Updates AlgebraicSolving

[ederc]

• Fixes Rings/solving.jl: Tests getting stuck / taking very long #3956 and reenables corresponding test.
• Enables computation of Gröbner bases over QQ w.r.t. degrevlex ordering via msolve.
• FIxes several minor bugs when computing real resp. rational solutions.

[fingolfin]

Thank you. Perhaps the docstrings could be adjusted to reflect the changed defaults for the algorithm kwarg, but otherwise this seems fine (and the weird CI issue seems to be gone now?)

[lgoettgens]

This now has conflicts

[codecov]

Codecov Report

Attention: Patch coverage is 53.33333% with 7 lines in your changes missing coverage. Please review.

Project coverage is 84.65%. Comparing base (b5d0b43) to head (a2a3831).
Report is 3 commits behind head on master.

Files with missing lines	Patch %	Lines
src/Rings/groebner.jl	53.33%	7 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #3987      +/-   ##
==========================================
- Coverage   84.66%   84.65%   -0.02%     
==========================================
  Files         626      626              
  Lines       84210    84226      +16     
==========================================
+ Hits        71300    71303       +3     
- Misses      12910    12923      +13     

Files with missing lines	Coverage Δ
src/Rings/groebner.jl	89.35% <53.33%> (-0.92%)	⬇️

... and 4 files with indirect coverage changes

[ederc]

@fingolfin @lgoettgens Conflicts were fixed, but I still have no idea why the short tests are failing.

[benlorenz]

The code seems to be running out of memory on the github workers.

Locally, running only the binomial-ideals testfile pushes the memory usage to 12GB, and it does not go down even after running GC.gc() manually several times. Running it a second time in the same session quickly exceeds my memory and causes the julia process to die:

julia> println(Base.format_bytes(Sys.maxrss()))
1.905 GiB

julia> Oscar.test_module("Rings/binomial-ideals.jl", new=false,tempproject=false)
Test Summary:   | Pass  Total   Time
Binomial Ideals |   32     32  46.2s

julia> GC.gc(); GC.gc();

julia> println(Base.format_bytes(Sys.maxrss()))
12.084 GiB

julia> Oscar.test_module("Rings/binomial-ideals.jl", new=false,tempproject=false)
Killed

For comparison, on master, the julia process needs about 2.7GB for running the binomial ideals test file.

julia> println(Base.format_bytes(Sys.maxrss()))
1.906 GiB

julia> Oscar.test_module("Rings/binomial-ideals.jl", new=false,tempproject=false)
Test Summary:   | Pass  Total   Time
Binomial Ideals |   32     32  38.9s

julia> GC.gc(); GC.gc();

julia> println(Base.format_bytes(Sys.maxrss()))
2.764 GiB

julia> Oscar.test_module("Rings/binomial-ideals.jl", new=false,tempproject=false)
Test Summary:   | Pass  Total  Time
Binomial Ideals |   32     32  4.8s

julia> GC.gc(); GC.gc();

julia> println(Base.format_bytes(Sys.maxrss()))
2.764 GiB

PS: The binomial ideals file was the last one that was started before this CI job died: https://github.com/oscar-system/Oscar.jl/actions/runs/10899937413/job/30250935622?pr=3987#step:9:3655

[ederc]

@benlorenz Thanks, I think now I know what the problem is. I will need to make a new minor msolve release first.

changes to msolve needed

[lgoettgens]

There is some (new) error in CI:
https://github.com/oscar-system/Oscar.jl/actions/runs/10939116703/job/30368701750?pr=3987#step:9:4503

      From worker 5:	flag pluecker ideal: Test Failed at /Users/aaruni/Desktop/oscar-runners/runner-4/_work/Oscar.jl/Oscar.jl/test/Rings/mpoly.jl:567
      From worker 5:	  Expression: [f1] == gens(I)
      From worker 5:	   Evaluated: QQMPolyRingElem[-x[[1, 4]]*x[[2, 3]] + x[[2, 4]]*x[[1, 3]] - x[[3, 4]]*x[[1, 2]]] == QQMPolyRingElem[-x[[1, 4]]*x[[2, 3]] + x[[2, 4]]*x[[1, 3]] - x[[3, 4]]*x[[1, 2]]]

[ederc]

Yes, I suspect this is coming from the new faster way to construct the polynomials in AlgebraicSolving. content_num is not set correctly. I asked @thofma for a possible solution.