Banff presentation rebased

[HechtiDerLachs]

Rebase of #3891 with some fixes.

[HechtiDerLachs]

It looks like some changes introduced here are quite expensive. vinberg1.jlcon did terminate, but probably took a long time. I will investigate and do some comparisons.

[benlorenz]

It looks like some changes introduced here are quite expensive. vinberg1.jlcon did terminate, but probably took a long time. I will investigate and do some comparisons.

The timings don't look that bad, vinberg_1 took about 1 minute:

Sat, 29 Jun 2024 19:39:02 GMT    vinberg_1.jlcon
....
Sat, 29 Jun 2024 19:40:26 GMT    vinberg_2.jlcon: skip

There are quite a few GC pauses in between but that is not unusual.

The booktests ran from 19:15 until it got stuck in zariski.jlcon at 19:42, which is the same time and result as we have on master. It will hopefully be fixed via singular.

[HechtiDerLachs]

Thanks for having a look!

I was afraid that the booktests only got cancelled, because the vinberg_1.jlcon had already taken up too much of the time. But indeed I have no trouble running this test locally.

I also tried the testfile for elliptic_surfaces.jl and got the following timings.
Before:

606.089320 seconds (3.03 G allocations: 116.741 GiB, 8.48% gc time)

620.284879 seconds (3.03 G allocations: 115.920 GiB, 8.75% gc time)

With this branch:

601.933683 seconds (2.93 G allocations: 114.751 GiB, 8.59% gc time)

662.196737 seconds (2.84 G allocations: 108.817 GiB, 8.91% gc time)

So memory consumption decreased with no significant impact on computing time. I think this is all good then.

[simonbrandhorst]

Please either remove the old code or add an algorithm option if you think that it still performs better in some cases.

[simonbrandhorst]

May I suggest to name this function
has_is_one(I::AbsIdealSheaf, U::AbsAffineScheme)
and in its body to call
has_is_one(I::Ideal) ?
We have the has_is_... syntax already in some other places.
It means: has cached the value already ( or it is trivial to compute).

[HechtiDerLachs]

I saw has_is_one being introduced. However, it is something different: Here we ask whether an ideal sheaf is known to be isomorphic to the unit ideal sheaf. has_is_one is asking whether it's easy to decide this question. The implementations are different and it would be more work to adapt this here. Since it's an internal function, I suggest to leave this for the moment.

[codecov]

Codecov Report

Attention: Patch coverage is 91.30435% with 8 lines in your changes missing coverage. Please review.

Project coverage is 83.85%. Comparing base (5acca40) to head (198d1ba).
Report is 4 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #3898      +/-   ##
==========================================
- Coverage   83.87%   83.85%   -0.02%     
==========================================
  Files         589      588       -1     
  Lines       80864    80891      +27     
==========================================
+ Hits        67823    67835      +12     
- Misses      13041    13056      +15     

Files	Coverage Δ
src/Rings/mpolyquo-localizations.jl	72.91% <100.00%> (ø)
...ental/Schemes/src/MorphismFromRationalFunctions.jl	51.00% <0.00%> (-0.19%)	⬇️
experimental/Schemes/src/IdealSheaves.jl	78.67% <92.30%> (+0.12%)	⬆️
experimental/Schemes/src/WeilDivisor.jl	71.24% <92.06%> (+2.19%)	⬆️

... and 9 files with indirect coverage changes