rok commented on PR #33925: URL: https://github.com/apache/arrow/pull/33925#issuecomment-1451695480
Thanks you for the input and description of MATLAB's Deep Learning Toolbox @extabgrad ! > Therefore our main input would be to request that permutation and dim_names are not mutually exclusive. As per [discussion on the mailing list](https://lists.apache.org/thread/zhdkotvjczcomzooxr5jtj2g0fmsyk5m) and this proposal I believe `permutation` and `dim_names` will not be mutually exclusive. > 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? > Questions: > > 1. Could you clarify whether shape, dim_names and permutation are listed in left-to-right or right-to-left order. As in, is the contiguous dimension the right-most (C, python) or left-most (MATLAB, Fortran) dimension? `shape`, `dim_names` and `permutation` would map to a row-major (C, python) physical layout of data. Data in the underlying buffer would have row-major layout. > 2. Are 1-dimensional and 0-dimensional arrays allowed? This wasn't discussed, but the current language doesn't forbid them, so I suppose they are allowed. Do you think we should explicitly allow them? > 3. “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. -- 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: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
