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
Better congruence testing for odd arithmetic subgroups #12949
Comments
comment:2
Vincent: I'm ccing you on this ticket, since it was you who originally suggested that Hsu's algorithm should work for odd subgroups too (#11598 comment:3) |
comment:3
Hello, Sorry to answer lately. This is quite nice that it works! I am starting the review... |
Reviewer: Vincent Delecroix |
Work Issues: bring back old code + minor doc |
comment:4
I would prefer to keep both implementations (naive one vs checking relations with default to the latter one of course ;-). Then you may add a dedicated test comparing the results in Is your result available to read ? A dissertation is a legitimate reference, but one may want to read it. Why did you remove the reference of the article of Alexey G. Gorinov ? I do not claim that it is not a good idea but just that it was not the purpose of the ticket. |
comment:5
Dear Vincent, Nice to hear from you; sorry it's taken me so long to look at this in turn...
|
Changed work issues from bring back old code + minor doc to ref issue + test |
comment:6
Hi, Replying to @loefflerd:
My mistake.
I would appreciate a copy but this is not the point. If you mention the Math dissertation as a reference for the algorithm it should be available for anyone. What about arXiv ?
You are right. It would be better to keep the fast algorithm only and to implement the naive test within the Best |
comment:7
I'd need Hamilton's permission to put the work on the arxiv, I'll see if that can be got. |
comment:8
Dear Vincent, Hamilton gave his permission for the work to be put on the Arxiv, so here's a new patch that references the Arxiv preprint. There is also a test in "tests.py" which compares Best regards, David |
Changed work issues from ref issue + test to none |
comment:9
patchbot is unhappy on 5.11.beta3. Is that normal? |
Rebased to 5.11.beta3 |
comment:10
Attachment: trac_12949.patch.gz It conflicted with #14014. I've rebased it. |
comment:11
Great! I am really happy that it was possible to put it on arxiv. Everything looks fine except for the tests: I also notice that
Best, |
Work Issues: test |
comment:36
Does this ticket really need super-long doctests to test things that absolutely cannot be tested by smaller parameters? Maximum time should be a few seconds, not minutes. There is no point in running huge computations if they don't test anyting that is already tested with smaller tests. Times out on slower buildbots. |
Changed commit from |
Changed branch from |
Changed branch from u/davidloeffler/337fc958ba03cfcbc3bd9a1b7b465f3394b65cb3 to u/davidloeffler/12949 |
comment:40
I've uploaded a new version without the time-wasting tests. (For some reason, this ticket had been linked to a Git branch which "git trac" refused to read, so I just spent the entire morning re-applying the changes by hand.) New commits:
|
Commit: |
This comment has been minimized.
This comment has been minimized.
comment:42
Hello, Replace Do you know why
Vincent |
Branch pushed to git repo; I updated commit sha1. New commits:
|
comment:44
I made the changes you suggested. I agree that the return type of "order" for perm group elements looks like a bug, so I changed that, although it has nothing really to do with this ticket. |
comment:45
Good! |
Changed dependencies from #12233 to none |
Changed branch from u/davidloeffler/12949 to |
This patch relates to the
is_congruence()
method of arithmetic subgroups of SL2Z (described in terms of permutations, cf.sage.modular.arithgroup.arithgroup_perm
). For even subgroups (containing -1), there is a very fast algorithm due to Tim Hsu, but for odd subgroups we were using a much, much slower brute-force algorithm.My student Thomas Hamilton checked in his MMath thesis that Hsu's algorithm also works for odd subgroups with minor modifications. This patch implements this generalized Hsu algorithm, resulting in a speedup of about three orders of magnitude in all the examples I tried.
CC: @videlec
Component: modular forms
Author: David Loeffler
Branch/Commit:
27b7508
Reviewer: Vincent Delecroix
Issue created by migration from https://trac.sagemath.org/ticket/12949
The text was updated successfully, but these errors were encountered: