#11589: faster zero matrix creation
------------------------------+---------------------------------------------
Reporter: malb | Owner: jason, was
Type: enhancement | Status: new
Priority: major | Milestone: sage-4.7.2
Component: linear algebra | Keywords:
Work_issues: | Upstream: N/A
Reviewer: | Author: Martin Albrecht, Simon King
Merged: | Dependencies:
------------------------------+---------------------------------------------
Comment(by SimonKing):
I was wrong in assuming that `entries=0` would provoke a second
initialisation of the diagonal.
However, using None as default argument does have a tiny advantage: The
test `if entries is None or entries==0` is a tad faster if it is actually
None.
With `entries=None` as default in the `__call__` method of matrix spaces,
timings do improve, and I obtain a threshold `nrows==ncols==40`: For
larger matrices, over GF(p) with p>2, creation from scratch is faster,
while for smaller matrices copying is faster.
However, with the original patch, I find a big surprise: Over GF(2),
copying is ''always'' faster! For GF(p), p>2, I can confirm your statement
that the threshold seems to be around 100.
There is one problem with default `entries=None`, though: Some matrix
classes (used for matrices over the reals or symbolics) don't accept
`entries=None`. So, their init methods should be adapted as well.
Question: Is it worth it?
Using None, I obtain for example:
{{{
sage: timeit("M = matrix(GF(2),30,30)", number=10000)
10000 loops, best of 3: 105 µs per loop
sage: timeit("M = matrix(GF(2),300,300)", number=10000)
10000 loops, best of 3: 107 µs per loop
sage: timeit("M = matrix(GF(3),30,30)", number=10000)
10000 loops, best of 3: 102 µs per loop
sage: timeit("M = matrix(GF(3),300,300)", number=10000)
10000 loops, best of 3: 210 µs per loop
}}}
Using 0, I obtain:
{{{
sage: timeit("M = matrix(GF(2),30,30)", number=10000)
10000 loops, best of 3: 113 µs per loop
sage: timeit("M = matrix(GF(2),300,300)", number=10000)
10000 loops, best of 3: 123 µs per loop
sage: timeit("M = matrix(GF(3),30,30)", number=10000)
10000 loops, best of 3: 102 µs per loop
sage: timeit("M = matrix(GF(3),300,300)", number=10000)
10000 loops, best of 3: 215 µs per loop
}}}
The difference is small, but it seems quite reproducible (not flaky) for
me. So, my suggestion is to go through the matrix classes and allow
`entries=None`.
--
Ticket URL: <http://trac.sagemath.org/sage_trac/ticket/11589#comment:8>
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 post to this group, send email to [email protected].
To unsubscribe from this group, send email to
[email protected].
For more options, visit this group at
http://groups.google.com/group/sage-trac?hl=en.
