tensorplay.functional.float_power

tensorplay.functional.float_power(input, exponent, *, out=None) → Tensor[source]

Raises input to the power of exponent, elementwise, in double precision. If neither input is complex returns a tensorplay.float64 tensor, and if one or more inputs is complex returns a tensorplay.complex128 tensor.

Note

This function always computes in double precision, unlike tensorplay.pow(), which implements more typical type promotion. This is useful when the computation needs to be performed in a wider or more precise dtype, or the results of the computation may contain fractional values not representable in the input dtypes, like when an integer base is raised to a negative integer exponent.

Parameters:

• input (Tensor or Number) – the base value(s)

• exponent (Tensor or Number) – the exponent value(s)

Keyword Arguments:

out (Tensor, optional) – the output tensor.

Example:

>>> a = tensorplay.randint(10, (4,))
>>> a
tensor([6, 4, 7, 1])
>>> tensorplay.float_power(a, 2)
tensor([36., 16., 49.,  1.], dtype=tensorplay.float64)

>>> a = tensorplay.arange(1, 5)
>>> a
tensor([ 1,  2,  3,  4])
>>> exp = tensorplay.tensor([2, -3, 4, -5])
>>> exp
tensor([ 2, -3,  4, -5])
>>> tensorplay.float_power(a, exp)
tensor([1.0000e+00, 1.2500e-01, 8.1000e+01, 9.7656e-04], dtype=tensorplay.float64)