lusol factor extraction #178
Conversation
superwhiskers
force-pushed
the
lusol-factor-extraction
branch
from
fa18d0b
to
9ad838e
Compare
|
wasn't able to solve it with a rebase. i ended up making a new branch locally and cherry-picking the commits from the old one |
|
this should be ready to merge now |
There was a problem hiding this comment.
The code needs to be better commented. There are some nontrivial indexing patterns and there need to be comments explaining what is being done in main loops in L and U factor getters. Also, it is not clear what each parameter from LUSOL parameter array is. This should be explained in comments in the code.
|
This is very good work. Apart from commenting code better, this should be ready to merge. |
…ssage for duplicates match the factor extraction method being used
* remove -1-filling of arrays from lusol factor extraction code * adjustments to deduplication criterion
pelesh
force-pushed
the
lusol-factor-extraction
branch
from
e4e7b9a
to
5fffbc9
Compare
| } | ||
|
|
||
| // handle rectangular l factors correctly | ||
| for (index_type column = 0; column < diagonal_bound; column++) { |
There was a problem hiding this comment.
is this supposed to sort columns through bubble sort type algorithm?
There was a problem hiding this comment.
I believe this loop just adds ones to the diagonal of L.
There was a problem hiding this comment.
correct. that's exactly what it does
There was a problem hiding this comment.
to elaborate, this loop corrects the column pointer of each column segment by the offsets necessary to accommodate an extra element in itself and all previous columns. it then inserts the one at the end of the column's allocated space, since it
- will always be the last element in a sorted column
- would add extra computational overhead if it were placed at the beginning (since it would incur an extra n - 1 swaps or something like that) to get the column sorted
sorting of the elements lusol writes to the l factor is done in the next loop, and that is done with a kind of insertion sort, as the comment above it states
There was a problem hiding this comment.
Thanks for helpful additional explanations, @superwhiskers.
Resolves #159.
what remains to be done:
developfromlusol-integration-testsL_andU_member variables instead of allocating new ones