New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Strongly Regular Graphs database #18948
Comments
This comment has been minimized.
This comment has been minimized.
Branch: u/ncohen/18948 |
New commits:
|
Commit: |
Branch pushed to git repo; I updated commit sha1. This was a forced push. New commits:
|
Branch pushed to git repo; I updated commit sha1. New commits:
|
comment:5
I think you do not need to edit module_list.py file anymore when there is no dependencies. See #15410. |
comment:6
I removed it and all doctests break because of an import problem. As I do not mind much either way personally, I leave it like that. Nathann |
comment:7
this comment landed on the wrong ticket (#18960), sorry:
Huh? Do you mean
All the similar docstrings have the same problem. Namely, one tests that a (v,k,l,m)-srg is |
comment:8
To me your phrasing also sounds incorrect, as it seems to test whether "every (v,k,l,mu)-strongly regular graph is a Paley graph". What about replacing 'a' by 'some'?
Nathann |
comment:10
Do you really want to keep mu as a required by functions parameter? I would rather get rid of it, for it is a simple computation to find mu given the 1st three parameters. |
comment:11
I prefer to keep it, for it belongs to the definition of strongly regular graphs. I don't see anything wrong it making it optional: we can replace 'None' by the computed value if necessary. Nathann |
Branch pushed to git repo; I updated commit sha1. New commits:
|
Branch pushed to git repo; I updated commit sha1. This was a forced push. New commits:
|
comment:14
Replying to @sagetrac-git:
How about functions like By the way, what is the point of doing |
comment:15
If you think that this feature would be useful, I have nothing against your adding it in another ticket. I do not see the point, as those functions are not even meant to be called directly by users, who can do so with 'strongly_regular_graph' (which can automatically guess 'mu' if needed).
That's in order to be able to tell if the construction can be done without doing it (which can take quite some time). Very useful to know which entries Sage can build to compare it to the list of those that are known to exist. I used a design similar to the one used in
Admittedly, the functions of the design code are computationally much more expensive than those. Nathann |
comment:16
it seems that
only returns correct results for the non-Paley (non-conference) case (i.e. the case where the eigenvalues are not integer). This should be documented, at least. |
Branch pushed to git repo; I updated commit sha1. New commits:
|
comment:54
Easy: you add constructions of strongly regular graphs. The only thing this doctest does is check the COUNT of them (note that the individual lines naming each graph are not check, except the first, because of '...'). Given that you add constructions, the final number changes, and the doctest breaks.
Just load ONLY the branch, and you will not have any problem. If you add other commits, now, I can do no magic. Nathann |
comment:55
If you read his 'history' file you will see that he tested more commits than that. He also had some commits of hiw own patch (#18972) in his history. If you load only the branch on my tickets things work, but you can't expect me to fix in my branches the errors introduced on non-reviewed tickets. The only thing that it proves is that #18972 needs to be rebased on #18988. And that's clearly not my job. Nathann |
comment:56
comment:33 is from the buildbot and does not include any non-reviewed tickets: |
comment:57
Replying to @nathanncohen:
Oh hell... How about we add a special tag (say, countexamples) to these counting test(s), so that they don't normally run.
The problem is that nobody expects the Spanish Inquisition hitting on your tests this way. Doctesting such counts is akin to doctesting number of Sage tests that pass... |
comment:58
Regardless of that, the doctests pass, in this ticket and in those above. Nathann |
comment:60
The buildbot always merges dependencies first; Currently I'm at
|
comment:61
So that is your explanation. Yesterday, or that night, you ran the doctests on a version of Sage which included #18934. This branch, at that time, did not have #18934 as a dependency. This morning, I rebased this ticket (and those above) over #18934, and I fixed it. So now it should work, because of this rebasing. So unless you find something wrong with this branch, please set it back to its original status. Nathann |
comment:62
And this is why it shouldn't be possible to change positively reviewed branches... we can implement a cool-off period just for you if you prefer ;-) |
comment:63
As the reviewer, I hereby humbly beg for the |
comment:64
Create a poll on sage-devel. |
comment:65
If we do it now, I will have to re-merge 3 linearly ordered patches again. So let's do that later. Nathann |
comment:66
Replying to @nathanncohen:
Can you do this on #18988, or whatever ticket that has the latest commits in this chain? |
comment:67
Yeah yeah. I'm against it, but after having been complaining pointlessly for 30 minutes on a ticket because you were all testing something different from the branch stored on this ticket, I won't even bother trying to explain. Nathann |
comment:68
Easier to blame the doctest, of course. |
comment:69
Let's create another ticket for that. It's unrelated to the topic of #18988.. |
comment:70
So? Are you now convinced that there is no broken doctest in this ticket, and that it can be set back to its original status ? |
comment:71
I'm not necessarily convinced but I can try again... hopefully I'll get through a testing cycle before the branch changes again ;-) |
comment:72
Yeah, sorry for that. I should have remembered to keep such tickets in a linear order at all times. |
comment:73
Replying to @nathanncohen:
As you like. Please do. |
Changed branch from u/ncohen/18948 to |
This ticket implements a new module names
strongly_regular_db
that lets us build one example of strongly regular graph, given four integer parametes (v,k,lambda,mu).It uses Andries Brouwer's database to return more meaningful non-existence results, and help us find which constructions are missing from the database.
With a bit of luck (and time, and work) it would be great if we could reproduce all SRG that are known to exist!
The module has a simple structure:
has a simple structure:
A
seems_feasible(v,k,l,mu)
function that performs the basic artihmeticchecks to figure out if
(v,k,l,mu)
is realizable. The'apparently_feasible_parameters(n)` returns the lists of all parameters that
pass these tests for v<n. When n=1301, the set of parameters it returns is
precisely those that appear on your database (this is checked in the code).
Several functions (is_paley, is_johnson, ...) test if a given set of
parameters (v,k,l,mu) can be realized with a graph of the corresponding family
(a Paley graph, a Johnson graph, ...). If they can, they return the parameters
of that graph so that it can be built easily.
The main function
strongly_regular_graph
can be called in two ways:strongly_regular_graph(v,k,l,mu,existence=True)
answers True if such agraph is known to exists, False if it is known to be infeasible, and Unknown
otherwise.
strongly_regular_graph(v,k,l,mu)
attempts to build and return therequested graph, and returns a meaningful exception if it cannot.
This branch also updates the package 'graphs', which now ships the database in json format.
http://www.steinertriples.fr/ncohen/tmp/graphs-20150724.tar.bz2
Nathann
Depends on #18934
CC: @dimpase
Component: graph theory
Author: Nathann Cohen
Branch/Commit:
d1d25a0
Reviewer: Dima Pasechnik
Issue created by migration from https://trac.sagemath.org/ticket/18948
The text was updated successfully, but these errors were encountered: