As of this date, the module does not handle (is not aware) of sparse matrices. It would be prudent for all system matrices from FEA to be considered sparse. This necessitates the conversion to sparse format. The following notes were made in a notebook earlier. The first experience was in using mode_expansion_from_model.
It appears that the issue below in the vibrationtesting module is that sparse matrices are not handled. Slicing, multiplying, and inverting will all have to be done separately for sparse matrices. See the sparse matrices manual .
In the short term, converting them to full matrices will be good enough. In the long term, there will need to be parallel paths, one for sparse matrices, one for non-sparse. The appropriate format using scipy sparse matrices appears to be lil_matrix. On fix might be to rewrite, and automatically convert all matrices to this format before the math, then convert to the originating format afterwards to make the user happy.
As of this date, the module does not handle (is not aware) of sparse matrices. It would be prudent for all system matrices from FEA to be considered sparse. This necessitates the conversion to sparse format. The following notes were made in a notebook earlier. The first experience was in using
mode_expansion_from_model.It appears that the issue below in the vibrationtesting module is that sparse matrices are not handled. Slicing, multiplying, and inverting will all have to be done separately for sparse matrices. See the sparse matrices manual .
In the short term, converting them to full matrices will be good enough. In the long term, there will need to be parallel paths, one for sparse matrices, one for non-sparse. The appropriate format using scipy sparse matrices appears to be
lil_matrix. On fix might be to rewrite, and automatically convert all matrices to this format before the math, then convert to the originating format afterwards to make the user happy.