Sparse logical

[ntjohnson1]

Resolves #256 and #257 but maybe raises more questions than it answers. It matches what you described as expected behavior. (Indicator tensors with 1.0, 0.0)

It raises the question on if we should track the dtype (could be a hidden variable or a property). Or even more explicitly could be set when constructing the tensor (I kind of hacked the vals in the sptensor to make things make sense; however would be nice to take a dtype argument on construction).

You mentioned in the issue we don't do boolean tensors but for equality we currently do. Should those also get the same indicator treatment?

>> import pyttb as ttb
>> T = ttb.tenones((2,2))
>> T == T
tensor of shape (2, 2)
data[:, :] =
[[ True  True]
 [ True  True]]

---

📚 Documentation preview 📚: https://pyttb--269.org.readthedocs.build/en/269/

[dmdunla]

Approved, but the bigger issue is explicitly handling dtypes for our tensor classes.

[ntjohnson1]

Do you mind filing a ticket for that issue just so we don't forget? Not sure if it's needed for 2.0 or not