by 2^^real.mant_dig + diff *= F.RECIP_EPSILON; + return bitsdiff + X.mant_dig - ((pd[F.EXPPOS_SHORT] & F.EXPMASK) >> F.EXPSHIFT); + } + + if (bitsdiff > 0) + return bitsdiff + 1; // add the 1 we subtracted before + + // Avoid out-by-1 errors when factor is almost 2. + if (bitsdiff == 0 + && ((pa[F.EXPPOS_SHORT] ^ pb[F.EXPPOS_SHORT]) & F.EXPMASK) == 0) + { + return 1; + } else return 0; + } +} + +@safe pure nothrow @nogc unittest +{ + void testFeqrel(F)() + { + // Exact equality + assert(feqrel(F.max, F.max) == F.mant_dig); + assert(feqrel!(F)(0.0, 0.0) == F.mant_dig); + assert(feqrel(F.infinity, F.infinity) == F.mant_dig); + + // a few bits away from exact equality + F w=1; + for (int i = 1; i < F.mant_dig - 1; ++i) + { + assert(feqrel!(F)(1.0 + w * F.epsilon, 1.0) == F.mant_dig-i); + assert(feqrel!(F)(1.0 - w * F.epsilon, 1.0) == F.mant_dig-i); + assert(feqrel!(F)(1.0, 1 + (w-1) * F.epsilon) == F.mant_dig - i + 1); + w*=2; + } + + assert(feqrel!(F)(1.5+F.epsilon, 1.5) == F.mant_dig-1); + assert(feqrel!(F)(1.5-F.epsilon, 1.5) == F.mant_dig-1); + assert(feqrel!(F)(1.5-F.epsilon, 1.5+F.epsilon) == F.mant_dig-2); + + + // Numbers that are close + assert(feqrel!(F)(0x1.Bp+84, 0x1.B8p+84) == 5); + assert(feqrel!(F)(0x1.8p+10, 0x1.Cp+10) == 2); + assert(feqrel!(F)(1.5 * (1 - F.epsilon), 1.0L) == 2); + assert(feqrel!(F)(1.5, 1.0) == 1); + assert(feqrel!(F)(2 * (1 - F.epsilon), 1.0L) == 1); + + // Factors of 2 + assert(feqrel(F.max, F.infinity) == 0); + assert(feqrel!(F)(2 * (1 - F.epsilon), 1.0L) == 1); + assert(feqrel!(F)(1.0, 2.0) == 0); + assert(feqrel!(F)(4.0, 1.0) == 0); + + // Extreme inequality + assert(feqrel(F.nan, F.nan) == 0); + assert(feqrel!(F)(0.0L, -F.nan) == 0); + assert(feqrel(F.nan, F.infinity) == 0); + assert(feqrel(F.infinity, -F.infinity) == 0); + assert(feqrel(F.max, -F.max) == 0); + + assert(feqrel(F.min_normal / 8, F.min_normal / 17) == 3); + + const F Const = 2; + immutable F Immutable = 2; + auto Compiles = feqrel(Const, Immutable); + } + + assert(feqrel(7.1824L, 7.1824L) == real.mant_dig); + + testFeqrel!(real)(); + testFeqrel!(double)(); + testFeqrel!(float)(); +} + +package: // Not public yet +/* Return the value that lies halfway between x and y on the IEEE number line. + * + * Formally, the result is the arithmetic mean of the binary significands of x + * and y, multiplied by the geometric mean of the binary exponents of x and y. + * x and y must have the same sign, and must not be NaN. + * Note: this function is useful for ensuring O(log n) behaviour in algorithms + * involving a 'binary chop'. + * + * Special cases: + * If x and y are within a factor of 2, (ie, feqrel(x, y) > 0), the return value + * is the arithmetic mean (x + y) / 2. + * If x and y are even powers of 2, the return value is the geometric mean, + * ieeeMean(x, y) = sqrt(x * y). + * + */ +T ieeeMean(T)(const T x, const T y) @trusted pure nothrow @nogc +in +{ + // both x and y must have the same sign, and must not be NaN. + assert(signbit(x) == signbit(y)); + assert(x == x && y == y); +} +body +{ + // Runtime behaviour for contract violation: + // If signs are opposite, or one is a NaN, return 0. + if (!((x >= 0 && y >= 0) || (x <= 0 && y <= 0))) return 0.0; + + // The implementation is simple: cast x and y to integers, + // average them (avoiding overflow), and cast the result back to a floating-point number. + + alias F = floatTraits!(T); + T u; + static if (F.realFormat == RealFormat.ieeeExtended) + { + // There's slight additional complexity because they are actually + // 79-bit reals... + ushort *ue = cast(ushort *)&u; + ulong *ul = cast(ulong *)&u; + ushort *xe = cast(ushort *)&x; + ulong *xl = cast(ulong *)&x; + ushort *ye = cast(ushort *)&y; + ulong *yl = cast(ulong *)&y; + + // Ignore the useless implicit bit. (Bonus: this prevents overflows) + ulong m = ((*xl) & 0x7FFF_FFFF_FFFF_FFFFL) + ((*yl) & 0x7FFF_FFFF_FFFF_FFFFL); + + // @@@ BUG? @@@ + // Cast shouldn't be here + ushort e = cast(ushort) ((xe[F.EXPPOS_SHORT] & F.EXPMASK) + + (ye[F.EXPPOS_SHORT] & F.EXPMASK)); + if (m & 0x8000_0000_0000_0000L) + { + ++e; + m &= 0x7FFF_FFFF_FFFF_FFFFL; + } + // Now do a multi-byte right shift + const uint c = e & 1; // carry + e >>= 1; + m >>>= 1; + if (c) + m |= 0x4000_0000_0000_0000L; // shift carry into significand + if (e) + *ul = m | 0x8000_0000_0000_0000L; // set implicit bit... + else + *ul = m; // ... unless exponent is 0 (subnormal or zero). + + ue[4]= e | (xe[F.EXPPOS_SHORT]& 0x8000); // restore sign bit + } + else static if (F.realFormat == RealFormat.ieeeQuadruple) + { + // This would be trivial if 'ucent' were implemented... + ulong *ul = cast(ulong *)&u; + ulong *xl = cast(ulong *)&x; + ulong *yl = cast(ulong *)&y; + + // Multi-byte add, then multi-byte right shift. + import core.checkedint : addu; + bool carry; + ulong ml = addu(xl[MANTISSA_LSB], yl[MANTISSA_LSB], carry); + + ulong mh = carry + (xl[MANTISSA_MSB] & 0x7FFF_FFFF_FFFF_FFFFL) + + (yl[MANTISSA_MSB] & 0x7FFF_FFFF_FFFF_FFFFL); + + ul[MANTISSA_MSB] = (mh >>> 1) | (xl[MANTISSA_MSB] & 0x8000_0000_0000_0000); + ul[MANTISSA_LSB] = (ml >>> 1) | (mh & 1) << 63; + } + else static if (F.realFormat == RealFormat.ieeeDouble) + { + ulong *ul = cast(ulong *)&u; + ulong *xl = cast(ulong *)&x; + ulong *yl = cast(ulong *)&y; + ulong m = (((*xl) & 0x7FFF_FFFF_FFFF_FFFFL) + + ((*yl) & 0x7FFF_FFFF_FFFF_FFFFL)) >>> 1; + m |= ((*xl) & 0x8000_0000_0000_0000L); + *ul = m; + } + else static if (F.realFormat == RealFormat.ieeeSingle) + { + uint *ul = cast(uint *)&u; + uint *xl = cast(uint *)&x; + uint *yl = cast(uint *)&y; + uint m = (((*xl) & 0x7FFF_FFFF) + ((*yl) & 0x7FFF_FFFF)) >>> 1; + m |= ((*xl) & 0x8000_0000); + *ul = m; + } + else + { + assert(0, "Not implemented"); + } + return u; +} + +@safe pure nothrow @nogc unittest +{ + assert(ieeeMean(-0.0,-1e-20)<0); + assert(ieeeMean(0.0,1e-20)>0); + + assert(ieeeMean(1.0L,4.0L)==2L); + assert(ieeeMean(2.0*1.013,8.0*1.013)==4*1.013); + assert(ieeeMean(-1.0L,-4.0L)==-2L); + assert(ieeeMean(-1.0,-4.0)==-2); + assert(ieeeMean(-1.0f,-4.0f)==-2f); + assert(ieeeMean(-1.0,-2.0)==-1.5); + assert(ieeeMean(-1*(1+8*real.epsilon),-2*(1+8*real.epsilon)) + ==-1.5*(1+5*real.epsilon)); + assert(ieeeMean(0x1p60,0x1p-10)==0x1p25); + + static if (floatTraits!(real).realFormat == RealFormat.ieeeExtended) + { + assert(ieeeMean(1.0L,real.infinity)==0x1p8192L); + assert(ieeeMean(0.0L,real.infinity)==1.5); + } + assert(ieeeMean(0.5*real.min_normal*(1-4*real.epsilon),0.5*real.min_normal) + == 0.5*real.min_normal*(1-2*real.epsilon)); +} + +public: + + +/*********************************** + * Evaluate polynomial A(x) = $(SUB a, 0) + $(SUB a, 1)x + $(SUB a, 2)$(POWER x,2) + * + $(SUB a,3)$(POWER x,3); ... + * + * Uses Horner's rule A(x) = $(SUB a, 0) + x($(SUB a, 1) + x($(SUB a, 2) + * + x($(SUB a, 3) + ...))) + * Params: + * x = the value to evaluate. + * A = array of coefficients $(SUB a, 0), $(SUB a, 1), etc. + */ +Unqual!(CommonType!(T1, T2)) poly(T1, T2)(T1 x, in T2[] A) @trusted pure nothrow @nogc +if (isFloatingPoint!T1 && isFloatingPoint!T2) +in +{ + assert(A.length > 0); +} +body +{ + static if (is(Unqual!T2 == real)) + { + return polyImpl(x, A); + } + else + { + return polyImplBase(x, A); + } +} + +/// +@safe nothrow @nogc unittest +{ + real x = 3.1; + static real[] pp = [56.1, 32.7, 6]; + + assert(poly(x, pp) == (56.1L + (32.7L + 6.0L * x) * x)); +} + +@safe nothrow @nogc unittest +{ + double x = 3.1; + static double[] pp = [56.1, 32.7, 6]; + double y = x; + y *= 6.0; + y += 32.7; + y *= x; + y += 56.1; + assert(poly(x, pp) == y); +} + +@safe unittest +{ + static assert(poly(3.0, [1.0, 2.0, 3.0]) == 34); +} + +private Unqual!(CommonType!(T1, T2)) polyImplBase(T1, T2)(T1 x, in T2[] A) @trusted pure nothrow @nogc +if (isFloatingPoint!T1 && isFloatingPoint!T2) +{ + ptrdiff_t i = A.length - 1; + typeof(return) r = A[i]; + while (--i >= 0) + { + r *= x; + r += A[i]; + } + return r; +} + +private real polyImpl(real x, in real[] A) @trusted pure nothrow @nogc +{ + version (D_InlineAsm_X86) + { + if (__ctfe) + { + return polyImplBase(x, A); + } + version (Windows) + { + // BUG: This code assumes a frame pointer in EBP. + asm pure nothrow @nogc // assembler by W. Bright + { + // EDX = (A.length - 1) * real.sizeof + mov ECX,A[EBP] ; // ECX = A.length + dec ECX ; + lea EDX,[ECX][ECX*8] ; + add EDX,ECX ; + add EDX,A+4[EBP] ; + fld real ptr [EDX] ; // ST0 = coeff[ECX] + jecxz return_ST ; + fld x[EBP] ; // ST0 = x + fxch ST(1) ; // ST1 = x, ST0 = r + align 4 ; + L2: fmul ST,ST(1) ; // r *= x + fld real ptr -10[EDX] ; + sub EDX,10 ; // deg-- + faddp ST(1),ST ; + dec ECX ; + jne L2 ; + fxch ST(1) ; // ST1 = r, ST0 = x + fstp ST(0) ; // dump x + align 4 ; + return_ST: ; + ; + } + } + else version (linux) + { + asm pure nothrow @nogc // assembler by W. Bright + { + // EDX = (A.length - 1) * real.sizeof + mov ECX,A[EBP] ; // ECX = A.length + dec ECX ; + lea EDX,[ECX*8] ; + lea EDX,[EDX][ECX*4] ; + add EDX,A+4[EBP] ; + fld real ptr [EDX] ; // ST0 = coeff[ECX] + jecxz return_ST ; + fld x[EBP] ; // ST0 = x + fxch ST(1) ; // ST1 = x, ST0 = r + align 4 ; + L2: fmul ST,ST(1) ; // r *= x + fld real ptr -12[EDX] ; + sub EDX,12 ; // deg-- + faddp ST(1),ST ; + dec ECX ; + jne L2 ; + fxch ST(1) ; // ST1 = r, ST0 = x + fstp ST(0) ; // dump x + align 4 ; + return_ST: ; + ; + } + } + else version (OSX) + { + asm pure nothrow @nogc // assembler by W. Bright + { + // EDX = (A.length - 1) * real.sizeof + mov ECX,A[EBP] ; // ECX = A.length + dec ECX ; + lea EDX,[ECX*8] ; + add EDX,EDX ; + add EDX,A+4[EBP] ; + fld real ptr [EDX] ; // ST0 = coeff[ECX] + jecxz return_ST ; + fld x[EBP] ; // ST0 = x + fxch ST(1) ; // ST1 = x, ST0 = r + align 4 ; + L2: fmul ST,ST(1) ; // r *= x + fld real ptr -16[EDX] ; + sub EDX,16 ; // deg-- + faddp ST(1),ST ; + dec ECX ; + jne L2 ; + fxch ST(1) ; // ST1 = r, ST0 = x + fstp ST(0) ; // dump x + align 4 ; + return_ST: ; + ; + } + } + else version (FreeBSD) + { + asm pure nothrow @nogc // assembler by W. Bright + { + // EDX = (A.length - 1) * real.sizeof + mov ECX,A[EBP] ; // ECX = A.length + dec ECX ; + lea EDX,[ECX*8] ; + lea EDX,[EDX][ECX*4] ; + add EDX,A+4[EBP] ; + fld real ptr [EDX] ; // ST0 = coeff[ECX] + jecxz return_ST ; + fld x[EBP] ; // ST0 = x + fxch ST(1) ; // ST1 = x, ST0 = r + align 4 ; + L2: fmul ST,ST(1) ; // r *= x + fld real ptr -12[EDX] ; + sub EDX,12 ; // deg-- + faddp ST(1),ST ; + dec ECX ; + jne L2 ; + fxch ST(1) ; // ST1 = r, ST0 = x + fstp ST(0) ; // dump x + align 4 ; + return_ST: ; + ; + } + } + else version (Solaris) + { + asm pure nothrow @nogc // assembler by W. Bright + { + // EDX = (A.length - 1) * real.sizeof + mov ECX,A[EBP] ; // ECX = A.length + dec ECX ; + lea EDX,[ECX*8] ; + lea EDX,[EDX][ECX*4] ; + add EDX,A+4[EBP] ; + fld real ptr [EDX] ; // ST0 = coeff[ECX] + jecxz return_ST ; + fld x[EBP] ; // ST0 = x + fxch ST(1) ; // ST1 = x, ST0 = r + align 4 ; + L2: fmul ST,ST(1) ; // r *= x + fld real ptr -12[EDX] ; + sub EDX,12 ; // deg-- + faddp ST(1),ST ; + dec ECX ; + jne L2 ; + fxch ST(1) ; // ST1 = r, ST0 = x + fstp ST(0) ; // dump x + align 4 ; + return_ST: ; + ; + } + } + else + { + static assert(0); + } + } + else + { + return polyImplBase(x, A); + } +} + + +/** + Computes whether two values are approximately equal, admitting a maximum + relative difference, and a maximum absolute difference. + + Params: + lhs = First item to compare. + rhs = Second item to compare. + maxRelDiff = Maximum allowable difference relative to `rhs`. + maxAbsDiff = Maximum absolute difference. + + Returns: + `true` if the two items are approximately equal under either criterium. + If one item is a range, and the other is a single value, then the result + is the logical and-ing of calling `approxEqual` on each element of the + ranged item against the single item. If both items are ranges, then + `approxEqual` returns `true` if and only if the ranges have the same + number of elements and if `approxEqual` evaluates to `true` for each + pair of elements. + */ +bool approxEqual(T, U, V)(T lhs, U rhs, V maxRelDiff, V maxAbsDiff = 1e-5) +{ + import std.range.primitives : empty, front, isInputRange, popFront; + static if (isInputRange!T) + { + static if (isInputRange!U) + { + // Two ranges + for (;; lhs.popFront(), rhs.popFront()) + { + if (lhs.empty) return rhs.empty; + if (rhs.empty) return lhs.empty; + if (!approxEqual(lhs.front, rhs.front, maxRelDiff, maxAbsDiff)) + return false; + } + } + else static if (isIntegral!U) + { + // convert rhs to real + return approxEqual(lhs, real(rhs), maxRelDiff, maxAbsDiff); + } + else + { + // lhs is range, rhs is number + for (; !lhs.empty; lhs.popFront()) + { + if (!approxEqual(lhs.front, rhs, maxRelDiff, maxAbsDiff)) + return false; + } + return true; + } + } + else + { + static if (isInputRange!U) + { + // lhs is number, rhs is range + for (; !rhs.empty; rhs.popFront()) + { + if (!approxEqual(lhs, rhs.front, maxRelDiff, maxAbsDiff)) + return false; + } + return true; + } + else static if (isIntegral!T || isIntegral!U) + { + // convert both lhs and rhs to real + return approxEqual(real(lhs), real(rhs), maxRelDiff, maxAbsDiff); + } + else + { + // two numbers + //static assert(is(T : real) && is(U : real)); + if (rhs == 0) + { + return fabs(lhs) <= maxAbsDiff; + } + static if (is(typeof(lhs.infinity)) && is(typeof(rhs.infinity))) + { + if (lhs == lhs.infinity && rhs == rhs.infinity || + lhs == -lhs.infinity && rhs == -rhs.infinity) return true; + } + return fabs((lhs - rhs) / rhs) <= maxRelDiff + || maxAbsDiff != 0 && fabs(lhs - rhs) <= maxAbsDiff; + } + } +} + +/** + Returns $(D approxEqual(lhs, rhs, 1e-2, 1e-5)). + */ +bool approxEqual(T, U)(T lhs, U rhs) +{ + return approxEqual(lhs, rhs, 1e-2, 1e-5); +} + +/// +@safe pure nothrow unittest +{ + assert(approxEqual(1.0, 1.0099)); + assert(!approxEqual(1.0, 1.011)); + float[] arr1 = [ 1.0, 2.0, 3.0 ]; + double[] arr2 = [ 1.001, 1.999, 3 ]; + assert(approxEqual(arr1, arr2)); + + real num = real.infinity; + assert(num == real.infinity); // Passes. + assert(approxEqual(num, real.infinity)); // Fails. + num = -real.infinity; + assert(num == -real.infinity); // Passes. + assert(approxEqual(num, -real.infinity)); // Fails. + + assert(!approxEqual(3, 0)); + assert(approxEqual(3, 3)); + assert(approxEqual(3.0, 3)); + assert(approxEqual([3, 3, 3], 3.0)); + assert(approxEqual([3.0, 3.0, 3.0], 3)); + int a = 10; + assert(approxEqual(10, a)); +} + +@safe pure nothrow @nogc unittest +{ + real num = real.infinity; + assert(num == real.infinity); // Passes. + assert(approxEqual(num, real.infinity)); // Fails. +} + + +@safe pure nothrow @nogc unittest +{ + float f = sqrt(2.0f); + assert(fabs(f * f - 2.0f) < .00001); + + double d = sqrt(2.0); + assert(fabs(d * d - 2.0) < .00001); + + real r = sqrt(2.0L); + assert(fabs(r * r - 2.0) < .00001); +} + +@safe pure nothrow @nogc unittest +{ + float f = fabs(-2.0f); + assert(f == 2); + + double d = fabs(-2.0); + assert(d == 2); + + real r = fabs(-2.0L); + assert(r == 2); +} + +@safe pure nothrow @nogc unittest +{ + float f = sin(-2.0f); + assert(fabs(f - -0.909297f) < .00001); + + double d = sin(-2.0); + assert(fabs(d - -0.909297f) < .00001); + + real r = sin(-2.0L); + assert(fabs(r - -0.909297f) < .00001); +} + +@safe pure nothrow @nogc unittest +{ + float f = cos(-2.0f); + assert(fabs(f - -0.416147f) < .00001); + + double d = cos(-2.0); + assert(fabs(d - -0.416147f) < .00001); + + real r = cos(-2.0L); + assert(fabs(r - -0.416147f) < .00001); +} + +@safe pure nothrow @nogc unittest +{ + float f = tan(-2.0f); + assert(fabs(f - 2.18504f) < .00001); + + double d = tan(-2.0); + assert(fabs(d - 2.18504f) < .00001); + + real r = tan(-2.0L); + assert(fabs(r - 2.18504f) < .00001); + + // Verify correct behavior for large inputs + assert(!isNaN(tan(0x1p63))); + assert(!isNaN(tan(0x1p300L))); + assert(!isNaN(tan(-0x1p63))); + assert(!isNaN(tan(-0x1p300L))); +} + +@safe pure nothrow unittest +{ + // issue 6381: floor/ceil should be usable in pure function. + auto x = floor(1.2); + auto y = ceil(1.2); +} + +@safe pure nothrow unittest +{ + // relative comparison depends on rhs, make sure proper side is used when + // comparing range to single value. Based on bugzilla issue 15763 + auto a = [2e-3 - 1e-5]; + auto b = 2e-3 + 1e-5; + assert(a[0].approxEqual(b)); + assert(!b.approxEqual(a[0])); + assert(a.approxEqual(b)); + assert(!b.approxEqual(a)); +} + +/*********************************** + * Defines a total order on all floating-point numbers. + * + * The order is defined as follows: + * $(UL + * $(LI All numbers in [-$(INFIN), +$(INFIN)] are ordered + * the same way as by built-in comparison, with the exception of + * -0.0, which is less than +0.0;) + * $(LI If the sign bit is set (that is, it's 'negative'), $(NAN) is less + * than any number; if the sign bit is not set (it is 'positive'), + * $(NAN) is greater than any number;) + * $(LI $(NAN)s of the same sign are ordered by the payload ('negative' + * ones - in reverse order).) + * ) + * + * Returns: + * negative value if $(D x) precedes $(D y) in the order specified above; + * 0 if $(D x) and $(D y) are identical, and positive value otherwise. + * + * See_Also: + * $(MYREF isIdentical) + * Standards: Conforms to IEEE 754-2008 + */ +int cmp(T)(const(T) x, const(T) y) @nogc @trusted pure nothrow +if (isFloatingPoint!T) +{ + alias F = floatTraits!T; + + static if (F.realFormat == RealFormat.ieeeSingle + || F.realFormat == RealFormat.ieeeDouble) + { + static if (T.sizeof == 4) + alias UInt = uint; + else + alias UInt = ulong; + + union Repainter + { + T number; + UInt bits; + } + + enum msb = ~(UInt.max >>> 1); + + import std.typecons : Tuple; + Tuple!(Repainter, Repainter) vars = void; + vars[0].number = x; + vars[1].number = y; + + foreach (ref var; vars) + if (var.bits & msb) + var.bits = ~var.bits; + else + var.bits |= msb; + + if (vars[0].bits < vars[1].bits) + return -1; + else if (vars[0].bits > vars[1].bits) + return 1; + else + return 0; + } + else static if (F.realFormat == RealFormat.ieeeExtended53 + || F.realFormat == RealFormat.ieeeExtended + || F.realFormat == RealFormat.ieeeQuadruple) + { + static if (F.realFormat == RealFormat.ieeeQuadruple) + alias RemT = ulong; + else + alias RemT = ushort; + + struct Bits + { + ulong bulk; + RemT rem; + } + + union Repainter + { + T number; + Bits bits; + ubyte[T.sizeof] bytes; + } + + import std.typecons : Tuple; + Tuple!(Repainter, Repainter) vars = void; + vars[0].number = x; + vars[1].number = y; + + foreach (ref var; vars) + if (var.bytes[F.SIGNPOS_BYTE] & 0x80) + { + var.bits.bulk = ~var.bits.bulk; + var.bits.rem = cast(typeof(var.bits.rem))(-1 - var.bits.rem); // ~var.bits.rem + } + else + { + var.bytes[F.SIGNPOS_BYTE] |= 0x80; + } + + version (LittleEndian) + { + if (vars[0].bits.rem < vars[1].bits.rem) + return -1; + else if (vars[0].bits.rem > vars[1].bits.rem) + return 1; + else if (vars[0].bits.bulk < vars[1].bits.bulk) + return -1; + else if (vars[0].bits.bulk > vars[1].bits.bulk) + return 1; + else + return 0; + } + else + { + if (vars[0].bits.bulk < vars[1].bits.bulk) + return -1; + else if (vars[0].bits.bulk > vars[1].bits.bulk) + return 1; + else if (vars[0].bits.rem < vars[1].bits.rem) + return -1; + else if (vars[0].bits.rem > vars[1].bits.rem) + return 1; + else + return 0; + } + } + else + { + // IBM Extended doubledouble does not follow the general + // sign-exponent-significand layout, so has to be handled generically + + const int xSign = signbit(x), + ySign = signbit(y); + + if (xSign == 1 && ySign == 1) + return cmp(-y, -x); + else if (xSign == 1) + return -1; + else if (ySign == 1) + return 1; + else if (x < y) + return -1; + else if (x == y) + return 0; + else if (x > y) + return 1; + else if (isNaN(x) && !isNaN(y)) + return 1; + else if (isNaN(y) && !isNaN(x)) + return -1; + else if (getNaNPayload(x) < getNaNPayload(y)) + return -1; + else if (getNaNPayload(x) > getNaNPayload(y)) + return 1; + else + return 0; + } +} + +/// Most numbers are ordered naturally. +@safe unittest +{ + assert(cmp(-double.infinity, -double.max) < 0); + assert(cmp(-double.max, -100.0) < 0); + assert(cmp(-100.0, -0.5) < 0); + assert(cmp(-0.5, 0.0) < 0); + assert(cmp(0.0, 0.5) < 0); + assert(cmp(0.5, 100.0) < 0); + assert(cmp(100.0, double.max) < 0); + assert(cmp(double.max, double.infinity) < 0); + + assert(cmp(1.0, 1.0) == 0); +} + +/// Positive and negative zeroes are distinct. +@safe unittest +{ + assert(cmp(-0.0, +0.0) < 0); + assert(cmp(+0.0, -0.0) > 0); +} + +/// Depending on the sign, $(NAN)s go to either end of the spectrum. +@safe unittest +{ + assert(cmp(-double.nan, -double.infinity) < 0); + assert(cmp(double.infinity, double.nan) < 0); + assert(cmp(-double.nan, double.nan) < 0); +} + +/// $(NAN)s of the same sign are ordered by the payload. +@safe unittest +{ + assert(cmp(NaN(10), NaN(20)) < 0); + assert(cmp(-NaN(20), -NaN(10)) < 0); +} + +@safe unittest +{ + import std.meta : AliasSeq; + foreach (T; AliasSeq!(float, double, real)) + { + T[] values = [-cast(T) NaN(20), -cast(T) NaN(10), -T.nan, -T.infinity, + -T.max, -T.max / 2, T(-16.0), T(-1.0).nextDown, + T(-1.0), T(-1.0).nextUp, + T(-0.5), -T.min_normal, (-T.min_normal).nextUp, + -2 * T.min_normal * T.epsilon, + -T.min_normal * T.epsilon, + T(-0.0), T(0.0), + T.min_normal * T.epsilon, + 2 * T.min_normal * T.epsilon, + T.min_normal.nextDown, T.min_normal, T(0.5), + T(1.0).nextDown, T(1.0), + T(1.0).nextUp, T(16.0), T.max / 2, T.max, + T.infinity, T.nan, cast(T) NaN(10), cast(T) NaN(20)]; + + foreach (i, x; values) + { + foreach (y; values[i + 1 .. $]) + { + assert(cmp(x, y) < 0); + assert(cmp(y, x) > 0); + } + assert(cmp(x, x) == 0); + } + } +} + +private enum PowType +{ + floor, + ceil +} + +pragma(inline, true) +private T powIntegralImpl(PowType type, T)(T val) +{ + import core.bitop : bsr; + + if (val == 0 || (type == PowType.ceil && (val > T.max / 2 || val == T.min))) + return 0; + else + { + static if (isSigned!T) + return cast(Unqual!T) (val < 0 ? -(T(1) << bsr(0 - val) + type) : T(1) << bsr(val) + type); + else + return cast(Unqual!T) (T(1) << bsr(val) + type); + } +} + +private T powFloatingPointImpl(PowType type, T)(T x) +{ + if (!x.isFinite) + return x; + + if (!x) + return x; + + int exp; + auto y = frexp(x, exp); + + static if (type == PowType.ceil) + y = ldexp(cast(T) 0.5, exp + 1); + else + y = ldexp(cast(T) 0.5, exp); + + if (!y.isFinite) + return cast(T) 0.0; + + y = copysign(y, x); + + return y; +} + +/** + * Gives the next power of two after $(D val). `T` can be any built-in + * numerical type. + * + * If the operation would lead to an over/underflow, this function will + * return `0`. + * + * Params: + * val = any number + * + * Returns: + * the next power of two after $(D val) + */ +T nextPow2(T)(const T val) +if (isIntegral!T) +{ + return powIntegralImpl!(PowType.ceil)(val); +} + +/// ditto +T nextPow2(T)(const T val) +if (isFloatingPoint!T) +{ + return powFloatingPointImpl!(PowType.ceil)(val); +} + +/// +@safe @nogc pure nothrow unittest +{ + assert(nextPow2(2) == 4); + assert(nextPow2(10) == 16); + assert(nextPow2(4000) == 4096); + + assert(nextPow2(-2) == -4); + assert(nextPow2(-10) == -16); + + assert(nextPow2(uint.max) == 0); + assert(nextPow2(uint.min) == 0); + assert(nextPow2(size_t.max) == 0); + assert(nextPow2(size_t.min) == 0); + + assert(nextPow2(int.max) == 0); + assert(nextPow2(int.min) == 0); + assert(nextPow2(long.max) == 0); + assert(nextPow2(long.min) == 0); +} + +/// +@safe @nogc pure nothrow unittest +{ + assert(nextPow2(2.1) == 4.0); + assert(nextPow2(-2.0) == -4.0); + assert(nextPow2(0.25) == 0.5); + assert(nextPow2(-4.0) == -8.0); + + assert(nextPow2(double.max) == 0.0); + assert(nextPow2(double.infinity) == double.infinity); +} + +@safe @nogc pure nothrow unittest +{ + assert(nextPow2(ubyte(2)) == 4); + assert(nextPow2(ubyte(10)) == 16); + + assert(nextPow2(byte(2)) == 4); + assert(nextPow2(byte(10)) == 16); + + assert(nextPow2(short(2)) == 4); + assert(nextPow2(short(10)) == 16); + assert(nextPow2(short(4000)) == 4096); + + assert(nextPow2(ushort(2)) == 4); + assert(nextPow2(ushort(10)) == 16); + assert(nextPow2(ushort(4000)) == 4096); +} + +@safe @nogc pure nothrow unittest +{ + foreach (ulong i; 1 .. 62) + { + assert(nextPow2(1UL << i) == 2UL << i); + assert(nextPow2((1UL << i) - 1) == 1UL << i); + assert(nextPow2((1UL << i) + 1) == 2UL << i); + assert(nextPow2((1UL << i) + (1UL<<(i-1))) == 2UL << i); + } +} + +@safe @nogc pure nothrow unittest +{ + import std.meta : AliasSeq; + + foreach (T; AliasSeq!(float, double, real)) + { + enum T subNormal = T.min_normal / 2; + + static if (subNormal) assert(nextPow2(subNormal) == T.min_normal); + + assert(nextPow2(T(0.0)) == 0.0); + + assert(nextPow2(T(2.0)) == 4.0); + assert(nextPow2(T(2.1)) == 4.0); + assert(nextPow2(T(3.1)) == 4.0); + assert(nextPow2(T(4.0)) == 8.0); + assert(nextPow2(T(0.25)) == 0.5); + + assert(nextPow2(T(-2.0)) == -4.0); + assert(nextPow2(T(-2.1)) == -4.0); + assert(nextPow2(T(-3.1)) == -4.0); + assert(nextPow2(T(-4.0)) == -8.0); + assert(nextPow2(T(-0.25)) == -0.5); + + assert(nextPow2(T.max) == 0); + assert(nextPow2(-T.max) == 0); + + assert(nextPow2(T.infinity) == T.infinity); + assert(nextPow2(T.init).isNaN); + } +} + +@safe @nogc pure nothrow unittest // Issue 15973 +{ + assert(nextPow2(uint.max / 2) == uint.max / 2 + 1); + assert(nextPow2(uint.max / 2 + 2) == 0); + assert(nextPow2(int.max / 2) == int.max / 2 + 1); + assert(nextPow2(int.max / 2 + 2) == 0); + assert(nextPow2(int.min + 1) == int.min); +} + +/** + * Gives the last power of two before $(D val). $(T) can be any built-in + * numerical type. + * + * Params: + * val = any number + * + * Returns: + * the last power of two before $(D val) + */ +T truncPow2(T)(const T val) +if (isIntegral!T) +{ + return powIntegralImpl!(PowType.floor)(val); +} + +/// ditto +T truncPow2(T)(const T val) +if (isFloatingPoint!T) +{ + return powFloatingPointImpl!(PowType.floor)(val); +} + +/// +@safe @nogc pure nothrow unittest +{ + assert(truncPow2(3) == 2); + assert(truncPow2(4) == 4); + assert(truncPow2(10) == 8); + assert(truncPow2(4000) == 2048); + + assert(truncPow2(-5) == -4); + assert(truncPow2(-20) == -16); + + assert(truncPow2(uint.max) == int.max + 1); + assert(truncPow2(uint.min) == 0); + assert(truncPow2(ulong.max) == long.max + 1); + assert(truncPow2(ulong.min) == 0); + + assert(truncPow2(int.max) == (int.max / 2) + 1); + assert(truncPow2(int.min) == int.min); + assert(truncPow2(long.max) == (long.max / 2) + 1); + assert(truncPow2(long.min) == long.min); +} + +/// +@safe @nogc pure nothrow unittest +{ + assert(truncPow2(2.1) == 2.0); + assert(truncPow2(7.0) == 4.0); + assert(truncPow2(-1.9) == -1.0); + assert(truncPow2(0.24) == 0.125); + assert(truncPow2(-7.0) == -4.0); + + assert(truncPow2(double.infinity) == double.infinity); +} + +@safe @nogc pure nothrow unittest +{ + assert(truncPow2(ubyte(3)) == 2); + assert(truncPow2(ubyte(4)) == 4); + assert(truncPow2(ubyte(10)) == 8); + + assert(truncPow2(byte(3)) == 2); + assert(truncPow2(byte(4)) == 4); + assert(truncPow2(byte(10)) == 8); + + assert(truncPow2(ushort(3)) == 2); + assert(truncPow2(ushort(4)) == 4); + assert(truncPow2(ushort(10)) == 8); + assert(truncPow2(ushort(4000)) == 2048); + + assert(truncPow2(short(3)) == 2); + assert(truncPow2(short(4)) == 4); + assert(truncPow2(short(10)) == 8); + assert(truncPow2(short(4000)) == 2048); +} + +@safe @nogc pure nothrow unittest +{ + foreach (ulong i; 1 .. 62) + { + assert(truncPow2(2UL << i) == 2UL << i); + assert(truncPow2((2UL << i) + 1) == 2UL << i); + assert(truncPow2((2UL << i) - 1) == 1UL << i); + assert(truncPow2((2UL << i) - (2UL<<(i-1))) == 1UL << i); + } +} + +@safe @nogc pure nothrow unittest +{ + import std.meta : AliasSeq; + + foreach (T; AliasSeq!(float, double, real)) + { + assert(truncPow2(T(0.0)) == 0.0); + + assert(truncPow2(T(4.0)) == 4.0); + assert(truncPow2(T(2.1)) == 2.0); + assert(truncPow2(T(3.5)) == 2.0); + assert(truncPow2(T(7.0)) == 4.0); + assert(truncPow2(T(0.24)) == 0.125); + + assert(truncPow2(T(-2.0)) == -2.0); + assert(truncPow2(T(-2.1)) == -2.0); + assert(truncPow2(T(-3.1)) == -2.0); + assert(truncPow2(T(-7.0)) == -4.0); + assert(truncPow2(T(-0.24)) == -0.125); + + assert(truncPow2(T.infinity) == T.infinity); + assert(truncPow2(T.init).isNaN); + } +} + +/** +Check whether a number is an integer power of two. + +Note that only positive numbers can be integer powers of two. This +function always return `false` if `x` is negative or zero. + +Params: + x = the number to test + +Returns: + `true` if `x` is an integer power of two. +*/ +bool isPowerOf2(X)(const X x) pure @safe nothrow @nogc +if (isNumeric!X) +{ + static if (isFloatingPoint!X) + { + int exp; + const X sig = frexp(x, exp); + + return (exp != int.min) && (sig is cast(X) 0.5L); + } + else + { + static if (isSigned!X) + { + auto y = cast(typeof(x + 0))x; + return y > 0 && !(y & (y - 1)); + } + else + { + auto y = cast(typeof(x + 0u))x; + return (y & -y) > (y - 1); + } + } +} +/// +@safe unittest +{ + assert( isPowerOf2(1.0L)); + assert( isPowerOf2(2.0L)); + assert( isPowerOf2(0.5L)); + assert( isPowerOf2(pow(2.0L, 96))); + assert( isPowerOf2(pow(2.0L, -77))); + + assert(!isPowerOf2(-2.0L)); + assert(!isPowerOf2(-0.5L)); + assert(!isPowerOf2(0.0L)); + assert(!isPowerOf2(4.315)); + assert(!isPowerOf2(1.0L / 3.0L)); + + assert(!isPowerOf2(real.nan)); + assert(!isPowerOf2(real.infinity)); +} +/// +@safe unittest +{ + assert( isPowerOf2(1)); + assert( isPowerOf2(2)); + assert( isPowerOf2(1uL << 63)); + + assert(!isPowerOf2(-4)); + assert(!isPowerOf2(0)); + assert(!isPowerOf2(1337u)); +} + +@safe unittest +{ + import std.meta : AliasSeq; + + immutable smallP2 = pow(2.0L, -62); + immutable bigP2 = pow(2.0L, 50); + immutable smallP7 = pow(7.0L, -35); + immutable bigP7 = pow(7.0L, 30); + + foreach (X; AliasSeq!(float, double, real)) + { + immutable min_sub = X.min_normal * X.epsilon; + + foreach (x; AliasSeq!(smallP2, min_sub, X.min_normal, .25L, 0.5L, 1.0L, + 2.0L, 8.0L, pow(2.0L, X.max_exp - 1), bigP2)) + { + assert( isPowerOf2(cast(X) x)); + assert(!isPowerOf2(cast(X)-x)); + } + + foreach (x; AliasSeq!(0.0L, 3 * min_sub, smallP7, 0.1L, 1337.0L, bigP7, X.max, real.nan, real.infinity)) + { + assert(!isPowerOf2(cast(X) x)); + assert(!isPowerOf2(cast(X)-x)); + } + } + + foreach (X; AliasSeq!(byte, ubyte, short, ushort, int, uint, long, ulong)) + { + foreach (x; [1, 2, 4, 8, (X.max >>> 1) + 1]) + { + assert( isPowerOf2(cast(X) x)); + static if (isSigned!X) + assert(!isPowerOf2(cast(X)-x)); + } + + foreach (x; [0, 3, 5, 13, 77, X.min, X.max]) + assert(!isPowerOf2(cast(X) x)); + } +} |