Universal Functions (ufunc)

CuPy provides universal functions (a.k.a. ufuncs) to support various elementwise operations. CuPy’s ufunc supports following features of NumPy’s one:

• Broadcasting
• Output type determination
• Casting rules

CuPy’s ufunc currently does not provide methods such as reduce, accumulate, reduceat, outer, and at.

Ufunc class

class cupy.ufunc

Universal function.

Variables:

• name (str) – The name of the universal function.
• nin (int) – Number of input arguments.
• nout (int) – Number of output arguments.
• nargs (int) – Number of all arguments.

__call__()

Applies the universal function to arguments elementwise.

Parameters:

• args – Input arguments. Each of them can be a cupy.ndarray object or a scalar. The output arguments can be omitted or be specified by the out argument.
• out (cupy.ndarray) – Output array. It outputs to new arrays default.
• dtype – Data type specifier.

Returns:

Output array or a tuple of output arrays.

types

A list of type signatures.

Each type signature is represented by type character codes of inputs and outputs separated by ‘->’.

Available ufuncs

Math operations

add subtract multiply divide logaddexp logaddexp2 true_divide floor_divide negative power remainder mod fmod absolute rint sign exp exp2 log log2 log10 expm1 log1p sqrt square reciprocal

Trigonometric functions

sin cos tan arcsin arccos arctan arctan2 hypot sinh cosh tanh arcsinh arccosh arctanh deg2rad rad2deg

Bit-twiddling functions

bitwise_and bitwise_or bitwise_xor invert left_shift right_shift

Comparison functions

greater greater_equal less less_equal not_equal equal logical_and logical_or logical_xor logical_not maximum minimum fmax fmin

Floating point values

isfinite isinf isnan signbit copysign nextafter modf ldexp frexp fmod floor ceil trunc

ufunc.at

Currently, CuPy does not support at for ufuncs in general. However, cupy.scatter_add() can substitute add.at as both behave identically.