ArrayBackend.get_dtype_identifier

ArrayBackend.get_dtype_identifier(**kwargs) → str[source]

Method for getting a dtype_identifier (str) from an array or a dtype. This is used to retrieve the dtype of a custom object as a string and pass it to another backend. The dtype must actually be a dtype object pertaining to the self backend. Strings or Python types are not allowed here. Use odl.core.util.dtype_utils._universal_dtype_identifier for a general conversion from dtype-ish objects to identifiers.

Parameters

**kwargs‘array’ or ‘dtype’: This function inputs either an array OR a dtype

Returns

dtype_identifier (str)

Examples

>>> backend = odl.lookup_array_backend('numpy')
>>> backend.get_dtype_identifier(array=np.zeros(10))
'float64'
>>> backend.get_dtype_identifier(array=np.zeros(10, dtype = 'float32'))
'float32'
>>> backend.get_dtype_identifier(array=np.zeros(10, float))
'float64'
>>> backend.get_dtype_identifier(dtype=np.dtype('float64'))
'float64'
>>> backend.get_dtype_identifier(dtype=np.zeros(10, dtype = 'float32').dtype)
'float32'
>>> backend.get_dtype_identifier(dtype=np.dtype(float))
'float64'
>>> backend.get_dtype_identifier(dtype=np.dtype(float), array=np.zeros(10, float))
Traceback (most recent call last):
AssertionError: "array and dtype are mutually exclusive parameters"
>>> backend.get_dtype_identifier(np.dtype(float))
Traceback (most recent call last):
TypeError: "ArrayBackend.get_dtype_identifier() takes 1 positional argument but 2 were given"