Split between basic and advanced indexing cannot handle all cases

[arcondello]

Consider

from dwave.optimization import Model

model = Model()

arr = np.arange(2*3*4*5).reshape(2, 3, 4, 5)
sym = model.constant(arr)


print(arr[:, 0, :, [2]].shape)
print(sym[:, 0, :, [2]].shape())

(1, 2, 4)
(2, 4, 1)

this is because we distribute the 0 index to basic indexing and [2] to advanced indexing. Basic indexing does not have a notion of what we call type 2 indexing. I.e.,

• The advanced indices are separated by a slice, Ellipsis or newaxis. For example x[arr1, :, arr2].
• The advanced indices are all next to each other. For example x[..., arr1, arr2, :] but not x[arr1, :, 1] since 1 is an advanced index in this regard.

[arcondello]

One workaround would be to turn int indexers into tuple[int] indexers. This would then use the AdvancedIndexing symbol, though we would need to follow it up with a Reshape symbol to remove the relevant dimension. And of course it would create an extra constant node.

[wbernoudy]

Isn't this more an issue of AdvancedIndexing not supporting int indexers? If it did, then the _split_indices() method could forward all those indices directly to it?