#19661: the srgs from Cossidente and Penttila construction of hemisystems in
H(3,q^2)
-------------------------+-------------------------------------------------
Reporter: | Owner:
dimpase | Status: needs_review
Type: | Milestone: sage-7.0
enhancement | Resolution:
Priority: major | Merged in:
Component: graph | Reviewers:
theory | Work issues:
Keywords: | Commit:
Authors: | 16f1f711fce54ae770c81dd957e5cad2baea8fe1
Report Upstream: N/A | Stopgaps:
Branch: |
public/19661 |
Dependencies: |
-------------------------+-------------------------------------------------
Comment (by dimpase):
Replying to [comment:29 ncohen]:
> Okay, before anything let's be clear: I won't give a review to this
patch if it injects 10 lines of GAP code. Perhaps you will find somebody
who will, but I will not. Now let us discuss.
>
> > Nothing fuzzy - it's the right tool for the job. You don't know a
better one, right? In fact many functions in this file would've become
much cleaner, had I known about it then...
>
> It may be a necessary tool for libgap, and I quite agree that this
function is necessary. I claim that it is not necessary here, and in
particular that the mathematical features you need from gap should not be
called *directly* from gap, and that we should instead expose in the Sage
mathematical objects those features from GAP that you need.
No, we don't need to force objects to cross the still nontrivial Sage/GAP
border unnecessarily. There is absolutely no reason for this piece of GAP
code, that takes a number as an input, and returns a graph, to expose its
finite fields and matrix groups guts to Sage, for this exposure does not
benefit anything here, it only degrades the performance.
>
> This code could then be !Sage/Python code that calls other !Sage/Python
functions. Those other functions may have reasons to rely on this
`function_factory` function, or they may not. I have absolutely no idea a
priori, and neither do I have any objection to that a priori.
You might see that the code already calls a !Sage/Python function, which
relies on this `function_factory` function. In that sense we are already
here, and I don't see what you're trying to tell me here.
>
> > > What I mean is that there are apparently some mathematical
computations tht you do not know how to perform with Sage only.
> >
> > yes, I do know how to do them in Sage, in C, or on a Turing machine if
I must...
> > But it's suboptimal, for reasons of efficiency, and of the code
length.
>
> We try to build a software that can handle all kind of maths. If you
must inject GAP code to get it to do what you want, then it means that we
should expose some feature somewhere. At least that's how I interpret it.
>
> > Well, every time I talk to you guys about a graph backend that can do
things with symmetric graphs as good as GRAPE does it in Sage, I don't
hear cheers and screams "yes, let's do it!" I rather hear "hmm, how do we
remove an edge then" or something equally compelling...
>
> I don't feel guilty of not writing your code for you.
>
> On the other hand, I have to admit that I still do not know what you
exactly want this backend to be. So no, I don't feel terribly overjoyed
when you answer my "you code is ugly" remarks with "it's because we don't
have a backend for symmetric graphs".
yes, we create graphs in GAP, and copy them into Sage, completely ignoring
the underlying symmetries. Instead, we could be using GRAPE/GAP Graph as a
backend. We could be handling sufficiently symmetric dense graphs on
10000 vertices with ease.
E.g. with GAP/GRAPE it takes about 3 sec to create the graph on the ticket
for q=13, it's 15386 vertices...
(testing adjacency is still instant, by the way)
--
Ticket URL: <http://trac.sagemath.org/ticket/19661#comment:34>
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 https://groups.google.com/group/sage-trac.
For more options, visit https://groups.google.com/d/optout.
