[GitHub] [arrow] jorisvandenbossche commented on pull request #33925: GH-33923: [Docs] Tensor canonical extension type specification

["jorisvandenbossche (via GitHub)" <gi...@apache.org>]

jorisvandenbossche commented on PR #33925:
URL: https://github.com/apache/arrow/pull/33925#issuecomment-1451727185

   > > We would also like to ensure that the format will support complex data.
   > 
   > Just to be clear: by complex you mean diverse not complex as in complex numbers?
   
   And if you mean complex numbers: the `value_type` in this proposal can be _any_ arrow data type. At the moment, there is no "complex" data type defined in the Arrow format spec, though. But there is a proposal to add a canonical extension type for complex data (https://github.com/apache/arrow/pull/10565), and assuming that this would be included in Arrow, then you can specify a tensor array that uses complex numbers as the individual tensor elements.
   
   > > 5\. “fixed_shape_array” or variations on multi-dimensional or nd array might be a more appropriate name given the mathematical implications of the term ‘tensor’.
   > 
   > That's an interesting point. Do you know of a source where this is discussed/argued? I'd be in favor of array-like naming too. The consideration here would be that deep learning frameworks are pushing tensor name and seem to have a lot of momentum.
   
   I don't have an opinion on the whole "array vs tensor" debate, but one aspect in favor of "Tensor" is that it avoids the confusion about whether "FixedShapeArray" is actually an array or type class, and the duplication in the name of "FixedShapeArrayArray".


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: github-unsubscribe@arrow.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org