math

Real-number functions and constants. Most functions accept an int or float and return a float; exceptions are noted below. Arguments to native functions are positional.

Available API

Group	Members
Constants	`pi`, `e`, `inf`, `nan`
Rounding	`ceil(x)`, `floor(x)`, `trunc(x)` return integers; `fabs(x)` returns a float.
Aggregation	`fsum(values)` for a list of numbers; `gcd(a, b)` for two integers.
Classification	`isfinite(x)`, `isinf(x)`, `isnan(x)`, `isclose(a, b)`
Powers/logarithms	`exp(x)`, `log(x)`, `log(x, base)`, `log2(x)`, `log10(x)`, `pow(x, y)`, `sqrt(x)`, `cbrt(x)`
Trigonometry	`sin(x)`, `cos(x)`, `tan(x)`, `asin(x)`, `acos(x)`, `atan(x)`, `atan2(y, x)`
Angles	`degrees(radians)`, `radians(degrees)`
Decomposition	`modf(x)`, `fmod(x, y)`, `copysign(x, y)`
Combinatorics	`factorial(n)` for a nonnegative integer.

Trigonometric arguments and results use radians. modf(x) returns (fractional_part, integer_part), both floats; pass a float to this function. copysign(x, y) takes the magnitude of x and the sign of y.

import math

assert math.ceil(1.2) == 2
assert type(math.floor(1.8)) is int
assert math.gcd(18, 24) == 6
assert math.fsum([0.1] * 10) == 1.0
assert math.isclose(math.sin(math.pi / 2), 1.0)
assert math.modf(2.5) == (0.5, 2.0)

Differences from CPython

fsum accepts a list, not an arbitrary iterable, and uses compensated summation rather than CPython's full algorithm.
isclose(a, b) checks abs(a - b) < 1e-9. It has no relative or absolute tolerance keyword arguments; infinities do not compare close by this rule.
gcd takes exactly two arguments.
factorial is limited by signed 64-bit integers. Keep n <= 20 for an exact representable result.
Domain handling follows the implementation's math functions and may produce NaN or infinity instead of CPython exceptions.

For complex values, use cmath. Implementation: math.c.