ak.to_numpy
Defined in awkward.operations.convert on line 163.
- ak.to_numpy(array, allow_missing=True)
Converts array
(many types supported, including all Awkward Arrays and
Records) into a NumPy array, if possible.
If the data are numerical and regular (nested lists have equal lengths in each dimension, as described by the #type), they can be losslessly converted to a NumPy array and this function returns without an error.
Otherwise, the function raises an error. It does not create a NumPy
array with dtype "O"
for np.object_
(see the
note on object_ type)
since silent conversions to dtype "O"
arrays would not only be a
significant performance hit, but would also break functionality, since
nested lists in a NumPy "O"
array are severed from the array and
cannot be sliced as dimensions.
If array
is a scalar, it is converted into a NumPy scalar.
If allow_missing
is True; NumPy
masked arrays
are a possible result; otherwise, missing values (None) cause this
function to raise an error.
See also ak.from_numpy
and ak.to_cupy
.