Open madeleineudell opened 10 years ago
Currently the transpose of the shared sparse matrix is computed in serial by calling transpose on a local version of the matrix: transpose(A) = share(transpose(localize(A))). We should be able to transpose much more quickly using parallelism.
transpose(A) = share(transpose(localize(A)))
@madeleineudell I was just going to open an issue about this :)
Thanks for the workaround. I am trying to use this package for a sparse coding problem in video using ADMM.
Currently the transpose of the shared sparse matrix is computed in serial by calling transpose on a local version of the matrix:
transpose(A) = share(transpose(localize(A)))
. We should be able to transpose much more quickly using parallelism.