add od_from_atlas_group, od_from_p_subgroup, and helpers
#3444
Conversation
| # Check equality of positive results in small cases. | ||
| l = all_od_infos(comment_matches => "AGR", dim => 1:8); | ||
| for entry in l | ||
| chi = Oscar.OrthogonalDiscriminants.character_of_entry(entry) |
There was a problem hiding this comment.
Tests fail in this line:
Got exception outside of a @test
MethodError: no method matching getindex(::Nothing, ::Int64)
Stacktrace:
[1] character_of_entry(d::Dict{Symbol, Any})
@ Oscar.OrthogonalDiscriminants ~/work/Oscar.jl/Oscar.jl/experimental/OrthogonalDiscriminants/src/data.jl:234
[2] macro expansion
@ ~/work/Oscar.jl/Oscar.jl/experimental/OrthogonalDiscriminants/test/direct.jl:5 [inlined]
|
We need |
benlorenz
added
the
backport 1.0.x
to the dictionaries returned by `all_atlas_group_infos`
i.e., prescribe a minimal characteristic for the returned group, in order to guarantee that the reduction of an irreducible group is also irreducible
- improve coercion from quadratic fields into `QQAbField`; the check for `is_quadratic_field` is not general enough, and evaluation of a field element requires not just some polynomial ring but the underlying polynomial ring of the number field - and add `_embedding` of `QQ` into `QQAbField`, for convenience
- add `known_class_fusions` - add optional argument to `induced_cyclic` - add `od_from_p_subgroup` - add helper `possible_permutation_characters_from_sylow_subgroup`
ThomasBreuer
force-pushed
the
TB_OD_pgroups
branch
from
ec986fd to
ca61976
Compare
| # Compute the orbits on rows. | ||
| row_orbits = Vector{Int}[] | ||
| reps = [] | ||
| poss = [] |
There was a problem hiding this comment.
whenever I see [] I am think "not type stable, beware". We see a bunch of these in this PR. But perhaps it is fine for other reasons -- just wanted to draw your attention to it.
|
@ThomasBreuer I just tried to backport this to finally get that part working but the return value for In the book ( julia> ch = character_table("R(27)")[16];
julia> degree(ch)
18278
julia> Oscar.OrthogonalDiscriminants.od_from_p_subgroup(ch, 3)
(true, "-3")Running this on the current Oscar master gives: julia> ch = character_table("R(27)")[16];
julia> degree(ch)
18278
julia> Oscar.OrthogonalDiscriminants.od_from_p_subgroup(ch, 3)
(false, "") |
|
@benlorenz I cannot reproduce the problem. |
|
@benlorenz Ah, now I understand what happens. In my situation, a newer version of the GAP package CTblLib is used. |
* add `conductor` for class functions * add `:constituents` key (if available) to the dictionaries returned by `all_atlas_group_infos` * support `min_char` in `isomorphic_group_over_finite_field` i.e., prescribe a minimal characteristic for the returned group, in order to guarantee that the reduction of an irreducible group is also irreducible * better handling of quadratic fields - improve coercion from quadratic fields into `QQAbField`; the check for `is_quadratic_field` is not general enough, and evaluation of a field element requires not just some polynomial ring but the underlying polynomial ring of the number field - and add `_embedding` of `QQ` into `QQAbField`, for convenience * added `orbits_group_automorphisms` and helpers * add `od_from_atlas_group` and helper * add p-group criterion - add `known_class_fusions` - add optional argument to `induced_cyclic` - add `od_from_p_subgroup` - add helper `possible_permutation_characters_from_sylow_subgroup` * adjust to the changes in Oscar since November * now the tests pass again * `FiniteField` does no longer exist * use `number_of_conjugacy_classes` * omit some tests that require AtlasRep at least 2.1.7 * for the moment, leave out tests that need web access (cherry picked from commit 556be5a)
benlorenz
removed
the
backport 1.0.x
- Add QQBar docs to the manual #3423 - do not show the OscarInterface banner #3422 - fix bugs in all_OD_infos #3419 - Ep/ Rename Spec to AffineScheme #3345 #3425 - Remove two mentions of Arb_jll #3431 - Tweak epimorphism_from_free_group #3430 - CI: re-enable nightly #3435 - support gen(G::GAPGroup, 0) #3332 - Align all_*_groups methods some more #3433 - Add all_perfect_groups #3434 - Add all_primitive_groups and all_transitive_groups variants taking a single int or int range #3404 - fix a docstring #3436 - Fixes multivariate division #3396 - Docu invariants tori #3428 - Improve docstrings for is_conjugate/is_conjugate_with_data. #3384 - Fix ambient_module(M::SubquoModule) #3448 - Bugfix for printing of affine schemes #3437 - Bugfix for bugfix for printing of affine schemes #3445 - Update OSCAR banner #3410 - Docu invariants lin. red. groups (Lakshmi Ramesh and Wolfram Decker) #3443 - add od_from_atlas_group, od_from_p_subgroup, and helpers #3444 - Unexport normalise #3453 - support group properties for character tables #3449 - add docstrings for acting_group and action_function #3432 (exports are used in new groups code for the book) - Adjust to renaming of rank(A::FinGenAbGroup) to torsion_free_rank(A::FinGenAbGroup) #3457 - Ensure fp_group(G) transfers group attributes #3464 - Added comment on convention #3467 - Export weierstrass_chart_on_minimal_model and patch transform_to_weierstrass #3458 - Fix a doc signature #3466 - Grading + caching for affine algebra of torus invariants #3469
@ThomasBreuer Does that mean that at some point there will be a new CTbLib package that will be automatically picked up by the current Oscar (without any code changes here) and corrects this output? |
|
@benlorenz Sorry for not being precise. Forcing a newer version of CTblLib in Oscar does not work because it gets loaded already at startup of GAP. (Installing something at runtime is not a good idea. Nevertheless, currently we do this for the "repsn" package. Here the situation is different, this package does not get loaded automatically by GAP, thus Oscar can first try to upgrade and then load the package.) |
|
Maybe we could do something like oscar-system/GAP.jl#855 and make a new package tarball that contains a newer CTblLib (but preferrably not many other changes) for GAP.jl. And then release this in a GAP.jl patch release 0.10.3. This would get picked up by the Oscar 1.0 release automatically and could fix the code that we want to have in the book? (Needs to be tested properly to make sure we don't break anything ...) |
|
We could do that, sure. Or we wait 1-2 weeks, then GAP 4.13.0 will be out, and will be used in GAP.jl 0.11, and we can use that in OSCAR 1.0.1 or 1.1? |
|
Can't really say how much the GAP minor version bump might affect Oscar and whether we can put this in a patch release but otherwise yes that would work as well. |
od_from_atlas_group(chi)tries to find a suitable representation affordingchiin the Atlas of Group Representations, and then to compute the orth. disc. from itod_from_p_subgroup(chi, p)tries to use the restriction of chi to the Sylowp-subgroupknown_class_fusionsinduced_cyclicpossible_permutation_characters_from_sylow_subgrouporbits_group_automorphismsis used to reduce the tests to only one member of each orbit on irreducible charactersmin_charargument inisomorphic_group_over_finite_fieldis used to guarantee that the reduction of an irreducible module is irreducibe@fieker These changes are essentially the ones from #3005, which got closed for technical reasons.