Eagon northcott complex

[HechtiDerLachs]

This is the branch with the full functionality for what's used in this preprint. I would like to bring these things to OSCAR, but piece by piece. The first round was in #4248 , so once that one is merged, I will rebase this PR again. For the moment, I would just like to work on the tests and polish this here a bit.

Ready for review. This introduces

• strands of complexes of graded modules for ZZ^m-gradings
• Eagon-Northcott complexes

[HechtiDerLachs]

Tests are failing, but I can not reproduce a single one of these failures. Is any of these problems known? Ping @benlorenz @aaruni96

[thofma]

keys for generators (comprehensions) is not available on 1.6

[codecov]

Codecov Report

Attention: Patch coverage is 88.98305% with 52 lines in your changes missing coverage. Please review.

Project coverage is 84.33%. Comparing base (4542bd3) to head (ed2cf5a).
Report is 7 commits behind head on master.

Files with missing lines	Patch %	Lines
experimental/Schemes/src/DerivedPushforward.jl	30.64%	43 Missing ⚠️
src/Modules/Iterators.jl	96.82%	4 Missing ⚠️
...HyperComplexes/src/Objects/new_koszul_complexes.jl	93.18%	3 Missing ⚠️
...tal/DoubleAndHyperComplexes/src/Objects/Methods.jl	88.88%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #4327      +/-   ##
==========================================
- Coverage   84.52%   84.33%   -0.19%     
==========================================
  Files         646      656      +10     
  Lines       85823    86875    +1052     
==========================================
+ Hits        72539    73268     +729     
- Misses      13284    13607     +323     

Files with missing lines	Coverage Δ
...perComplexes/src/Morphisms/simplified_complexes.jl	94.71% <100.00%> (ø)
...perComplexes/src/Morphisms/strand_functionality.jl	100.00% <100.00%> (ø)
...l/DoubleAndHyperComplexes/src/Morphisms/strands.jl	100.00% <100.00%> (ø)
...mplexes/src/Objects/cartan_eilenberg_resolution.jl	75.67% <100.00%> (ø)
...erComplexes/src/Objects/eagon_northcott_complex.jl	100.00% <100.00%> (ø)
...DoubleAndHyperComplexes/src/Objects/induced_ENC.jl	100.00% <100.00%> (ø)
...plexes/src/Objects/tensor_product_functionality.jl	100.00% <100.00%> (ø)
...leAndHyperComplexes/src/Objects/tensor_products.jl	90.56% <ø> (-0.67%)	⬇️
src/Modules/UngradedModules/HomologicalAlgebra.jl	92.81% <100.00%> (ø)
src/Modules/UngradedModules/Methods.jl	88.94% <100.00%> (+2.89%)	⬆️
... and 5 more

... and 20 files with indirect coverage changes

[HechtiDerLachs]

keys for generators (comprehensions) is not available on 1.6

Sorry, but I still do not understand what's going on here. I can not reproduce the error. You are talking about 1.6 of what? Oscar? Julia? Hecke? Nemo?

[benlorenz]

It would be good to improve the coverage a bit

Attention: Patch coverage is 53.81356% with 218 lines in your changes missing coverage. Please review.

especially src/Modules/Iterators.jl where just 8% of the new code is tested.

[HechtiDerLachs]

It would be good to improve the coverage a bit

Yes. I was thinking of putting the easier examples from my paper here as test files. That way people also have the possibility to reproduce the computations. Is this a legitimate approach? @JHanselman ?

[lgoettgens]

It would be good to improve the coverage a bit

Yes. I was thinking of putting the easier examples from my paper here as test files. That way people also have the possibility to reproduce the computations. Is this a legitimate approach? @JHanselman ?

That would of course help. However, please then add a comment somewhere that these are more like regression tests aka you are not certain that the current result is correct, but you want failures if the result changes. And (if possible) please add some basic tests that can be either verified by hand or using other software (although this may be hard to do/impossible)

[benlorenz]

One file with very low patch coverage is src/Modules/Iterators.jl, it should be easy to add some basic tests for iterating or counting some AllModuleMonomials and AllModuleExponents objects. Having separate tests for such code makes maintenance a lot easier than just testing this indirectly via some higher-level examples.

[HereAround]

@wdecker seems interested, and will touch by next week (likely this week) with @HechtiDerLachs to discuss more.

[fingolfin]

@HechtiDerLachs I am pretty sure @thofma referred to Julia 1.6 (which we currently still support). After all he was referring to language feature: "keys for generators (comprehensions) is not available on 1.6". In addition, none of Oscar, Hecke, Nemo have reached version 1.6...

[HechtiDerLachs]

@benlorenz gave a more detailed explanation what's going on below and I could already do similar fixes on other PRs myself after that. So this question has been clarified. But thank you for your explanation.

[thofma]

There are plenty of tests added which are disabled, which I have mixed feelings about. If this has to do with having the code for some paper be available somewhere, I wonder whether uncommented tests in a specific Oscar commit is the right place for this. (This also "violates" the Mardi (FAIR?) principle, since it is not reproducible.) I don't think this should block this PR here, but it might be better to discuss this specific issue in-person in case this comes up again.

[HechtiDerLachs]

Thanks for the feedback. I also do not find this solution really viable at the moment and will eventually move on to something else. I will consult with the appropriate people in person to learn about better approaches.