#15104: Special case modn_dense matrix operations to improve performance
-------------------------------------+-------------------------------------
Reporter: nbruin | Owner:
Type: enhancement | Status: needs_work
Priority: major | Milestone: sage-6.2
Component: linear algebra | Resolution:
Keywords: | Merged in:
Authors: Nils Bruin, Simon | Reviewers:
King | Work issues: regression in
Report Upstream: N/A | right_kernel_matrix
Branch: | Commit:
u/SimonKing/ticket/15104 | 0f008f9266e3ed6fbd67e7f3d357474825bdb160
Dependencies: | Stopgaps:
-------------------------------------+-------------------------------------
Comment (by nbruin):
Replying to [comment:31 nthiery]:
> Would it make sense as a short term fix to implement now the
> alternative option mentioned in the documentation of
> MatrixSpace.full_category_initialisation? After all, having an object
> whose category is not completely initialized is prone to issues, and
> it's not unreasonable to have to ask explicitly for it when needed,
> rather than the converse.
The problem there is that the person not requiring a fully initialized
matrix space isn't asking for a matrix space at all: he/she is probably
just asking for a matrix and is doing that, quite reasonably, via linear
algebra routines. Those routines *internally* find the need to
find/construct an appropriate matrix space and these routines need to be
able to operate both in time critical code where full initialization of a
never-used object is too expensive and in situations where the result
needs to be a full member of the sage universe. I suspect it'll be too
difficult to get the intent of the caller to the location where that
intent is relevant.
In cases where the same size matrices end up being created repeatedly,
ensuring that matrix spaces have a circular reference (most parents do)
should let them stick around long enough (till next GC) for weak caching
to be effective.
In number theoretic code, where one for instance needs to compute a lot of
characteristic polynomials of same-sized matrices over a lot of distinct
finite fields, the matrix spaces don't get used/reused at all, so any time
spent on them is wasted.
--
Ticket URL: <http://trac.sagemath.org/ticket/15104#comment:36>
Sage <http://www.sagemath.org>
Sage: Creating a Viable Open Source Alternative to Magma, Maple, Mathematica,
and MATLAB
--
You received this message because you are subscribed to the Google Groups
"sage-trac" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/sage-trac.
For more options, visit https://groups.google.com/groups/opt_out.
