Saturation for modules

[wdecker]

No description provided.

[wdecker]

The error which appears here in some tests stems from an example which is NOT NEW, but which I moved from a .jl to a .md file. The function called is in MPolyQuo.jl and reads a follows:

function intersect(a::MPolyQuoIdeal{T}, b::MPolyQuoIdeal{T}...) where T
  for g in b
    @req base_ring(g) == base_ring(a) "base rings must match"
  end
  as = Singular.intersection(singular_generators(a.gens), [singular_generators(g.gens) for g in b]...)
  return MPolyQuoIdeal(base_ring(a), as)
end

The result is the ideal generated by x*y to which the Singular calculation NOW adds the generator x*y^2 which is not wrong but superfluous. @hannes14, @ederc: Is there a recent change in Singular(.jl) which causes this?

[lgoettgens]

is this change intended? why don't you want to include the docstring here?

[wdecker]

Please see here:
https://docs.oscar-system.org/dev/CommutativeAlgebra/affine_algebras/#Ideals-in-Affine-Algebras
When I include the docstring, a whole Rattenschwanz of methods for intersection will be printed. These happens occasionally and so far, none of our gurus had a solution for this problem. Please note: I duplicated the example using jldoctest in the .md file, so the example will run anyway.

[lgoettgens]

I think @thofma has some tool that might help in these cases. Maybe he could try it on this instance of the problem.

[thofma]

There is no docstring for the signature

intersect(V::Vector{MPolyQuoIdeal{T}}) where T

so it just includes all of them. I think simply removing this should fix everything.

[thofma]

and keep the @docs

[thofma]

Suggested change

intersect(V::Vector{MPolyQuoIdeal{T}}) where T

[thofma]

@wdecker I wrote a little package to help with the docstrings (which is how I found out about the things mentioned earlier):

julia> using WhereIsMyDocstring, Oscar

julia> @docmatch intersect(V::Vector{MPolyQuoIdeal{T}}) where T
6-element Vector{WhereIsMyDocstring.DocStr}:
[...]

The @docmatch bla tells you which docstrings will be included, if you write this in a ```@docs section.

You can also do

julia> @docmatch intersect
25-element Vector{WhereIsMyDocstring.DocStr}:
[...]

to list all docstrings for a specific function. This also gives you instructions on how to include a specific one in the documentation (which works most of the time).

[codecov]

Codecov Report

Attention: Patch coverage is 86.66667% with 6 lines in your changes missing coverage. Please review.

Project coverage is 84.59%. Comparing base (16517cf) to head (6faf888).
Report is 4 commits behind head on master.

Files with missing lines	Patch %	Lines
src/Modules/UngradedModules/SubquoModule.jl	86.66%	6 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #4249      +/-   ##
==========================================
- Coverage   84.59%   84.59%   -0.01%     
==========================================
  Files         631      631              
  Lines       85072    85109      +37     
==========================================
+ Hits        71970    72001      +31     
- Misses      13102    13108       +6     

Files with missing lines	Coverage Δ
src/Modules/UngradedModules/SubquoModule.jl	78.72% <86.66%> (+1.19%)	⬆️

... and 1 file with indirect coverage changes

[fingolfin]

Huh, it's indeed odd: the docstring for intersect(a::MPolyQuoIdeal{T}, b::MPolyQuoIdeal{T}...) where T in src/Rings/MPolyQuo.jl:505 contains a jldoctest which does not seem to be executed for some reason?

It looks like this:

@doc raw"""
    intersect(a::MPolyQuoIdeal{T}, bs::MPolyQuoIdeal{T}...) where T
    intersect(V::Vector{MPolyQuoIdeal{T}}) where T

Return the intersection of two or more ideals.

# Examples
```jldoctest
julia> R, (x, y) = polynomial_ring(QQ, [:x, :y]);

julia> A, _ = quo(R, ideal(R, [x^2-y^3, x-y]));

julia> a = ideal(A, [y^2])
Ideal generated by
  y^2

julia> b = ideal(A, [x])
Ideal generated by
  x

julia> intersect(a,b)
Ideal generated by
  x*y

julia> intersect([a,b])
Ideal generated by
  x*y
```
"""
function intersect(a::MPolyQuoIdeal{T}, b::MPolyQuoIdeal{T}...) where T
  for g in b
    @req base_ring(g) == base_ring(a) "base rings must match"
  end
  as = Singular.intersection(singular_generators(a.gens), [singular_generators(g.gens) for g in b]...)
  return MPolyQuoIdeal(base_ring(a), as)
end

And clearly CI tests pass. But if I run this test manually in the REPL, I get different output (just like @wdecker):

julia> R, (x, y) = polynomial_ring(QQ, [:x, :y]);

julia> A, _ = quo(R, ideal(R, [x^2-y^3, x-y]));

julia> a = ideal(A, [y^2])
Ideal generated by
  y^2

julia> b = ideal(A, [x])
Ideal generated by
  x

julia> intersect(a,b)
Ideal generated by
  x*y
  x*y^2

julia> intersect([a,b])
Ideal generated by
  x*y
  x*y^2

If I run

Oscar.doctest_fix("src/Rings/MPolyQuo.jl")

it modifies the docstring to the new output.

But CI does not complain. What's going on there?

[thofma]

The following claims that this is run

https://github.com/oscar-system/Oscar.jl/actions/runs/11552275709/job/32151021946?pr=4249#step:8:2272

[hannes14]

The problem with additional element in intersect is fixed in Singular with
Singular/Singular@de8b882,
will be in Oscar with the next Singular(.jl) release

[wdecker]

thx!

…

Am 29.10.2024 um 11:18 schrieb Hans Schoenemann ***@***.***>: The problem with additional element in intersect is fixed in Singular with ***@***.*** <Singular/Singular@de8b882>, will be in Oscar with the next Singular(.jl) release — Reply to this email directly, view it on GitHub <#4249 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AASSXEVVOMKIJAA4PKNF7NLZ55OI5AVCNFSM6AAAAABQVW6P3WVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDINBTHAYDQMZZGI>. You are receiving this because you were mentioned.