#4495: [with patch, positive review] weight distribution for binary codes
---------------------------+------------------------------------------------
Reporter: rlm | Owner: rlm
Type: defect | Status: new
Priority: major | Milestone: sage-3.4
Component: coding theory | Resolution:
Keywords: |
---------------------------+------------------------------------------------
Comment (by mabshoff):
Replying to [comment:6 wdj]:
Hi David,
> This is a great patch. It applies cleanly and I've done lots of testing,
which it passes.
>
> I have a question though and this is what the Wall time on the following
test means?
Fixing the formatting:
{{{
for i in range(20):
C = RandomLinearCode(100, 25, GF(2))
time s1 = C.spectrum()
time s2 = C.spectrum(method="gap")
s1 == s2
}}}
The timings:
{{{
Time: CPU 0.84 s, Wall: 0.84 s
Time: CPU 0.94 s, Wall: 5.05 s
True
Time: CPU 0.87 s, Wall: 0.90 s
Time: CPU 1.05 s, Wall: 4.99 s
True
Time: CPU 0.87 s, Wall: 0.90 s
Time: CPU 0.95 s, Wall: 4.94 s
True
Time: CPU 0.84 s, Wall: 0.86 s
Time: CPU 0.96 s, Wall: 4.96 s
True
Time: CPU 0.88 s, Wall: 0.88 s
Time: CPU 0.98 s, Wall: 4.98 s
True
Time: CPU 0.85 s, Wall: 0.85 s
Time: CPU 0.92 s, Wall: 4.69 s
True
Time: CPU 0.85 s, Wall: 0.86 s
Time: CPU 0.84 s, Wall: 4.76 s
True
Time: CPU 0.84 s, Wall: 0.84 s
Time: CPU 0.98 s, Wall: 5.01 s
True
Time: CPU 0.85 s, Wall: 0.85 s
Time: CPU 0.94 s, Wall: 4.97 s
True
Time: CPU 0.88 s, Wall: 0.89 s
Time: CPU 0.94 s, Wall: 4.95 s
True
Time: CPU 0.85 s, Wall: 0.86 s
Time: CPU 0.93 s, Wall: 4.99 s
True
Time: CPU 0.85 s, Wall: 0.85 s
Time: CPU 1.00 s, Wall: 4.93 s
True
Time: CPU 0.93 s, Wall: 0.95 s
Time: CPU 1.03 s, Wall: 4.97 s
True
Time: CPU 0.85 s, Wall: 0.85 s
Time: CPU 1.04 s, Wall: 4.92 s
True
Time: CPU 0.84 s, Wall: 0.85 s
Time: CPU 1.02 s, Wall: 5.00 s
True
Time: CPU 0.86 s, Wall: 0.87 s
Time: CPU 0.85 s, Wall: 4.66 s
True
Time: CPU 0.85 s, Wall: 0.86 s
Time: CPU 0.83 s, Wall: 4.59 s
True
Time: CPU 0.85 s, Wall: 0.85 s
Time: CPU 0.86 s, Wall: 4.75 s
True
Time: CPU 0.91 s, Wall: 0.91 s
Time: CPU 0.90 s, Wall: 4.73 s
True
Time: CPU 0.91 s, Wall: 0.92 s
Time: CPU 0.88 s, Wall: 4.71 s
True
}}]
>
> I wonder if this means that GAP's kernel computation (method="gap" is
the slowest of the three)
What three? I see only two computations.
> beats binary some percentage of the time but GAP's interface takes a
long
> time to parse that information back to Sage (via pexpect and whatever
> fiddling GAP does), as indicated by the Wall time?
That means the new code is beating the pants off GAP+Guava:
{{{
Time: CPU 0.91 s, Wall: 0.92 s
Time: CPU 0.88 s, Wall: 4.71 s
}}}
What counts it the total time, i.e. about 0.92s vs. 4.71s. The first line
tells us that the new code spends all its time in Sage while the second
line tells us that 4.71-0.88=3.83s were spend in GAP. I assume if we pick
larger examples the favor will shift toward Robert's implementation, but
that needs to be tested. One aspect here might be that the pexpect
transfer to and from GAP is inefficient, but that can also easily be
determined.
Cheers,
Michael
--
Ticket URL: <http://trac.sagemath.org/sage_trac/ticket/4495#comment:7>
Sage <http://sagemath.org/>
Sage - Open Source Mathematical Software: Building the Car Instead of
Reinventing the Wheel
--~--~---------~--~----~------------~-------~--~----~
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
-~----------~----~----~----~------~----~------~--~---
