Source Edit

Nim's Python-like math, but raises no catchable exception, using errno to report math error. a.k.a. non will be raised (but not for Defects)

which allows fast code.

For Python-like error handler (exception-based), see Lib/math

Also this is not just a module that wraps Nim's std/math,

but also providing some extensions that Nim's std/math lacks, for example:

JavaScript Backend and Compile-time (Nim VM) support for ldexp, frexp
JavaScript Backend support for erf, erfc, gamma, lgamma
fsum, prod, etc.

And fix some non-C99 behavior on some systems, e.g. log(-ve) -> -Inf on Solaris

NOTE: Currently int is not acceptable when it comes to float about functions

Imports

math_impl/platformUtils, math_impl/errnoUtils, math_impl/ldexp, math_impl/cbrt, math_impl/frexp, math_impl/nextafter_ulp, errno, math_impl/patch/fma, math_impl/vec_op, math_impl/isX, math_impl/patch/gamma, math_impl/patch/lgamma

Consts

e = 2.718281828459045: Source Edit
inf = 0x7FF0000000000000'f64: Source Edit
nan = 0x7FF7FFFFFFFFFFFF'f64: Source Edit
pi = 3.141592653589793: Source Edit
tau = 6.283185307179586: Source Edit

Procs

func acos[F: SomeFloat](x: F): F: Source Edit
func acosh[F: SomeFloat](x: F): F: Source Edit
func asin[F: SomeFloat](x: F): F: Source Edit
func asinh[F: SomeFloat](x: F): F: Source Edit
func atan[F: SomeFloat](x: F): F: Source Edit
func atan2[F: SomeFloat](x: F): F: Source Edit
func atanh[F: SomeFloat](x: F): F: Source Edit
func ceil(x: SomeFloat): int: Hint: unlike C/Nim's returning float, Python's returns int and raises errors for nans and infinities
Source Edit
func comb(n, k: int): int {....raises: [], tags: [], forbids: [].}: Hint: Python's math.comb does not accept negative value for n, k but Nim's std/math.binom allows, so this function allows too. For consistent behavior with Python, see Lib/math.comb
Source Edit
func degrees[F: SomeFloat](x: F): F: Source Edit
func expm1[F: SomeFloat](x: F): F: Source Edit
func fabs[F: SomeFloat](x: F): F: Source Edit
func factorial(x: Natural): int {....raises: [], tags: [], forbids: [].}: Source Edit
func floor(x: SomeFloat): int: Hint: unlike C/Nim's returning float, Python's returns int and raises errors for nans and infinities
Source Edit
func fmod[F: SomeFloat](x: F; y: F): F: Source Edit
func fsum[T: SomeFloat](iterable: Iterable[T]): T: Source Edit
func gamma[F: SomeFloat](x: F): F: Source Edit
func gcd[I: SomeInteger](x, y, z: I; args: varargs[I]): I: Source Edit
func gcd[I: SomeInteger](x, y: I): I: Source Edit
func hypot[I: SomeFloat](x, y, z: I; args: varargs[I]): I: Source Edit
func hypot[I: SomeFloat](x, y: I): I: Source Edit
func isclose(a, b: SomeFloat; rel_tol = 1e-9; abs_tol = 0.0): bool: Source Edit
func isqrt(n: Natural): int {....raises: [], tags: [], forbids: [].}: Example:

assert 2 == isqrt 5
Source Edit
func isqrt[T: SomeFloat](x: T): int {....raises: [].}: Hint: assuming x > 0 (raise RangeDefect otherwise (if not danger build))

use math.isqrt if expecting raising ValueError
Source Edit
func isqrtPositive(n: Positive): int {.inline, ...raises: [], tags: [], forbids: [].}: EXT: isqrt for Positive only, as we all know, in Python:
isqrt(0) == 0

isqrt(-int)

Source Edit
func lcm[I: SomeInteger](x, y, z: I; args: varargs[I]): I: Source Edit
func lcm[I: SomeInteger](x, y: I): I: Source Edit
func ldexp[F: SomeFloat](x: F; i: int): F {....raises: [].}: Hint: only set errno
Source Edit
func ldexp[F: SomeFloat](x: F; i: int; exc: var ref Exception): F {....raises: [].}: set exception to exc instead of raising it

exc:

set to nil if no error

set to suitable exc if some error occurs

Source Edit
func lgamma[F: SomeFloat](x: F): F: Source Edit
func log[F: SomeFloat](x, base: F): F: Source Edit
func log[F: SomeFloat](x: F): F: Source Edit
func log1p[F: SomeFloat](x: F): F: Source Edit
func log2[F: SomeFloat](x: F): F: Source Edit
func log10[F: SomeFloat](x: F): F: Source Edit
func math_is_error(x: SomeFloat; exc: var ref Exception): bool: inner usage (used by Lib/math).

Call this when errno != 0, and where x is the result libm returned. This will usually set up an exception and return true, but may return false without setting up an exception.
Source Edit
func modf[F: SomeFloat](x: F): tuple[intpart: F, floatpart: F]: Source Edit
func perm(n, k: Natural): int {....raises: [], tags: [], forbids: [].}: Source Edit
func perm(n: int): int {....raises: [], tags: [], forbids: [].}: equal to perm(n,n), returns n! Source Edit
func pow[F: SomeFloat](x, y: F): F: Source Edit
func prod[T](iterable: Iterable[T]; start = 1.T): T: Source Edit
func radians[F: SomeFloat](x: F): F: Source Edit
func remainder[F: SomeFloat](x: F; y: F): F: Source Edit
func trunc(x: SomeFloat): int: Hint: unlike C/Nim's returning float, Python's returns int and raises errors for nans and infinities
Source Edit

Templates

template py_math_isclose_impl(abs): inner use. Implementation of isclose.

mixin a, b, isinf, rel_tol, abs_tol
Source Edit

Exports

cbrt, nextafter, nextafter, ulp, UNRELIABLE_FMA, fma, fma, c_fma, c_fma, sumprod, dist, dist, sumprod, OpenarrayOrNimIter, dist, isinf, isfinite, frexp, frexp

src/pylib/Lib/n_math

Imports

Consts

Procs

Templates

Exports