[
https://issues.apache.org/jira/browse/ARROW-6671?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16942394#comment-16942394
]
Kenta Murata commented on ARROW-6671:
-------------------------------------
There are three tickets about adding new sparse tensor index types:
* ARROW-4225: Add CSC sparse matrix support
* ARROW-4226: Add CSF sparse tensor support
* ARROW-6485: Support the format of a COO sparse matrix that has separated row
and column indices
In ARROW-6485, we will introduce another COO sparse index that manages indices
in two vectors, the same format as scipy.sparse.coo_matrix, rather than a
single matrix, that is the current SparseCOOTensor's format.
If we can use both "matrix" and "tensor" for sparse index type names, we will
have the following sparse index types after the above tickets are closed:
* SparseCOOTensor
* SparseCSFTensor
* SparseCOOMatrix
* SparseCSRMatrix
* SparseCSCMatrix
By looking at the type names, we can recognize what types can represent tensors
with more than 2 dimensions.
> [C++] Sparse tensor naming
> --------------------------
>
> Key: ARROW-6671
> URL: https://issues.apache.org/jira/browse/ARROW-6671
> Project: Apache Arrow
> Issue Type: Wish
> Components: C++
> Reporter: Antoine Pitrou
> Assignee: Kenta Murata
> Priority: Minor
>
> Currently there's {{SparseCOOIndex}} and {{SparseCSRIndex}}, but also
> {{SparseTensorCOO}} and {{SparseTensorCSR}}.
> For consistency, it would be nice to rename the latter {{SparseCOOTensor}}
> and {{SparseCSRTensor}}.
> Also, it's not obvious the {{SparseMatrixCSR}} alias is useful.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
