phobos/std/math/package.d
2021-04-17 16:59:34 +02:00

3315 lines
101 KiB
D
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

// 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 `std.mathspecial`.
*
$(SCRIPT inhibitQuickIndex = 1;)
$(DIVC quickindex,
$(BOOKTABLE ,
$(TR $(TH Category) $(TH Members) )
$(TR $(TDNW $(SUBMODULE Constants, constants)) $(TD
$(SUBREF constants, E)
$(SUBREF constants, PI)
$(SUBREF constants, PI_2)
$(SUBREF constants, PI_4)
$(SUBREF constants, M_1_PI)
$(SUBREF constants, M_2_PI)
$(SUBREF constants, M_2_SQRTPI)
$(SUBREF constants, LN10)
$(SUBREF constants, LN2)
$(SUBREF constants, LOG2)
$(SUBREF constants, LOG2E)
$(SUBREF constants, LOG2T)
$(SUBREF constants, LOG10E)
$(SUBREF constants, SQRT2)
$(SUBREF constants, SQRT1_2)
))
$(TR $(TDNW $(SUBMODULE Algebraic, algebraic)) $(TD
$(SUBREF algebraic, abs)
$(SUBREF algebraic, fabs)
$(SUBREF algebraic, sqrt)
$(SUBREF algebraic, cbrt)
$(SUBREF algebraic, hypot)
$(SUBREF algebraic, poly)
$(SUBREF algebraic, nextPow2)
$(SUBREF algebraic, truncPow2)
))
$(TR $(TDNW $(SUBMODULE Trigonometry, trigonometry)) $(TD
$(SUBREF trigonometry, sin)
$(SUBREF trigonometry, cos)
$(SUBREF trigonometry, tan)
$(SUBREF trigonometry, asin)
$(SUBREF trigonometry, acos)
$(SUBREF trigonometry, atan)
$(SUBREF trigonometry, atan2)
$(SUBREF trigonometry, sinh)
$(SUBREF trigonometry, cosh)
$(SUBREF trigonometry, tanh)
$(SUBREF trigonometry, asinh)
$(SUBREF trigonometry, acosh)
$(SUBREF trigonometry, atanh)
))
$(TR $(TDNW $(SUBMODULE Rounding, rounding)) $(TD
$(SUBREF rounding, ceil)
$(SUBREF rounding, floor)
$(SUBREF rounding, round)
$(SUBREF rounding, lround)
$(SUBREF rounding, trunc)
$(SUBREF rounding, rint)
$(SUBREF rounding, lrint)
$(SUBREF rounding, nearbyint)
$(SUBREF rounding, rndtol)
$(SUBREF rounding, quantize)
))
$(TR $(TDNW $(SUBMODULE Exponentiation & Logarithms, exponential)) $(TD
$(SUBREF exponential, pow)
$(SUBREF exponential, exp)
$(SUBREF exponential, exp2)
$(SUBREF exponential, expm1)
$(SUBREF exponential, ldexp)
$(SUBREF exponential, frexp)
$(SUBREF exponential, log)
$(SUBREF exponential, log2)
$(SUBREF exponential, log10)
$(SUBREF exponential, logb)
$(SUBREF exponential, ilogb)
$(SUBREF exponential, log1p)
$(SUBREF exponential, scalbn)
))
$(TR $(TDNW $(SUBMODULE Remainder, remainder)) $(TD
$(SUBREF remainder, fmod)
$(SUBREF remainder, modf)
$(SUBREF remainder, remainder)
$(SUBREF remainder, remquo)
))
$(TR $(TDNW Floating-point operations) $(TD
$(MYREF approxEqual) $(MYREF feqrel) $(MYREF fdim) $(MYREF fmax)
$(MYREF fmin) $(MYREF fma) $(MYREF isClose) $(MYREF nextDown) $(MYREF nextUp)
$(MYREF nextafter) $(MYREF NaN) $(MYREF getNaNPayload)
$(MYREF cmp)
))
$(TR $(TDNW $(SUBMODULE Introspection, traits)) $(TD
$(SUBREF traits, isFinite)
$(SUBREF traits, isIdentical)
$(SUBREF traits, isInfinity)
$(SUBREF traits, isNaN)
$(SUBREF traits, isNormal)
$(SUBREF traits, isSubnormal)
$(SUBREF traits, signbit)
$(SUBREF traits, sgn)
$(SUBREF traits, copysign)
$(SUBREF traits, isPowerOf2)
))
$(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.
*
* 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 = &#915;
* THETA = &theta;
* INTEGRAL = &#8747;
* INTEGRATE = $(BIG &#8747;<sub>$(SMALL $1)</sub><sup>$2</sup>)
* POWER = $1<sup>$2</sup>
* SUB = $1<sub>$2</sub>
* BIGSUM = $(BIG &Sigma; <sup>$2</sup><sub>$(SMALL $1)</sub>)
* CHOOSE = $(BIG &#40;) <sup>$(SMALL $1)</sup><sub>$(SMALL $2)</sub> $(BIG &#41;)
* PLUSMN = &plusmn;
* INFIN = &infin;
* PLUSMNINF = &plusmn;&infin;
* PI = &pi;
* LT = &lt;
* GT = &gt;
* SQRT = &radic;
* HALF = &frac12;
*
* SUBMODULE = $(MREF_ALTTEXT $1, std, math, $2)
* SUBREF = $(REF_ALTTEXT $(TT $2), $2, std, math, $1)$(NBSP)
*
* Copyright: Copyright The D Language Foundation 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/package.d)
*/
module std.math;
public import std.math.algebraic;
public import std.math.constants;
public import std.math.exponential;
public import std.math.floats;
public import std.math.hardware;
public import std.math.remainder;
public import std.math.rounding;
public import std.math.traits;
public import std.math.trigonometry;
static import core.math;
static import core.stdc.math;
static import core.stdc.fenv;
import std.traits : CommonType, isFloatingPoint, isIntegral, isNumeric,
isSigned, isUnsigned, Largest, Unqual;
// @@@DEPRECATED_2.102@@@
// Note: Exposed accidentally, should be deprecated / removed
deprecated("std.meta.AliasSeq was unintentionally available from std.math "
~ "and will be removed after 2.102. Please import std.meta instead")
public import std.meta : AliasSeq;
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 (S390) version = IBMZ_Any;
version (SPARC) version = SPARC_Any;
version (SPARC64) version = SPARC_Any;
version (SystemZ) version = IBMZ_Any;
version (RISCV32) version = RISCV_Any;
version (RISCV64) version = RISCV_Any;
version (D_InlineAsm_X86) version = InlineAsm_X86_Any;
version (D_InlineAsm_X86_64) version = InlineAsm_X86_Any;
version (InlineAsm_X86_Any) version = InlineAsm_X87;
version (InlineAsm_X87)
{
static assert(real.mant_dig == 64);
version (CRuntime_Microsoft) version = InlineAsm_X87_MSVC;
}
version (X86_64) version = StaticallyHaveSSE;
version (X86) version (OSX) version = StaticallyHaveSSE;
version (StaticallyHaveSSE)
{
private enum bool haveSSE = true;
}
else version (X86)
{
static import core.cpuid;
private alias haveSSE = core.cpuid.sse;
}
version (D_SoftFloat)
{
// Some soft float implementations may support IEEE floating flags.
// The implementation here supports hardware flags only and is so currently
// only available for supported targets.
}
else version (X86_Any) version = IeeeFlagsSupport;
else version (PPC_Any) version = IeeeFlagsSupport;
else version (RISCV_Any) version = IeeeFlagsSupport;
else version (MIPS_Any) version = IeeeFlagsSupport;
else version (ARM_Any) version = IeeeFlagsSupport;
// Struct FloatingPointControl is only available if hardware FP units are available.
version (D_HardFloat)
{
// FloatingPointControl.clearExceptions() depends on version IeeeFlagsSupport
version (IeeeFlagsSupport) version = FloatingPointControlSupport;
}
version (IeeeFlagsSupport)
{
/** 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
{
nothrow @nogc:
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).
// The RISC-V (32 & 64 bit) fcsr is 32-bit register.
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,
}
}
static uint getIeeeFlags() @trusted pure
{
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 version (RISCV_Any)
{
mixin(`
uint result = void;
asm pure nothrow @nogc
{
"frflags %0" : "=r" (result);
}
return result;
`);
}
else
assert(0, "Not yet supported");
}
static void resetIeeeFlags() @trusted
{
version (InlineAsm_X86_Any)
{
asm 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 version (RISCV_Any)
{
mixin(`
uint newValues = 0x0;
asm pure nothrow @nogc
{
"fsflags %0" : : "r" (newValues);
}
`);
}
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:
/**
* The result cannot be represented exactly, so rounding occurred.
* Example: `x = sin(0.1);`
*/
@property bool inexact() @safe const { return (flags & INEXACT_MASK) != 0; }
/**
* A zero was generated by underflow
* Example: `x = real.min*real.epsilon/2;`
*/
@property bool underflow() @safe const { return (flags & UNDERFLOW_MASK) != 0; }
/**
* An infinity was generated by overflow
* Example: `x = real.max*2;`
*/
@property bool overflow() @safe const { return (flags & OVERFLOW_MASK) != 0; }
/**
* An infinity was generated by division by zero
* Example: `x = 3/0.0;`
*/
@property bool divByZero() @safe const { return (flags & DIVBYZERO_MASK) != 0; }
/**
* A machine NaN was generated.
* Example: `x = real.infinity * 0.0;`
*/
@property bool invalid() @safe const { return (flags & INVALID_MASK) != 0; }
}
///
@safe unittest
{
static void func() {
int a = 10 * 10;
}
pragma(inline, false) static void blockopt(ref real x) {}
real a = 3.5;
// Set all the flags to zero
resetIeeeFlags();
assert(!ieeeFlags.divByZero);
blockopt(a); // avoid constant propagation by the optimizer
// Perform a division by zero.
a /= 0.0L;
assert(a == real.infinity);
assert(ieeeFlags.divByZero);
blockopt(a); // avoid constant propagation by the optimizer
// 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);
}
@safe unittest
{
import std.meta : AliasSeq;
static struct Test
{
void delegate() @trusted action;
bool function() @trusted ieeeCheck;
}
static 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 += 0.1L; },
() => 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());
}
}}
}
/// Set all of the floating-point status flags to false.
void resetIeeeFlags() @trusted nothrow @nogc
{
IeeeFlags.resetIeeeFlags();
}
///
@safe unittest
{
pragma(inline, false) static void blockopt(ref real x) {}
resetIeeeFlags();
real a = 3.5;
blockopt(a); // avoid constant propagation by the optimizer
a /= 0.0L;
blockopt(a); // avoid constant propagation by the optimizer
assert(a == real.infinity);
assert(ieeeFlags.divByZero);
resetIeeeFlags();
assert(!ieeeFlags.divByZero);
}
/// Returns: snapshot of the current state of the floating-point status flags
@property IeeeFlags ieeeFlags() @trusted pure nothrow @nogc
{
return IeeeFlags(IeeeFlags.getIeeeFlags());
}
///
@safe nothrow unittest
{
pragma(inline, false) static void blockopt(ref real x) {}
resetIeeeFlags();
real a = 3.5;
blockopt(a); // avoid constant propagation by the optimizer
a /= 0.0L;
assert(a == real.infinity);
assert(ieeeFlags.divByZero);
blockopt(a); // avoid constant propagation by the optimizer
a *= 0.0L;
assert(isNaN(a));
assert(ieeeFlags.invalid);
}
} // IeeeFlagsSupport
version (FloatingPointControlSupport)
{
/** 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
// The set hardware exceptions and rounding modes will be disabled when
// leaving this scope.
}
----
*/
struct FloatingPointControl
{
nothrow @nogc:
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
*
* Changing the rounding mode in the middle of a function can interfere
* with optimizations of floating point expressions, as the optimizer assumes
* that the rounding mode does not change.
* It is best to change the rounding mode only at the
* beginning of the function, and keep it until the function returns.
* It is also best to add the line:
* ---
* pragma(inline, false);
* ---
* as the first line of the function so it will not get inlined.
* Params:
* newMode = the new rounding mode
*/
@property void rounding(RoundingMode newMode) @trusted
{
initialize();
setControlState((getControlState() & (-1 - roundingMask)) | (newMode & roundingMask));
}
/// Returns: the currently active rounding mode
@property static RoundingMode rounding() @trusted pure
{
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 (PPC_Any)
{
enum : ExceptionMask
{
inexactException = 0x0008,
divByZeroException = 0x0010,
underflowException = 0x0020,
overflowException = 0x0040,
invalidException = 0x0080,
severeExceptions = overflowException | divByZeroException
| invalidException,
allExceptions = severeExceptions | underflowException
| inexactException,
}
}
else version (RISCV_Any)
{
enum : ExceptionMask
{
inexactException = 0x01,
divByZeroException = 0x02,
underflowException = 0x04,
overflowException = 0x08,
invalidException = 0x10,
severeExceptions = overflowException | divByZeroException
| invalidException,
allExceptions = severeExceptions | underflowException
| inexactException,
}
}
else version (HPPA)
{
enum : ExceptionMask
{
inexactException = 0x01,
underflowException = 0x02,
overflowException = 0x04,
divByZeroException = 0x08,
invalidException = 0x10,
severeExceptions = overflowException | divByZeroException
| invalidException,
allExceptions = severeExceptions | underflowException
| inexactException,
}
}
else version (MIPS_Any)
{
enum : ExceptionMask
{
inexactException = 0x0080,
divByZeroException = 0x0400,
overflowException = 0x0200,
underflowException = 0x0100,
invalidException = 0x0800,
severeExceptions = overflowException | divByZeroException
| invalidException,
allExceptions = severeExceptions | underflowException
| inexactException,
}
}
else version (SPARC_Any)
{
enum : ExceptionMask
{
inexactException = 0x0800000,
divByZeroException = 0x1000000,
overflowException = 0x4000000,
underflowException = 0x2000000,
invalidException = 0x8000000,
severeExceptions = overflowException | divByZeroException
| invalidException,
allExceptions = severeExceptions | underflowException
| inexactException,
}
}
else version (IBMZ_Any)
{
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");
version (ARM_Any)
{
static bool hasExceptionTraps_impl() @safe
{
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;
}
}
/// Returns: true if the current FPU supports exception trapping
@property static bool hasExceptionTraps() @safe pure
{
version (X86_Any)
return true;
else version (PPC_Any)
return true;
else version (MIPS_Any)
return true;
else version (ARM_Any)
{
// The hasExceptionTraps_impl function is basically pure,
// as it restores all global state
auto fptr = ( () @trusted => cast(bool function() @safe
pure nothrow @nogc)&hasExceptionTraps_impl)();
return fptr();
}
else
assert(0, "Not yet supported");
}
/// Enable (unmask) specific hardware exceptions. Multiple exceptions may be ORed together.
void enableExceptions(ExceptionMask exceptions) @trusted
{
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) @trusted
{
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() @trusted pure
{
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() @trusted
{
clearExceptions();
if (initialized)
setControlState(savedState);
}
private:
ControlState savedState;
bool initialized = false;
version (ARM_Any)
{
alias ControlState = uint;
}
else version (HPPA)
{
alias ControlState = uint;
}
else version (PPC_Any)
{
alias ControlState = uint;
}
else version (RISCV_Any)
{
alias ControlState = uint;
}
else version (MIPS_Any)
{
alias ControlState = uint;
}
else version (SPARC_Any)
{
alias ControlState = ulong;
}
else version (IBMZ_Any)
{
alias ControlState = uint;
}
else version (X86_Any)
{
alias ControlState = ushort;
}
else
static assert(false, "Not implemented for this architecture");
void initialize() @safe
{
// BUG: This works around the absence of this() constructors.
if (initialized) return;
clearExceptions();
savedState = getControlState();
initialized = true;
}
// Clear all pending exceptions
static void clearExceptions() @safe
{
version (IeeeFlagsSupport)
resetIeeeFlags();
else
static assert(false, "Not implemented for this architecture");
}
// Read from the control register
package(std.math) static ControlState getControlState() @trusted pure
{
version (D_InlineAsm_X86)
{
short cont;
asm pure nothrow @nogc
{
xor EAX, EAX;
fstcw cont;
}
return cont;
}
else version (D_InlineAsm_X86_64)
{
short cont;
asm pure nothrow @nogc
{
xor RAX, RAX;
fstcw cont;
}
return cont;
}
else version (RISCV_Any)
{
mixin(`
ControlState cont;
asm pure nothrow @nogc
{
"frcsr %0" : "=r" (cont);
}
return cont;
`);
}
else
assert(0, "Not yet supported");
}
// Set the control register
package(std.math) static void setControlState(ControlState newState) @trusted
{
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 version (RISCV_Any)
{
mixin(`
asm pure nothrow @nogc
{
"fscsr %0" : : "r" (newState);
}
`);
}
else
assert(0, "Not yet supported");
}
}
///
@safe unittest
{
FloatingPointControl fpctrl;
fpctrl.rounding = FloatingPointControl.roundDown;
assert(lrint(1.5) == 1.0);
fpctrl.rounding = FloatingPointControl.roundUp;
assert(lrint(1.4) == 2.0);
fpctrl.rounding = FloatingPointControl.roundToNearest;
assert(lrint(1.5) == 2.0);
}
@safe unittest
{
void ensureDefaults()
{
assert(FloatingPointControl.rounding
== FloatingPointControl.roundToNearest);
if (FloatingPointControl.hasExceptionTraps)
assert(FloatingPointControl.enabledExceptions == 0);
}
{
FloatingPointControl ctrl;
}
ensureDefaults();
{
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();
}
@safe unittest // rounding
{
import std.meta : AliasSeq;
static foreach (T; AliasSeq!(float, double, real))
{{
/* Be careful with changing the rounding mode, it interferes
* with common subexpressions. Changing rounding modes should
* be done with separate functions that are not inlined.
*/
{
static T addRound(T)(uint rm)
{
pragma(inline, false) static void blockopt(ref T x) {}
pragma(inline, false);
FloatingPointControl fpctrl;
fpctrl.rounding = rm;
T x = 1;
blockopt(x); // avoid constant propagation by the optimizer
x += 0.1L;
return x;
}
T u = addRound!(T)(FloatingPointControl.roundUp);
T d = addRound!(T)(FloatingPointControl.roundDown);
T z = addRound!(T)(FloatingPointControl.roundToZero);
assert(u > d);
assert(z == d);
}
{
static T subRound(T)(uint rm)
{
pragma(inline, false) static void blockopt(ref T x) {}
pragma(inline, false);
FloatingPointControl fpctrl;
fpctrl.rounding = rm;
T x = -1;
blockopt(x); // avoid constant propagation by the optimizer
x -= 0.1L;
return x;
}
T u = subRound!(T)(FloatingPointControl.roundUp);
T d = subRound!(T)(FloatingPointControl.roundDown);
T z = subRound!(T)(FloatingPointControl.roundToZero);
assert(u > d);
assert(z == u);
}
}}
}
} // FloatingPointControlSupport
// 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 ||
F.realFormat == RealFormat.ieeeExtended53)
{
// 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
}
if (__ctfe)
{
v = 1; // We use a double in CTFE.
assert(payload >>> 51 == 0,
"Cannot set more than 51 bits of NaN payload in CTFE.");
}
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;
if (__ctfe)
{
v |= 0x7FF0_0000_0000_0000;
return *cast(double*) &v;
}
else 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;
}
}
///
@safe @nogc pure nothrow unittest
{
real a = NaN(1_000_000);
assert(isNaN(a));
assert(getNaNPayload(a) == 1_000_000);
}
@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);
ulong m = void;
if (__ctfe)
{
double y = x;
m = *cast(ulong*) &y;
// 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.ieeeDouble)
{
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)
{
m = *cast(ulong*)(6+cast(ubyte*)(&x));
}
else
{
m = *cast(ulong*)(2+cast(ubyte*)(&x));
}
m >>= 1; // there's no implicit bit
}
else
{
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;
}
///
@safe @nogc pure nothrow unittest
{
real a = NaN(1_000_000);
assert(isNaN(a));
assert(getNaNPayload(a) == 1_000_000);
}
@safe @nogc pure nothrow unittest
{
enum real a = NaN(1_000_000);
static assert(isNaN(a));
static assert(getNaNPayload(a) == 1_000_000);
real b = NaN(1_000_000);
assert(isIdentical(b, a));
// The CTFE version of getNaNPayload relies on it being impossible
// for a CTFE-constructed NaN to have more than 51 bits of payload.
enum nanNaN = NaN(getNaNPayload(real.nan));
assert(isIdentical(real.nan, nanNaN));
static if (real.init != real.init)
{
enum initNaN = NaN(getNaNPayload(real.init));
assert(isIdentical(real.init, initNaN));
}
}
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)
{
if (__ctfe)
{
if (x == -real.infinity)
return -real.max;
if (!(x < real.infinity)) // Infinity or NaN.
return x;
real delta;
// Start with a decent estimate of delta.
if (x <= 0x1.ffffffffffffep+1023 && x >= -double.max)
{
const double d = cast(double) x;
delta = (cast(real) nextUp(d) - cast(real) d) * 0x1p-11L;
while (x + (delta * 0x1p-100L) > x)
delta *= 0x1p-100L;
}
else
{
delta = 0x1p960L;
while (!(x + delta > x) && delta < real.max * 0x1p-100L)
delta *= 0x1p100L;
}
if (x + delta > x)
{
while (x + (delta / 2) > x)
delta /= 2;
}
else
{
do { delta += delta; } while (!(x + delta > x));
}
if (x < 0 && x + delta == 0)
return -0.0L;
return x + delta;
}
}
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 ||
F.realFormat == RealFormat.ieeeExtended53)
{
// For 80-bit reals, the "implied bit" is a nuisance...
ushort *pe = cast(ushort *)&x;
ulong *ps = cast(ulong *)&x;
// EPSILON is 1 for 64-bit, and 2048 for 53-bit precision reals.
enum ulong EPSILON = 2UL ^^ (64 - real.mant_dig);
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 -= EPSILON;
// 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 += EPSILON;
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 s = *cast(ulong *)&x;
if ((s & 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 (s & 0x8000_0000_0000_0000) // Negative number
{
if (s == 0x8000_0000_0000_0000) // it was negative zero
{
s = 0x0000_0000_0000_0001; // change to smallest subnormal
return *cast(double*) &s;
}
--s;
}
else
{ // Positive number
++s;
}
return *cast(double*) &s;
}
/** ditto */
float nextUp(float x) @trusted pure nothrow @nogc
{
uint s = *cast(uint *)&x;
if ((s & 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 (s & 0x8000_0000) // Negative number
{
if (s == 0x8000_0000) // it was negative zero
{
s = 0x0000_0001; // change to smallest subnormal
return *cast(float*) &s;
}
--s;
}
else
{
// Positive number
++s;
}
return *cast(float*) &s;
}
///
@safe @nogc pure nothrow unittest
{
assert(nextUp(1.0 - 1.0e-6).feqrel(0.999999) > 16);
assert(nextUp(1.0 - real.epsilon).feqrel(1.0) > 16);
}
/**
* 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 ||
floatTraits!(real).realFormat == RealFormat.ieeeDouble ||
floatTraits!(real).realFormat == RealFormat.ieeeExtended53 ||
floatTraits!(real).realFormat == RealFormat.ieeeQuadruple)
{
// Tests for reals
assert(isIdentical(nextUp(NaN(0xABC)), NaN(0xABC)));
//static 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);
static assert( nextUp(-real.infinity) == -real.max );
static assert( nextUp(-1.0L-real.epsilon) == -1.0 );
static 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) );
static assert( nextUp(-real.min_normal) == -real.min_normal*(1-real.epsilon) );
static assert( nextUp(-real.min_normal*(1-real.epsilon)) == -real.min_normal*(1-2*real.epsilon) );
static assert( -0.0L is nextUp(-real.min_normal*real.epsilon) );
static assert( nextUp(-0.0L) == real.min_normal*real.epsilon );
static assert( nextUp(0.0L) == real.min_normal*real.epsilon );
static assert( nextUp(real.min_normal*(1-real.epsilon)) == real.min_normal );
static 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 );
static assert( nextUp(1.0L) == 1.0 + real.epsilon );
static assert( nextUp(2.0L-real.epsilon) == 2.0 );
static assert( nextUp(real.max) == real.infinity );
static assert( nextUp(real.infinity)==real.infinity );
// ctfe near double.max boundary
static assert(nextUp(nextDown(cast(real) double.max)) == cast(real) double.max);
}
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);
// CTFE
enum double ctfe_n = NaN(0xABC);
//static assert(isIdentical(nextUp(ctfe_n), ctfe_n)); // FIXME: https://issues.dlang.org/show_bug.cgi?id=20197
static assert(nextUp(double.nan) is double.nan);
// negative numbers
static assert( nextUp(-double.infinity) == -double.max );
static assert( nextUp(-1-double.epsilon) == -1.0 );
static assert( nextUp(-2.0) == -2.0 + double.epsilon);
// subnormals and zero
static assert( nextUp(-double.min_normal) == -double.min_normal*(1-double.epsilon) );
static assert( nextUp(-double.min_normal*(1-double.epsilon)) == -double.min_normal*(1-2*double.epsilon) );
static assert( -0.0 is nextUp(-double.min_normal*double.epsilon) );
static assert( nextUp(0.0) == double.min_normal*double.epsilon );
static assert( nextUp(-0.0) == double.min_normal*double.epsilon );
static assert( nextUp(double.min_normal*(1-double.epsilon)) == double.min_normal );
static assert( nextUp(double.min_normal) == double.min_normal*(1+double.epsilon) );
// positive numbers
static assert( nextUp(1.0) == 1.0 + double.epsilon );
static assert( nextUp(2.0-double.epsilon) == 2.0 );
static assert( nextUp(double.max) == double.infinity );
enum float ctfe_fn = NaN(0xABC);
//static assert(isIdentical(nextUp(ctfe_fn), ctfe_fn)); // FIXME: https://issues.dlang.org/show_bug.cgi?id=20197
static assert(nextUp(float.nan) is float.nan);
static assert(nextUp(-float.min_normal) == -float.min_normal*(1-float.epsilon));
static assert(nextUp(1.0f) == 1.0f+float.epsilon);
static assert(nextUp(-0.0f) == float.min_normal*float.epsilon);
static assert(nextUp(float.infinity)==float.infinity);
static assert(nextDown(1.0L+real.epsilon)==1.0);
static assert(nextDown(1.0+double.epsilon)==1.0);
static assert(nextDown(1.0f+float.epsilon)==1.0);
static 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.
* If x or y is a NaN, the result is a NaN.
*
* 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 || isNaN(y))
{
return y;
}
if (isNaN(x))
{
return x;
}
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);
assert(isNaN(nextafter(a, a.nan)));
assert(isNaN(nextafter(a.nan, a)));
double b = 2;
assert(is(typeof(nextafter(b, b)) == double));
assert(nextafter(b, b.infinity) > b);
assert(isNaN(nextafter(b, b.nan)));
assert(isNaN(nextafter(b.nan, b)));
real c = 3;
assert(is(typeof(nextafter(c, c)) == real));
assert(nextafter(c, c.infinity) > c);
assert(isNaN(nextafter(c, c.nan)));
assert(isNaN(nextafter(c.nan, c)));
}
@safe pure nothrow @nogc unittest
{
// CTFE
enum float a = 1;
static assert(is(typeof(nextafter(a, a)) == float));
static assert(nextafter(a, a.infinity) > a);
static assert(isNaN(nextafter(a, a.nan)));
static assert(isNaN(nextafter(a.nan, a)));
enum double b = 2;
static assert(is(typeof(nextafter(b, b)) == double));
static assert(nextafter(b, b.infinity) > b);
static assert(isNaN(nextafter(b, b.nan)));
static assert(isNaN(nextafter(b.nan, b)));
enum real c = 3;
static assert(is(typeof(nextafter(c, c)) == real));
static assert(nextafter(c, c.infinity) > c);
static assert(isNaN(nextafter(c, c.nan)));
static assert(isNaN(nextafter(c.nan, c)));
enum real negZero = nextafter(+0.0L, -0.0L);
static assert(negZero == -0.0L);
static assert(signbit(negZero));
static assert(nextafter(c, c) == c);
}
//real nexttoward(real x, real y) { return core.stdc.math.nexttowardl(x, y); }
/**
* Returns the positive difference between x and y.
*
* Equivalent to `fmax(x-y, 0)`.
*
* 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) ? +0.0 : x - y;
}
///
@safe pure nothrow @nogc unittest
{
assert(fdim(2.0, 0.0) == 2.0);
assert(fdim(-2.0, 0.0) == 0.0);
assert(fdim(real.infinity, 2.0) == real.infinity);
assert(isNaN(fdim(real.nan, 2.0)));
assert(isNaN(fdim(2.0, real.nan)));
assert(isNaN(fdim(real.nan, real.nan)));
}
/**
* Returns the larger of `x` and `y`.
*
* If one of the arguments is a `NaN`, the other is returned.
*
* See_Also: $(REF max, std,algorithm,comparison) is faster because it does not perform the `isNaN` test.
*/
F fmax(F)(const F x, const F y) @safe pure nothrow @nogc
if (__traits(isFloating, F))
{
// Do the more predictable test first. Generates 0 branches with ldc and 1 branch with gdc.
// See https://godbolt.org/z/erxrW9
if (isNaN(x)) return y;
return y > x ? y : x;
}
///
@safe pure nothrow @nogc unittest
{
import std.meta : AliasSeq;
static foreach (F; AliasSeq!(float, double, real))
{
assert(fmax(F(0.0), F(2.0)) == 2.0);
assert(fmax(F(-2.0), 0.0) == F(0.0));
assert(fmax(F.infinity, F(2.0)) == F.infinity);
assert(fmax(F.nan, F(2.0)) == F(2.0));
assert(fmax(F(2.0), F.nan) == F(2.0));
}
}
/**
* Returns the smaller of `x` and `y`.
*
* If one of the arguments is a `NaN`, the other is returned.
*
* See_Also: $(REF min, std,algorithm,comparison) is faster because it does not perform the `isNaN` test.
*/
F fmin(F)(const F x, const F y) @safe pure nothrow @nogc
if (__traits(isFloating, F))
{
// Do the more predictable test first. Generates 0 branches with ldc and 1 branch with gdc.
// See https://godbolt.org/z/erxrW9
if (isNaN(x)) return y;
return y < x ? y : x;
}
///
@safe pure nothrow @nogc unittest
{
import std.meta : AliasSeq;
static foreach (F; AliasSeq!(float, double, real))
{
assert(fmin(F(0.0), F(2.0)) == 0.0);
assert(fmin(F(-2.0), F(0.0)) == -2.0);
assert(fmin(F.infinity, F(2.0)) == 2.0);
assert(fmin(F.nan, F(2.0)) == 2.0);
assert(fmin(F(2.0), F.nan) == 2.0);
}
}
/**************************************
* Returns (x * y) + z, rounding only once according to the
* current rounding mode.
*
* BUGS: Not currently implemented - rounds twice.
*/
pragma(inline, true)
real fma(real x, real y, real z) @safe pure nothrow @nogc { return (x * y) + z; }
///
@safe pure nothrow @nogc unittest
{
assert(fma(0.0, 2.0, 2.0) == 2.0);
assert(fma(2.0, 2.0, 2.0) == 6.0);
assert(fma(real.infinity, 2.0, 2.0) == real.infinity);
assert(fma(real.nan, 2.0, 2.0) is real.nan);
assert(fma(2.0, 2.0, real.nan) is real.nan);
}
/** Computes the value of a positive integer `x`, raised to the power `n`, modulo `m`.
*
* Params:
* x = base
* n = exponent
* m = modulus
*
* Returns:
* `x` to the power `n`, modulo `m`.
* The return type is the largest of `x`'s and `m`'s type.
*
* The function requires that all values have unsigned types.
*/
Unqual!(Largest!(F, H)) powmod(F, G, H)(F x, G n, H m)
if (isUnsigned!F && isUnsigned!G && isUnsigned!H)
{
import std.meta : AliasSeq;
alias T = Unqual!(Largest!(F, H));
static if (T.sizeof <= 4)
{
alias DoubleT = AliasSeq!(void, ushort, uint, void, ulong)[T.sizeof];
}
static T mulmod(T a, T b, T c)
{
static if (T.sizeof == 8)
{
static T addmod(T a, T b, T c)
{
b = c - b;
if (a >= b)
return a - b;
else
return c - b + a;
}
T result = 0, tmp;
b %= c;
while (a > 0)
{
if (a & 1)
result = addmod(result, b, c);
a >>= 1;
b = addmod(b, b, c);
}
return result;
}
else
{
DoubleT result = cast(DoubleT) (cast(DoubleT) a * cast(DoubleT) b);
return result % c;
}
}
T base = x, result = 1, modulus = m;
Unqual!G exponent = n;
while (exponent > 0)
{
if (exponent & 1)
result = mulmod(result, base, modulus);
base = mulmod(base, base, modulus);
exponent >>= 1;
}
return result;
}
///
@safe pure nothrow @nogc unittest
{
assert(powmod(1U, 10U, 3U) == 1);
assert(powmod(3U, 2U, 6U) == 3);
assert(powmod(5U, 5U, 15U) == 5);
assert(powmod(2U, 3U, 5U) == 3);
assert(powmod(2U, 4U, 5U) == 1);
assert(powmod(2U, 5U, 5U) == 2);
}
@safe pure nothrow @nogc unittest
{
ulong a = 18446744073709551615u, b = 20u, c = 18446744073709551610u;
assert(powmod(a, b, c) == 95367431640625u);
a = 100; b = 7919; c = 18446744073709551557u;
assert(powmod(a, b, c) == 18223853583554725198u);
a = 117; b = 7919; c = 18446744073709551557u;
assert(powmod(a, b, c) == 11493139548346411394u);
a = 134; b = 7919; c = 18446744073709551557u;
assert(powmod(a, b, c) == 10979163786734356774u);
a = 151; b = 7919; c = 18446744073709551557u;
assert(powmod(a, b, c) == 7023018419737782840u);
a = 168; b = 7919; c = 18446744073709551557u;
assert(powmod(a, b, c) == 58082701842386811u);
a = 185; b = 7919; c = 18446744073709551557u;
assert(powmod(a, b, c) == 17423478386299876798u);
a = 202; b = 7919; c = 18446744073709551557u;
assert(powmod(a, b, c) == 5522733478579799075u);
a = 219; b = 7919; c = 18446744073709551557u;
assert(powmod(a, b, c) == 15230218982491623487u);
a = 236; b = 7919; c = 18446744073709551557u;
assert(powmod(a, b, c) == 5198328724976436000u);
a = 0; b = 7919; c = 18446744073709551557u;
assert(powmod(a, b, c) == 0);
a = 123; b = 0; c = 18446744073709551557u;
assert(powmod(a, b, c) == 1);
immutable ulong a1 = 253, b1 = 7919, c1 = 18446744073709551557u;
assert(powmod(a1, b1, c1) == 3883707345459248860u);
uint x = 100 ,y = 7919, z = 1844674407u;
assert(powmod(x, y, z) == 1613100340u);
x = 134; y = 7919; z = 1844674407u;
assert(powmod(x, y, z) == 734956622u);
x = 151; y = 7919; z = 1844674407u;
assert(powmod(x, y, z) == 1738696945u);
x = 168; y = 7919; z = 1844674407u;
assert(powmod(x, y, z) == 1247580927u);
x = 185; y = 7919; z = 1844674407u;
assert(powmod(x, y, z) == 1293855176u);
x = 202; y = 7919; z = 1844674407u;
assert(powmod(x, y, z) == 1566963682u);
x = 219; y = 7919; z = 1844674407u;
assert(powmod(x, y, z) == 181227807u);
x = 236; y = 7919; z = 1844674407u;
assert(powmod(x, y, z) == 217988321u);
x = 253; y = 7919; z = 1844674407u;
assert(powmod(x, y, z) == 1588843243u);
x = 0; y = 7919; z = 184467u;
assert(powmod(x, y, z) == 0);
x = 123; y = 0; z = 1844674u;
assert(powmod(x, y, z) == 1);
immutable ubyte x1 = 117;
immutable uint y1 = 7919;
immutable uint z1 = 1844674407u;
auto res = powmod(x1, y1, z1);
assert(is(typeof(res) == uint));
assert(res == 9479781u);
immutable ushort x2 = 123;
immutable uint y2 = 203;
immutable ubyte z2 = 113;
auto res2 = powmod(x2, y2, z2);
assert(is(typeof(res2) == ushort));
assert(res2 == 42u);
}
/**************************************
* 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.ieeeSingle
|| F.realFormat == RealFormat.ieeeDouble
|| F.realFormat == RealFormat.ieeeExtended
|| F.realFormat == RealFormat.ieeeExtended53
|| 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;
}
else
{
static assert(false, "Not implemented for this architecture");
}
}
///
@safe pure unittest
{
assert(feqrel(2.0, 2.0) == 53);
assert(feqrel(2.0f, 2.0f) == 24);
assert(feqrel(2.0, double.nan) == 0);
// Test that numbers are within n digits of each
// other by testing if feqrel > n * log2(10)
// five digits
assert(feqrel(2.0, 2.00001) > 16);
// ten digits
assert(feqrel(2.0, 2.00000000001) > 33);
}
@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)();
}
/**
Computes whether a values is approximately equal to a reference value,
admitting a maximum relative difference, and a maximum absolute difference.
Warning:
This template is considered out-dated. It will be removed from
Phobos in 2.106.0. Please use $(LREF isClose) instead.
Params:
value = Value to compare.
reference = Reference value.
maxRelDiff = Maximum allowable difference relative to `reference`.
Setting to 0.0 disables this check. Defaults to `1e-2`.
maxAbsDiff = Maximum absolute difference. This is mainly usefull
for comparing values to zero. Setting to 0.0 disables this check.
Defaults to `1e-5`.
Returns:
`true` if `value` is approximately equal to `reference` under
either criterium. It is sufficient, when `value ` satisfies
one of the two criteria.
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.
See_Also:
Use $(LREF feqrel) to get the number of equal bits in the mantissa.
*/
deprecated("approxEqual will be removed in 2.106.0. Please use isClose instead.")
bool approxEqual(T, U, V)(T value, U reference, V maxRelDiff = 1e-2, V maxAbsDiff = 1e-5)
{
import std.range.primitives : empty, front, isInputRange, popFront;
static if (isInputRange!T)
{
static if (isInputRange!U)
{
// Two ranges
for (;; value.popFront(), reference.popFront())
{
if (value.empty) return reference.empty;
if (reference.empty) return value.empty;
if (!approxEqual(value.front, reference.front, maxRelDiff, maxAbsDiff))
return false;
}
}
else static if (isIntegral!U)
{
// convert reference to real
return approxEqual(value, real(reference), maxRelDiff, maxAbsDiff);
}
else
{
// value is range, reference is number
for (; !value.empty; value.popFront())
{
if (!approxEqual(value.front, reference, maxRelDiff, maxAbsDiff))
return false;
}
return true;
}
}
else
{
static if (isInputRange!U)
{
// value is number, reference is range
for (; !reference.empty; reference.popFront())
{
if (!approxEqual(value, reference.front, maxRelDiff, maxAbsDiff))
return false;
}
return true;
}
else static if (isIntegral!T || isIntegral!U)
{
// convert both value and reference to real
return approxEqual(real(value), real(reference), maxRelDiff, maxAbsDiff);
}
else
{
// two numbers
//static assert(is(T : real) && is(U : real));
if (reference == 0)
{
return fabs(value) <= maxAbsDiff;
}
static if (is(typeof(value.infinity)) && is(typeof(reference.infinity)))
{
if (value == value.infinity && reference == reference.infinity ||
value == -value.infinity && reference == -reference.infinity) return true;
}
return fabs((value - reference) / reference) <= maxRelDiff
|| maxAbsDiff != 0 && fabs(value - reference) <= maxAbsDiff;
}
}
}
deprecated @safe pure nothrow unittest
{
assert(approxEqual(1.0, 1.0099));
assert(!approxEqual(1.0, 1.011));
assert(approxEqual(0.00001, 0.0));
assert(!approxEqual(0.00002, 0.0));
assert(approxEqual(3.0, [3, 3.01, 2.99])); // several reference values is strange
assert(approxEqual([3, 3.01, 2.99], 3.0)); // better
float[] arr1 = [ 1.0, 2.0, 3.0 ];
double[] arr2 = [ 1.001, 1.999, 3 ];
assert(approxEqual(arr1, arr2));
}
deprecated @safe pure nothrow unittest
{
// relative comparison depends on reference, make sure proper
// side is used when comparing range to single value. Based on
// https://issues.dlang.org/show_bug.cgi?id=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));
}
deprecated @safe pure nothrow @nogc unittest
{
assert(!approxEqual(0.0,1e-15,1e-9,0.0));
assert(approxEqual(0.0,1e-15,1e-9,1e-9));
assert(!approxEqual(1.0,3.0,0.0,1.0));
assert(approxEqual(1.00000000099,1.0,1e-9,0.0));
assert(!approxEqual(1.0000000011,1.0,1e-9,0.0));
}
deprecated @safe pure nothrow @nogc unittest
{
// maybe unintuitive behavior
assert(approxEqual(1000.0,1010.0));
assert(approxEqual(9_090_000_000.0,9_000_000_000.0));
assert(approxEqual(0.0,1e30,1.0));
assert(approxEqual(0.00001,1e-30));
assert(!approxEqual(-1e-30,1e-30,1e-2,0.0));
}
deprecated @safe pure nothrow @nogc unittest
{
int a = 10;
assert(approxEqual(10, a));
assert(!approxEqual(3, 0));
assert(approxEqual(3, 3));
assert(approxEqual(3.0, 3));
assert(approxEqual(3, 3.0));
assert(approxEqual(0.0,0.0));
assert(approxEqual(-0.0,0.0));
assert(approxEqual(0.0f,0.0));
}
deprecated @safe pure nothrow @nogc unittest
{
real num = real.infinity;
assert(num == real.infinity);
assert(approxEqual(num, real.infinity));
num = -real.infinity;
assert(num == -real.infinity);
assert(approxEqual(num, -real.infinity));
assert(!approxEqual(1,real.nan));
assert(!approxEqual(real.nan,real.max));
assert(!approxEqual(real.nan,real.nan));
}
deprecated @safe pure nothrow unittest
{
assert(!approxEqual([1.0,2.0,3.0],[1.0,2.0]));
assert(!approxEqual([1.0,2.0],[1.0,2.0,3.0]));
assert(approxEqual!(real[],real[])([],[]));
assert(approxEqual(cast(real[])[],cast(real[])[]));
}
/**
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 relative difference.
Setting to 0.0 disables this check. Default depends on the type of
`lhs` and `rhs`: It is approximately half the number of decimal digits of
precision of the smaller type.
maxAbsDiff = Maximum absolute difference. This is mainly usefull
for comparing values to zero. Setting to 0.0 disables this check.
Defaults to `0.0`.
Returns:
`true` if the two items are approximately equal under either criterium.
It is sufficient, when `value ` satisfies one of the two criteria.
If one item is a range, and the other is a single value, then
the result is the logical and-ing of calling `isClose` on
each element of the ranged item against the single item. If
both items are ranges, then `isClose` returns `true` if
and only if the ranges have the same number of elements and if
`isClose` evaluates to `true` for each pair of elements.
See_Also:
Use $(LREF feqrel) to get the number of equal bits in the mantissa.
*/
bool isClose(T, U, V = CommonType!(FloatingPointBaseType!T,FloatingPointBaseType!U))
(T lhs, U rhs, V maxRelDiff = CommonDefaultFor!(T,U), V maxAbsDiff = 0.0)
{
import std.range.primitives : empty, front, isInputRange, popFront;
import std.complex : Complex;
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 (!isClose(lhs.front, rhs.front, maxRelDiff, maxAbsDiff))
return false;
}
}
else
{
// lhs is range, rhs is number
for (; !lhs.empty; lhs.popFront())
{
if (!isClose(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 (!isClose(lhs, rhs.front, maxRelDiff, maxAbsDiff))
return false;
}
return true;
}
else static if (is(T TE == Complex!TE))
{
static if (is(U UE == Complex!UE))
{
// Two complex numbers
return isClose(lhs.re, rhs.re, maxRelDiff, maxAbsDiff)
&& isClose(lhs.im, rhs.im, maxRelDiff, maxAbsDiff);
}
else
{
// lhs is complex, rhs is number
return isClose(lhs.re, rhs, maxRelDiff, maxAbsDiff)
&& isClose(lhs.im, 0.0, maxRelDiff, maxAbsDiff);
}
}
else static if (is(U UE == Complex!UE))
{
// lhs is number, rhs is complex
return isClose(lhs, rhs.re, maxRelDiff, maxAbsDiff)
&& isClose(0.0, rhs.im, maxRelDiff, maxAbsDiff);
}
else
{
// two numbers
if (lhs == rhs) return true;
static if (is(typeof(lhs.infinity)) && is(typeof(rhs.infinity)))
{
if (lhs == lhs.infinity || rhs == rhs.infinity ||
lhs == -lhs.infinity || rhs == -rhs.infinity) return false;
}
auto diff = abs(lhs - rhs);
return diff <= maxRelDiff*abs(lhs)
|| diff <= maxRelDiff*abs(rhs)
|| diff <= maxAbsDiff;
}
}
///
@safe pure nothrow @nogc unittest
{
assert(isClose(1.0,0.999_999_999));
assert(isClose(0.001, 0.000_999_999_999));
assert(isClose(1_000_000_000.0,999_999_999.0));
assert(isClose(17.123_456_789, 17.123_456_78));
assert(!isClose(17.123_456_789, 17.123_45));
// use explicit 3rd parameter for less (or more) accuracy
assert(isClose(17.123_456_789, 17.123_45, 1e-6));
assert(!isClose(17.123_456_789, 17.123_45, 1e-7));
// use 4th parameter when comparing close to zero
assert(!isClose(1e-100, 0.0));
assert(isClose(1e-100, 0.0, 0.0, 1e-90));
assert(!isClose(1e-10, -1e-10));
assert(isClose(1e-10, -1e-10, 0.0, 1e-9));
assert(!isClose(1e-300, 1e-298));
assert(isClose(1e-300, 1e-298, 0.0, 1e-200));
// different default limits for different floating point types
assert(isClose(1.0f, 0.999_99f));
assert(!isClose(1.0, 0.999_99));
static if (real.sizeof > double.sizeof)
assert(!isClose(1.0L, 0.999_999_999L));
}
///
@safe pure nothrow unittest
{
assert(isClose([1.0, 2.0, 3.0], [0.999_999_999, 2.000_000_001, 3.0]));
assert(!isClose([1.0, 2.0], [0.999_999_999, 2.000_000_001, 3.0]));
assert(!isClose([1.0, 2.0, 3.0], [0.999_999_999, 2.000_000_001]));
assert(isClose([2.0, 1.999_999_999, 2.000_000_001], 2.0));
assert(isClose(2.0, [2.0, 1.999_999_999, 2.000_000_001]));
}
@safe pure nothrow unittest
{
assert(!isClose([1.0, 2.0, 3.0], [0.999_999_999, 3.0, 3.0]));
assert(!isClose([2.0, 1.999_999, 2.000_000_001], 2.0));
assert(!isClose(2.0, [2.0, 1.999_999_999, 2.000_000_999]));
}
@safe pure nothrow @nogc unittest
{
immutable a = 1.00001f;
const b = 1.000019;
assert(isClose(a,b));
assert(isClose(1.00001f,1.000019f));
assert(isClose(1.00001f,1.000019));
assert(isClose(1.00001,1.000019f));
assert(!isClose(1.00001,1.000019));
real a1 = 1e-300L;
real a2 = a1.nextUp;
assert(isClose(a1,a2));
}
@safe pure nothrow unittest
{
float[] arr1 = [ 1.0, 2.0, 3.0 ];
double[] arr2 = [ 1.00001, 1.99999, 3 ];
assert(isClose(arr1, arr2));
}
@safe pure nothrow @nogc unittest
{
assert(!isClose(1000.0,1010.0));
assert(!isClose(9_090_000_000.0,9_000_000_000.0));
assert(isClose(0.0,1e30,1.0));
assert(!isClose(0.00001,1e-30));
assert(!isClose(-1e-30,1e-30,1e-2,0.0));
}
@safe pure nothrow @nogc unittest
{
assert(!isClose(3, 0));
assert(isClose(3, 3));
assert(isClose(3.0, 3));
assert(isClose(3, 3.0));
assert(isClose(0.0,0.0));
assert(isClose(-0.0,0.0));
assert(isClose(0.0f,0.0));
}
@safe pure nothrow @nogc unittest
{
real num = real.infinity;
assert(num == real.infinity);
assert(isClose(num, real.infinity));
num = -real.infinity;
assert(num == -real.infinity);
assert(isClose(num, -real.infinity));
assert(!isClose(1,real.nan));
assert(!isClose(real.nan,real.max));
assert(!isClose(real.nan,real.nan));
}
@safe pure nothrow @nogc unittest
{
assert(isClose!(real[],real[],real)([],[]));
assert(isClose(cast(real[])[],cast(real[])[]));
}
@safe pure nothrow @nogc unittest
{
import std.conv : to;
float f = 31.79f;
double d = 31.79;
double f2d = f.to!double;
assert(isClose(f,f2d));
assert(!isClose(d,f2d));
}
@safe pure nothrow @nogc unittest
{
import std.conv : to;
double d = 31.79;
float f = d.to!float;
double f2d = f.to!double;
assert(isClose(f,f2d));
assert(!isClose(d,f2d));
assert(isClose(d,f2d,1e-4));
}
package(std.math) template CommonDefaultFor(T,U)
{
import std.algorithm.comparison : min;
alias baseT = FloatingPointBaseType!T;
alias baseU = FloatingPointBaseType!U;
enum CommonType!(baseT, baseU) CommonDefaultFor = 10.0L ^^ -((min(baseT.dig, baseU.dig) + 1) / 2 + 1);
}
private template FloatingPointBaseType(T)
{
import std.range.primitives : ElementType;
static if (isFloatingPoint!T)
{
alias FloatingPointBaseType = Unqual!T;
}
else static if (isFloatingPoint!(ElementType!(Unqual!T)))
{
alias FloatingPointBaseType = Unqual!(ElementType!(Unqual!T));
}
else
{
alias FloatingPointBaseType = real;
}
}
@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(-0x1p63)));
static if (real.mant_dig >= 64)
{
assert(!isNaN(tan(0x1p300L)));
assert(!isNaN(tan(-0x1p300L)));
}
}
/***********************************
* 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 `x` precedes `y` in the order specified above;
* 0 if `x` and `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;
static 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);
}
}}
}
package(std): // 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);
}
do
{
// 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 ||
F.realFormat == RealFormat.ieeeExtended53)
{
// 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));
}
// 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 == 113,
"Only 64-bit and 128-bit reals are supported for BigEndian CPUs.");
}
// 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 Unqual!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;
// For IBM doubledouble the larger magnitude double comes first.
// It's really a double[2] and arrays don't index differently
// between little and big-endian targets.
enum DOUBLEPAIR_MSB = 0;
enum DOUBLEPAIR_LSB = 1;
// The exponent/sign byte is for most significant part.
version (LittleEndian)
{
enum EXPPOS_SHORT = 3;
enum SIGNPOS_BYTE = 7;
}
else
{
enum EXPPOS_SHORT = 0;
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;
}