pe DiGraph. Then
Poset would have _f(x), _g(x) and so on as an implementation.
For example minimal_elements() of a Poset is just a sources() of DiGraph
wrapped. Do we need two-phase wrapper?
--
Jori Mäntysalo
On Thu, 21 Jan 2016, jhonrubia6 wrote:
Does it makes sense to you if I modify the plot.py docstrings in order to
include PLOT:: blocks? Wouldn't it be more clear for newbies?
Of course it would. And same applies to different plot()-functions; at
least graphs and posets have these.
--
Jori
Is it just me, or do you also get an error from
./sage --docbuild --warn-links all html
?
--
Jori Mäntysalo
On Wed, 20 Jan 2016, Dima Pasechnik wrote:
Here is a ticket with the new nauty release, and a proposal to make it a
standard package.
I vote 'yes'.
--
Jori Mäntysalo
._hasse_diagram
and then computing on H, using H.neighbors_in(...) and so on.
--
Jori Mäntysalo
[polynomia] IndexError: list assignment index out of range
Is there a quick fix for this.
* * *
This is something I would like to see tested in every release. We don't
have a checklist for this, I think.
--
Jori Mäntysalo
trying to read docstrings from functions
which have no docstrings, and these seem to be causing the errors.)
Why we have functions without docstring?
(And can Sphinx be used to found function without, say, EXAMPLES-blocks?)
--
Jori Mäntysalo
There is a quite long discussion between me and Nathann at
http://trac.sagemath.org/ticket/19659
Basic question is the rationale behing HasseDiagram-Poset -classes. What
kind of functions are meant to be implemented in HasseDiagram?
--
Jori Mäntysalo
On Fri, 15 Jan 2016, Volker Braun wrote:
The trac vm disk was full again.
df -h | mail -s "Disk usage report" some.addr...@somewhere.net
to cron.weekly? Kind of poor man's Nagios, but better than nothing.
--
Jori Mäntysalo
/sage-6.10 , i.e. nothing. Or found a
sage-release telling that bug "strongly_regular_graph() crashes when mu=0"
is corrected. "Hmmm... graph and this was something about math. I think I
took graph theory course 25 years ago."
--
Jori Mäntysalo
poset and lattice code. In
that part I think Sage has largest coverage of functions, even with good
margin to competitors.
--
Jori Mäntysalo
to deprecate any
"mobius" instances to "moebius", with the usual period.
OK. Now there is 122 lines in the source containing "mobius" and 92
"moebius". Maucalay2 and Mathematica posets packages uses "moebius". So I
will make a patch to do a shift from mobius to moebius.
--
Jori Mäntysalo
account. And more, a new user with same username
will see them. (Btw, that happens also on plain Linux, if you delete a
user with largest uid and there was files not looked at by userdel -r.)
--
Jori Mäntysalo
ic usage:
2) Can Sage compute Levenstein distance or words?
3) The reason? See (already reviewed)
http://trac.sagemath.org/ticket/19852 .
--
Jori Mäntysalo
On Sat, 9 Jan 2016, Dima Pasechnik wrote:
all what "# optional - blah" does is to look up "blah" in the --optional
command line argument.
So you can doctest stuff dependent on "blah", too.
Ah, OK. So I was (once again) thinking too complicated. Thanks.
--
Jori Mäntysalo
but
still you might want to install some package that is not included there."
I actually tested some semigroup package some time ago IIRC.
--
Jori Mäntysalo
Now we have .mobius_function() at posets and .moebius_algebra() at
lattices. I see this as an error.
Native english speakers, which one seems better?
--
Jori Mäntysalo
t;Did I see wrong?" - "Should I get married
with grandfather?".)
--
Jori Mäntysalo
additional packages
installed to a "sub-program" of Sage, like some R pacakge?
--
Jori Mäntysalo
l and have internal user management. (But to be
honest, they do not have internal quota AFAIK.)
--
Jori Mäntysalo
of these features are going to be on notebooks talked about in this
list? As an example: Will it be possible to restrict memory use of
students to lower level than staff?
--
Jori Mäntysalo
o throught wishlists in
any case?
--
Jori Mäntysalo
.
--
Jori Mäntysalo
Isn't it an indicator of bad code quality if the software works in 32-bit
but not in 64-bit or vice versa?
just my two cents...
--
Jori Mäntysalo
ort for 32-bit?
--
Jori Mäntysalo
, as this one is not a clear bug.
(Compare to http://trac.sagemath.org/ticket/18936 which just needs someone
to check that I resolved the conflict correctly. It has waited that for
five weeks.)
--
Jori Mäntysalo
On Sun, 27 Dec 2015, Travis Scrimshaw wrote:
Poset({1:[2], 2:[3], 4:[5]}).show()
I have straight lines for each cover relation with 7.0.beta1
With or without dot2tex? I have it installed.
--
Jori Mäntysalo
:[5]})
P.hasse_diagram().plot(layout='acyclic')
Your explanation is very good. So if somebody (not me!) wants to touch
this code, then it should split the poset to connected components, draw
them, and then adjoin the pictures.
--
Jori Mäntysalo
On Sun, 27 Dec 2015, Dima Pasechnik wrote:
make distclean && make
Thanks, that worked.
--
Jori Mäntysalo
?
--
Jori Mäntysalo
Please try in Sage 6.10 with dot2tex installed:
Poset({1:[2], 2:[3], 4:[5]})
I expect to have two direct lines, but instead longer line bends at
element 2. Is there some reason for this?
--
Jori Mäntysalo
it for non-integer arguments.
--
Jori Mäntysalo
On Sun, 13 Dec 2015, Nathann Cohen wrote:
There is no bliss_version() function, but you can obtain the version
of any installed package from the output of installed_packages()
Thanks. The non-working version is bliss-0.72.p1, on Sage 6.9.
--
Jori Mäntysalo
system now.
--
Jori Mäntysalo
_version() -commands.)
On production system (i.e. on 6.9, not beta) as a normal user:
. . .
self.mirrorfile = open(self.filename, 'w+t')
IOError: [Errno 13] Permission denied: . . .
. . .
So a user can not check the versions of bliss package installed.
--
Jori Mäntysalo
::vector&, unsigned int&,
bliss::Partition::Cell*&): Assertion `sh_return' failed.
/tmp2/sage-6.9/local/lib/python2.7/site-packages/sage/ext/interrupt/interrupt.so(+0x3d45)[0x7fea4f830d45]
. . .
--
Jori Mäntysalo
re looking that.)
--
Jori Mäntysalo
?
--
Jori Mäntysalo
fomance
tips" with Nathann Cohen's suggestions of installing packages like bliss.
--
Jori Mäntysalo
ave an example where this is *not* the case already?
On SageNB
G = graphs.PetersenGraph()
G.automorphism_group(algorithm='bliss')
says
Traceback (click to the left of this block for traceback)
...
ImportError: You must install the 'bliss' package to run this command.
--
Jori Mäntysalo
I have no idea about how to do this, only a clear vision of what should be
done.
--
Jori Mäntysalo
d this sounds complicated. And there are many places where this will not
help: use sparse matrix, have %cython etc.
--
Jori Mäntysalo
Does anyone know javascript here? We have graph_editor() that works nicely
when starting from scratch. (But forgots labels when modifying existing
graph.)
But there is no way to draw a DIgraph, and it is a little irritating.
--
Jori Mäntysalo
shape of vertices one-by-one, as now we can set colors.
--
Jori Mäntysalo
?
Restart the discussion? Close the ticket as wontfix?
--
Jori Mäntysalo
if there is no bottom, no
need to check that again at every step.
True. But I guess that it is nice to show right error also then. I'll try
to do that without re-checking.
--
Jori Mäntysalo
On Thu, 12 Nov 2015, Volker Braun wrote:
Just define your own exception
class LatticeMeetException(ValueError):
def __init__(self, msg, x, y):
...
Tried something: http://trac.sagemath.org/ticket/19163 . May belong under
"bad examples" -header on Python style book
r x=1 y=2"
but exception with value more complicated than a string.
Question: Will this break something? I.e. can "internal" classes raise
these kind of exceptions?
Example implementation: http://trac.sagemath.org/ticket/19163 . NOT in
needs_review yet.
--
Jori Mäntysalo
() returns empty graph, but that's different because graphs
are mutable.
--
Jori Mäntysalo
e.g. in
ValueError("No meet for x=%s y=%s"%(x,y), x, y)
It sounds stupid to first generate a string and then parse it, even if the
string is very simple.
(Sidenote: Actually LatticePoset() could raise error giving FOUR elements,
those that form "forbidden X" that makes a poset non-lattice.)
--
Jori Mäntysalo
On Sun, 8 Nov 2015, Jori Mantysalo wrote:
Here is a result of quick grepping:
algorithm 382
method 37
...and here is the link to trac: http://trac.sagemath.org/ticket/6094 . So
part of those are already done.
--
Jori Mäntysalo
Anything new on this? Will 6.10 (at last) remove SSLv3?
--
Jori Mäntysalo
that will help user.
OTOH {[1,2]:1} will give basically same error message without further
help.
(In ideal world interactive loop would of course detect also this... And
give again some good help text. But that won't happen, I guess.)
--
Jori Mäntysalo
It seems that there is different opinion of exceptions format. See
http://trac.sagemath.org/ticket/19542 . More views?
(We need vote.sagemath.org. And this is not a joke. A www-service to make
a query and to vote between options.)
--
Jori Mäntysalo
and "solver". I would vote to unify all these to "implementation".
Here is a result of quick grepping:
algorithm 382
backend 3
impl 4
implementation 29
method 37
solver 55
Maybe first thing could be change method -> algorithm? (With
deprecation, of course.)
--
Jori Mäntysalo
int.py should be corrected.
3) Should we use keyword "algorithm" or "implementation"? Or have both
with different meaning?
--
Jori Mäntysalo
;algorithm"? "backend" clearly refers to some
bigger difference, like different data structure, and applies mostly to
functions creating some object.
And this is another thing to add to developer manual.
--
Jori Mäntysalo
On Fri, 6 Nov 2015, John H Palmieri wrote:
As I said in the ticket:
[X] 'foo?' should NOT display TESTS blocks.
--
Jori Mäntysalo
cket/19517 , but I suppose
that it is not really related.)
--
Jori Mäntysalo
On Wed, 4 Nov 2015, William Stein wrote:
Does this mean computers aren't constantly getting faster anymore? l miss
the 90s... :-)
# long time - 90bogoseconds
?
--
Jori Mäntysalo
On Sun, 1 Nov 2015, Dominique Laurain wrote:
I don't know the math subject (lattice, poset...)..but...is this paper
interesting ?
www.emis.de/journals/MB/125.2/mb125_2_1.pdf
Uh. Example code is ready for Obfuscated C Code Contest. But trying to
understand...
--
Jori Mäntysalo
first 10 lines of a
file starts with any of r""" nodoctest - -". So can we just add those
extra tests to files marked with that?
--
Jori Mäntysalo
Does anyone have some algorithm for generating random lattices of size,
say, 15..50 elements? If so, I could code it and maybe add to SageMath.
--
Jori Mäntysalo
in the docstring, I can imagine it working.
Stupid question, but is there any list of possible blocks? (I.e. how well
structured our documentation is?) Artificial example: if I add functions
doing some numerical computation, can I just add
NUMERICAL STABILITY::
-blocks?
--
Jori Mäntysalo
will be used in a test but not shown.
--
Jori Mäntysalo
On Thu, 29 Oct 2015, John H Palmieri wrote:
Should it hide "TESTS" blocks? Same questions when you ask for
interactive help via "foo?".
foo??? should do what foo?? does now, i.e. show the source. foo?? should
do what foo? does now, and foo? should hide TESTS, ALGORITHM a
robi
'foo'
--
Jori Mäntysalo
?? ?
--
Jori Mäntysalo
of function one must put
sorted() to output or do someting like (1 in result, 2 in result) -->
True.
Should we show them to user? I think that this could be argued to both
directions.
--
Jori Mäntysalo
Stupid question, but is there a page that lists ways to specify colors for
mathplotlib? "black", "#00", "0.0", what else?
Compare to http://matplotlib.org/api/markers_api.html that is linked on
plot() function of graphs.
--
Jori Mäntysalo
n the core that is not free???
--
Jori Mäntysalo
This question did not get any answer, but neither anybody has made a
ticket to change this:
Does somebody want LatticePoset({42:[]}) == Poset({42:[]}) to return
False?
--
Jori Mäntysalo
e no
loops), and not twice that amount.
Whatever. I can give reasons to do it either way...
But we should have a documented unified policy to this. As an example,
if Graph(.., f) assumes f to be symmetric without checking, then
Poset(..., f) should assume that f does not define a loop on eleme
new on this?
--
Jori Mäntysalo
phism, due to McKay (1981) - - While
they seem to perform well on random graphs, a major drawback of these
algorithms is their exponential time performance in the worst case."
So maybe we should have both current algorithm and Nauty.
--
Jori Mäntysalo
! Anybody else to volunteer...?
--
Jori Mäntysalo
significance.
It started with http://trac.sagemath.org/ticket/14110 . I jumped in and
now we kind of have code to generate posets faster. But it can not be
integrated into Sage.
--
Jori Mäntysalo
of source and try
docbuild then?
--
Jori Mäntysalo
or similar.
--
Jori Mäntysalo
ve noticed that building docs starts as
usual, but older hardware got almost stuck in some later phase of
docbuild. I guess that it is some kind of crosslink checking or
integrating some previously made parts.
--
Jori Mäntysalo
(ncpus=1) gave
CPU time: 0.07 s, Wall time: 47.23 s
--
Jori Mäntysalo
in Lc:
L = LatticePoset(x)
if not foobar(L):
print "Counter-example!"
L.show()
break
but this must be modified to
i=0
for x in Lc:
i += 1
if i%10: gc.collect()
. . .
---
(*) Would be a good addition to SageMath...
--
Jori Mäntysalo
This is OK for me. But usually it seems better to say something like
def foo(self, x):
"""
Return foo value of x of the thing.
For a thing `A` and a real number `x` we define foo value to be
... where `y` is the number of elements in `A` such that `x` ...
"""
--
Jori Mäntysalo
to release memory. Or --- will gc.collect() do this
also?
--
Jori Mäntysalo
Is this a feature or a bug?
a = Poset({1:[]})
b = LatticePoset({1:[]})
c1 = Poset(b)
c2 = Poset(b.hasse_diagram())
a==c1, a==c2
outputs False, True. But this is quite a surprise for a user.
--
Jori Mäntysalo
of mathematics, where Sage would
be quite good but not excellent?
(If growing wheat is a strong point, next try oak - no carrots.)
--
Jori Mäntysalo
clean the cache" or something like that? Week ago I
wanted to run a little test over all posets of 14 elements. I had to do it
part by part, as it ate all memory. (There are about 16 million of those.)
--
Jori Mäntysalo
les?
--
Jori Mäntysalo
= P.width()
i += 1
Compare with P.height() that does not eat memory.
--
Jori Mäntysalo
in
Sage? Where it now is The Software(tm) to use?
And how could we expand those to some near area?
--
Jori Mäntysalo
similar for "binary power rings"?
--
Jori Mäntysalo
/* and most elsewhere... (yeah yeah, was already
rejected.)
--
Jori Mäntysalo
ept that INPUT shall
not contain `self`.)
I like "the thing" instead of "``self``", but whatever is OK to me, as
long as it is unified.
--
Jori Mäntysalo
for CGI? Linux? Ubuntu, Centos, Fedora...? Normal
Apache, something else?
--
Jori Mäntysalo
uot;n-1 in NN" gives bad error message if n is
for example string. "n in NN and n>0"?
--
Jori Mäntysalo
integers, nonnegative integers and positive integers. I am quite sure that
there are places for each of those.
(Of course there are other restrictions too, like >= 3 for DiamondPoset.
But I guess that they are rare.)
--
Jori Mäntysalo
inatorics?
2) What about `self` in docstrings? I have avoided it, as was suggested to
me maybe a year ago. I write for example "Return True if the poset is
connected - -", i.e. "the poset" instead of "``self``".
--
Jori Mäntysalo
eterminant() for matrices are good
examples here. We do not have inv(), even if it would save four letters
and is very common operation. We have det() as an alias because it used in
mathematical texts.
--
Jori Mäntysalo
ever abbreviate to save only few letters. Always
mention commonly used abbreviations and aliases in docstring."
--
Jori Mäntysalo
. At
least we should not use something like 'sublats' for 'sublattices' and so
on. On the other hand (as I have said before), Posets.PentagonPoset()
could just be Posets.Pentagon().
--
Jori Mäntysalo
301 - 400 of 508 matches
Mail list logo