Equivalent NumPy-compatible index expression.

The index expression consists of a tuple of rank slice objects that specify the lower and upper bounds for each dimension, where an infinite bound in the domain corresponds to a bound of None in the slice object.

The index expression can be used with this library as a NumPy-style indexing expression or to directly index a NumPy array.

>>> ts.IndexDomain(rank=2).index_exp
(slice(None, None, None), slice(None, None, None))
>>> ts.IndexDomain(inclusive_min=[1, 2], exclusive_max=[5, 10]).index_exp
(slice(1, 5, None), slice(2, 10, None))
>>> arr = np.array([[1, 2, 3, 4, 5], [6, 7, 8, 9, 10]])
>>> domain = ts.IndexDomain(inclusive_min=[0, 2], shape=[3, 2])
>>> arr[domain.index_exp]
array([[3, 4],
       [8, 9]])

Raises:¶

ValueError – If any finite bound in inclusive_min or exclusive_max is negative. In this case the index expression would not actually NumPy-compatible since NumPy does not support actual negative indices, and instead interprets negative numbers as counting from the end.