decide_du_val_singularity: Ideal does not describe a single K-point #2531

benlorenz · 2023-07-07T07:55:36Z

Describe the bug
Infrequent bug during the tests after the merge of #2468 by @afkafkafk13, observed on julia 1.8 (ubuntu) and julia nightly (ubuntu)

To Reproduce
Maybe run the tests repeatedly?

Full error message:

du Val Tester: Error During Test at /home/runner/work/Oscar.jl/Oscar.jl/test/AlgebraicGeometry/Schemes/duValSing.jl:13
  Test threw exception
  Expression: ((decide_du_val_singularity(X, J2))[1])[3] == (:A, 3)
  Ideal does not describe a single K-point
  Stacktrace:
   [1] error(s::String)
     @ Base ./error.jl:35
   [2] rational_point_coordinates(I::MPolyIdeal{AbstractAlgebra.Generic.MPoly{nf_elem}})
     @ Oscar ~/work/Oscar.jl/Oscar.jl/experimental/Schemes/SpaceGerms.jl:128
   [3] _check_duval_at_point(IX::MPolyIdeal{AbstractAlgebra.Generic.MPoly{nf_elem}}, Ipt::MPolyIdeal{AbstractAlgebra.Generic.MPoly{nf_elem}})
     @ Oscar ~/work/Oscar.jl/Oscar.jl/experimental/Schemes/duValSing.jl:219
   [4] decide_du_val_singularity(X::Spec{AnticNumberField, MPolyQuoRing{AbstractAlgebra.Generic.MPoly{nf_elem}}}, I::MPolyIdeal{AbstractAlgebra.Generic.MPoly{nf_elem}})
     @ Oscar ~/work/Oscar.jl/Oscar.jl/experimental/Schemes/duValSing.jl:173
   [5] decide_du_val_singularity(X::Spec{QQField, MPolyQuoRing{QQMPolyRingElem}}, I::MPolyIdeal{QQMPolyRingElem})
     @ Oscar ~/work/Oscar.jl/Oscar.jl/experimental/Schemes/duValSing.jl:185
   [6] macro expansion
     @ /opt/hostedtoolcache/julia/nightly/x64/share/julia/stdlib/v1.11/Test/src/Test.jl:669 [inlined]
   [7] macro expansion
     @ ~/work/Oscar.jl/Oscar.jl/test/AlgebraicGeometry/Schemes/duValSing.jl:13 [inlined]
   [8] macro expansion
     @ /opt/hostedtoolcache/julia/nightly/x64/share/julia/stdlib/v1.11/Test/src/Test.jl:1577 [inlined]
   [9] top-level scope
     @ ~/work/Oscar.jl/Oscar.jl/test/AlgebraicGeometry/Schemes/duValSing.jl:2
Test Summary: | Pass  Error  Total   Time
du Val Tester |    7      1      8  31.4s
ERROR: LoadError: LoadError: Some tests did not pass: 7 passed, 0 failed, 1 errored, 0 broken.
in expression starting at /home/runner/work/Oscar.jl/Oscar.jl/test/AlgebraicGeometry/Schemes/duValSing.jl:1

System (please complete the following information):
Current Oscar master on github actions, see logs for details.

The text was updated successfully, but these errors were encountered:

afkafkafk13 · 2023-07-07T13:16:02Z

I tried to reproduce the bug with 50 runs of the offending test on julia 1.8.5 and the current master of Oscar (each time restarting julia from scratch to avoid influence of prior runs -- e.g. on the random seed of Singular or on cached data), but the error did not manifest itself. I also tried 50 runs on the same setup, but without restarting Julia/Oscar. Again the error did not manifest itself.

I am a bit at a loss on how to reproduce the error. Any hint is welcome.

afkafkafk13 · 2023-07-08T16:48:50Z

I tracked it down:

absolute_primary_decomposition may return something not fully decomposed in rare circomstances:

julia> I = ideal(w, z, y^3 + 2*y, x^3 + x, x^7 + 2*x^5 + x^3 + y^8 + 6*y^6 + 12*y^4 + 8*y^2 + z^4);

julia> decomp = absolute_primary_decomposition(I);

julia> decomp
Tuple{MPolyIdeal{QQMPolyRingElem}, MPolyIdeal{QQMPolyRingElem}, MPolyIdeal{AbstractAlgebra.Generic.MPoly{nf_elem}}, Int64}[(ideal(w, z, y, x), ideal(x, y, z, w), ideal(w, z, y, x), 1), (ideal(w, z, y^2 + 2, x), ideal(w, z, y^2 + 2, x), ideal(-45*x + 3*z + w, z, x, y^2 + 2), 1), (ideal(w, z, y, x^2 + 1), ideal(w, z, y, x^2 + 1), ideal(-17*x - 10*y - 45*z + w - _a, z, y, w), 2), (ideal(w, z, y^2 + 2, x^2 + 1), ideal(w, z, y^2 + 2, x^2 + 1), ideal(-36*x + 35*y - 34*z + w - _a, z, 36*_a^2*x + (-35*_a^2 + 80780)*y - _a^2*w - 8646*_a, w), 4)]

More readable version of the components:

(ideal(w, z, y, x), ideal(x, y, z, w), ideal(w, z, y, x), 1), 
(ideal(w, z, y^2 + 2, x), ideal(w, z, y^2 + 2, x), ideal(-45*x + 3*z + w, z, x, y^2 + 2), 1), 
(ideal(w, z, y, x^2 + 1), ideal(w, z, y, x^2 + 1), ideal(-17*x - 10*y - 45*z + w - _a, z, y, w), 2),
 (ideal(w, z, y^2 + 2, x^2 + 1), ideal(w, z, y^2 + 2, x^2 + 1), ideal(-36*x + 35*y - 34*z + w - _a, z, 36*_a^2*x + (-35*_a^2 + 80780)*y - _a^2*w - 8646*_a, w), 4)

The offending component is the second one and the whole decomposition is handed to Oscar by Singular.LibPrimdec.absPrimdecGTZ.

@wdecker : Is it possible that some test on, whether a choice is sufficiently general, misses a case down in the Singular library (or even further down)?

fingolfin · 2023-07-19T09:42:38Z

We just talked about this in our weekly meeting, @wdecker and @hannes14 will look into it.

fingolfin · 2023-07-19T09:48:19Z

@afkafkafk13 your reproducing example is incomplete, there is no ring definition, could you provide one so we can more easily reproduce?

fingolfin · 2023-07-19T09:49:56Z

Here is my guess:

julia> R, (x,y,z,w) = QQ[:x,:y,:z,:w]
(Multivariate polynomial ring in 4 variables over QQ, QQMPolyRingElem[x, y, z, w])

julia> I = ideal(R, [w, z, y^3 + 2*y, x^3 + x, x^7 + 2*x^5 + x^3 + y^8 + 6*y^6 + 12*y^4 + 8*y^2 + z^4]);

julia> decomp = absolute_primary_decomposition(I);

fingolfin · 2023-07-19T09:53:45Z

It seems [x[end] for x in decomp] usually is 1,2,2,4 but in some cases 1,2,2,2 which I assume is wrong.

afkafkafk13 · 2023-07-19T10:03:38Z

Yes, you see it in the multiplicities, and no, the erroneous one can also be 1,1,2,4.

Usually, it decomposes into 9 parts as it should, but sometimes it misses a decomposition.

I reran my test script again and made sure to copy properly today.
Here is run 73 of this example (1 to 72 were fine):

julia> R,(x,y,z,w) = QQ["x","y","z","w"]
(Multivariate polynomial ring in 4 variables over QQ, QQMPolyRingElem[x, y, z, w])

julia> I = ideal(R,[w, z, y^3 + 2*y, x^3 + x, x^7 + 2*x^5 + x^3 + y^8 + 6*y^6 + 12*y^4 + 8*y^2 + z^4])
ideal(w, z, y^3 + 2*y, x^3 + x, x^7 + 2*x^5 + x^3 + y^8 + 6*y^6 + 12*y^4 + 8*y^2 + z^4)

julia> absolute_primary_decomposition(I)
4-element Vector{Tuple{MPolyIdeal{QQMPolyRingElem}, MPolyIdeal{QQMPolyRingElem}, MPolyIdeal{AbstractAlgebra.Generic.MPoly{nf_elem}}, Int64}}:
 (ideal(w, z, y, x), ideal(x, y, z, w), ideal(w, z, y, x), 1)
 (ideal(w, z, y^2 + 2, x), ideal(w, z, y^2 + 2, x), ideal(27*x + 39*z + w, z, x, y^2 + 2), 1)
 (ideal(w, z, y, x^2 + 1), ideal(w, z, y, x^2 + 1), ideal(46*x + 34*y + 31*z + w - _a, z, y, -w), 2)
 (ideal(w, z, y^2 + 2, x^2 + 1), ideal(w, z, y^2 + 2, x^2 + 1), ideal(-16*x - 42*y + 30*z + w - _a, z, -16*_a^2*x + (-42*_a^2 + 274848)*y + _a^2*w + 10840*_a, w), 4)

wdecker · 2023-07-19T11:33:59Z

@hannes14 The bug seems to boil down to this strange behaviour:

julia> R, (x,y,z) = polynomial_ring(QQ, ["x", "y", "z"])
(Multivariate polynomial ring in 3 variables over QQ, QQMPolyRingElem[x, y, z])

julia> I = ideal(R, [y^2+2]);

julia> decompI = absolute_primary_decomposition(I)
1-element Vector{Tuple{MPolyIdeal{QQMPolyRingElem}, MPolyIdeal{QQMPolyRingElem}, MPolyIdeal{AbstractAlgebra.Generic.MPoly{nf_elem}}, Int64}}:
 (ideal(y^2 + 2), ideal(y^2 + 2), ideal(y - _a), 2)

julia> J = ideal(R, [x, y^2+2]);

julia> decompJ = absolute_primary_decomposition(J)
1-element Vector{Tuple{MPolyIdeal{QQMPolyRingElem}, MPolyIdeal{QQMPolyRingElem}, MPolyIdeal{AbstractAlgebra.Generic.MPoly{nf_elem}}, Int64}}:
 (ideal(y^2 + 2, x), ideal(y^2 + 2, x), ideal(y - _a, x), 2)

julia> K = ideal(R, [z, y^2+2]);

julia> decompK = absolute_primary_decomposition(K)
1-element Vector{Tuple{MPolyIdeal{QQMPolyRingElem}, MPolyIdeal{QQMPolyRingElem}, MPolyIdeal{AbstractAlgebra.Generic.MPoly{nf_elem}}, Int64}}:
 (ideal(z, y^2 + 2), ideal(z, y^2 + 2), ideal(17*y + z - _a, -z), 2)

wdecker · 2023-07-19T11:36:35Z

The problem is already on the Singular side

fingolfin · 2023-07-19T11:41:04Z

@afkafkafk13 just to clarify, I got 1,2,2,2 in a test run (out of many) so this can fail in multiple ways it seems

afkafkafk13 · 2023-07-19T11:58:12Z

@fingolfin This is precisely, what I understood.

thofma · 2023-08-01T11:58:46Z

Is someone working on this? This pops up regularly in CI.

afkafkafk13 · 2023-08-01T12:18:30Z

Yes, @wdecker and @hannes14 are working on it. It is a Singular problem.

fieker · 2023-08-09T09:50:48Z

fixed by Singular/Singular#1182

benlorenz · 2024-02-07T13:31:32Z

It looks like this was not really fixed, it recently happened here:
https://github.com/oscar-system/Oscar.jl/actions/runs/7766397045/job/21182180188?pr=3255#step:8:1271
(Probably depending on which other tests use the random number generation before...)

It has since disappeared again on that PR but a similar error can be reproduced by running through a few random seeds:

using Oscar, Test

mod = "$(Oscar.oscardir)/test/AlgebraicGeometry/Schemes/duValSing.jl";
seed = UInt32(0)
while (true)
   seed = seed+1
   println("running with seed $seed")
   Oscar.randseed!(seed)
   Oscar.set_seed!(seed)
   Oscar.test_module(mod; new=false)
end

On my setup this fails at seed 25:

running with seed 25
du Val Tester: Error During Test at /home/datastore/lorenz/software/julia/Oscar-duval.jl/test/AlgebraicGeometry/Schemes/duValSing.jl:12
  Test threw exception
  Expression: is_du_val_singularity(X, J3) == false
  Ideal does not describe a single K-point
  Stacktrace:
   [1] error(s::String)
     @ Base ./error.jl:35
   [2] rational_point_coordinates(I::MPolyIdeal{AbstractAlgebra.Generic.MPoly{AbsSimpleNumFieldElem}})
     @ Oscar /home/datastore/lorenz/software/julia/Oscar-duval.jl/experimental/Schemes/SpaceGerms.jl:343
   [3] _check_duval_at_point(IX::MPolyIdeal{AbstractAlgebra.Generic.MPoly{AbsSimpleNumFieldElem}}, Ipt::MPolyIdeal{AbstractAlgebra.Generic.MPoly{AbsSimpleNumFieldElem}})
     @ Oscar /home/datastore/lorenz/software/julia/Oscar-duval.jl/experimental/Schemes/duValSing.jl:242
   [4] is_du_val_singularity(X::Spec{AbsSimpleNumField, MPolyQuoRing{AbstractAlgebra.Generic.MPoly{AbsSimpleNumFieldElem}}}, I::MPolyIdeal{AbstractAlgebra.Generic.MPoly{AbsSimpleNumFieldElem}})
     @ Oscar /home/datastore/lorenz/software/julia/Oscar-duval.jl/experimental/Schemes/duValSing.jl:116
   [5] is_du_val_singularity(X::Spec{QQField, MPolyQuoRing{QQMPolyRingElem}}, I::MPolyIdeal{QQMPolyRingElem})
     @ Oscar /home/datastore/lorenz/software/julia/Oscar-duval.jl/experimental/Schemes/duValSing.jl:127
   [6] macro expansion
     @ /net/site-local.linux64/julia/julia-1.10.0/share/julia/stdlib/v1.10/Test/src/Test.jl:669 [inlined]
   [7] macro expansion
     @ /home/datastore/lorenz/software/julia/Oscar-duval.jl/test/AlgebraicGeometry/Schemes/duValSing.jl:12 [inlined]
   [8] macro expansion
     @ /net/site-local.linux64/julia/julia-1.10.0/share/julia/stdlib/v1.10/Test/src/Test.jl:1577 [inlined]
   [9] top-level scope
     @ /home/datastore/lorenz/software/julia/Oscar-duval.jl/test/AlgebraicGeometry/Schemes/duValSing.jl:2
du Val Tester: Error During Test at /home/datastore/lorenz/software/julia/Oscar-duval.jl/test/AlgebraicGeometry/Schemes/duValSing.jl:15
  Test threw exception
  Expression: ((decide_du_val_singularity(X, J3))[1])[1] == false
  Ideal does not describe a single K-point
  Stacktrace:
   [1] error(s::String)
     @ Base ./error.jl:35
   [2] rational_point_coordinates(I::MPolyIdeal{AbstractAlgebra.Generic.MPoly{AbsSimpleNumFieldElem}})
     @ Oscar /home/datastore/lorenz/software/julia/Oscar-duval.jl/experimental/Schemes/SpaceGerms.jl:343
   [3] _check_duval_at_point(IX::MPolyIdeal{AbstractAlgebra.Generic.MPoly{AbsSimpleNumFieldElem}}, Ipt::MPolyIdeal{AbstractAlgebra.Generic.MPoly{AbsSimpleNumFieldElem}})
     @ Oscar /home/datastore/lorenz/software/julia/Oscar-duval.jl/experimental/Schemes/duValSing.jl:242
   [4] decide_du_val_singularity(X::Spec{AbsSimpleNumField, MPolyQuoRing{AbstractAlgebra.Generic.MPoly{AbsSimpleNumFieldElem}}}, I::MPolyIdeal{AbstractAlgebra.Generic.MPoly{AbsSimpleNumFieldElem}})
     @ Oscar /home/datastore/lorenz/software/julia/Oscar-duval.jl/experimental/Schemes/duValSing.jl:196
   [5] decide_du_val_singularity(X::Spec{QQField, MPolyQuoRing{QQMPolyRingElem}}, I::MPolyIdeal{QQMPolyRingElem})
     @ Oscar /home/datastore/lorenz/software/julia/Oscar-duval.jl/experimental/Schemes/duValSing.jl:208
   [6] macro expansion
     @ /net/site-local.linux64/julia/julia-1.10.0/share/julia/stdlib/v1.10/Test/src/Test.jl:669 [inlined]
   [7] macro expansion
     @ /home/datastore/lorenz/software/julia/Oscar-duval.jl/test/AlgebraicGeometry/Schemes/duValSing.jl:15 [inlined]
   [8] macro expansion
     @ /net/site-local.linux64/julia/julia-1.10.0/share/julia/stdlib/v1.10/Test/src/Test.jl:1577 [inlined]
   [9] top-level scope
     @ /home/datastore/lorenz/software/julia/Oscar-duval.jl/test/AlgebraicGeometry/Schemes/duValSing.jl:2
Test Summary: | Pass  Error  Total  Time
du Val Tester |    6      2      8  0.7s
ERROR: LoadError: Some tests did not pass: 6 passed, 0 failed, 2 errored, 0 broken.
in expression starting at /home/datastore/lorenz/software/julia/Oscar-duval.jl/test/AlgebraicGeometry/Schemes/duValSing.jl:1

This is on master, commit a91222c.

Note that the failure of the reproducer is at Expression: is_du_val_singularity(X, J3) == false but the CI test died at Expression: has_du_val_singularities(X) == false.

Seed 94 gives the same error as the CI:

running with seed 94
du Val Tester: Error During Test at /home/datastore/lorenz/software/julia/Oscar-duval.jl/test/AlgebraicGeometry/Schemes/duValSing.jl:16
  Test threw exception
  Expression: has_du_val_singularities(X) == false
  Ideal does not describe a single K-point

cc: @antonydellavecchia

afkafkafk13 · 2024-02-07T19:13:19Z

I confirm: It is the same old bug coming up from Singular. It is again an incomplete absolute primary decomposition.

In the context of the code to reproduce provided by @benlorenz today, I see the following absolute primary decompositon in my debug output (newlines included for readability):

decomp = Tuple{MPolyIdeal{QQMPolyRingElem}, MPolyIdeal{QQMPolyRingElem}, MPolyIdeal{AbstractAlgebra.Generic.MPoly{nf_elem}}, Int64}
[(ideal(w, z, y^2 + 2, x), ideal(w, z, y^2 + 2, x), ideal(-6*x - 7*z + w, z, x, y^2 + 2), 1)]

Note that this describes two conjugate points over the algebraic closure (i.e. this is not absolutely primary) , but the number at the end indicates just 1 (i.e. absolutely primary). This mistake arises from the fact the the third ideal in the list, which should provide an ideal of any of the two points over a algebraic suitable field extension, is in reality just the same ideal as the original one.

@fieker originally tracked this down in the Singular library. The codeblock he changed in the Singular library, works fine now! Unfortunately, there seem to be additional copies of this precise codeblock in the file, which are still unchanged. (Look for randomLast in this file)

afkafkafk13 · 2024-02-07T19:56:04Z

Could @fieker and @hannes14 please jointly have a look at this? This is also linked to Singular's#1186 and it seems that @hannes14 might have reverted the bugfix in the meantime -- on the other hand, the original problem (independently reported by fchyzak to Singular) might still have persisted after the original fix, just because only one of three or four occurrences had been fixed by @fieker .

hannes14 · 2024-02-08T09:27:50Z

sShould be fixed with Singular/Singular@70c8c34
(atleast the test above has now error till seed 1700)

afkafkafk13 · 2024-02-08T11:56:17Z

Thank you for the quick reaction.

lgoettgens · 2024-08-29T15:47:38Z

should be fixed upstream. If someone sees it again, please re-open.

benlorenz added the bug Something isn't working label Jul 7, 2023

benlorenz mentioned this issue Jul 19, 2023

nightly + 1.10-alpha1: GC marking segfault during GAP tests #2543

Closed

fingolfin added the singular coding sprint label Jul 19, 2023

fingolfin assigned hannes14 and wdecker Jul 19, 2023

fingolfin mentioned this issue Jul 25, 2023

document Julia.GAP.Obj(x,true), add tests oscar-system/GAP.jl#925

Merged

joschmitt mentioned this issue Jul 26, 2023

Resolve two small issues regarding direct products of groups #2588

Merged

thofma mentioned this issue Aug 2, 2023

minor improvements #2609

Merged

fingolfin closed this as completed Aug 24, 2023

benlorenz reopened this Feb 7, 2024

lgoettgens closed this as completed Aug 29, 2024

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

decide_du_val_singularity: Ideal does not describe a single K-point #2531

decide_du_val_singularity: Ideal does not describe a single K-point #2531

benlorenz commented Jul 7, 2023 •

edited

Loading

afkafkafk13 commented Jul 7, 2023

afkafkafk13 commented Jul 8, 2023 •

edited

Loading

fingolfin commented Jul 19, 2023

fingolfin commented Jul 19, 2023

fingolfin commented Jul 19, 2023

fingolfin commented Jul 19, 2023

afkafkafk13 commented Jul 19, 2023 •

edited

Loading

wdecker commented Jul 19, 2023 •

edited by fingolfin

Loading

wdecker commented Jul 19, 2023

fingolfin commented Jul 19, 2023

afkafkafk13 commented Jul 19, 2023

thofma commented Aug 1, 2023

afkafkafk13 commented Aug 1, 2023

fieker commented Aug 9, 2023 •

edited by lgoettgens

Loading

benlorenz commented Feb 7, 2024 •

edited

Loading

afkafkafk13 commented Feb 7, 2024 •

edited

Loading

afkafkafk13 commented Feb 7, 2024

hannes14 commented Feb 8, 2024 •

edited by lgoettgens

Loading

afkafkafk13 commented Feb 8, 2024

lgoettgens commented Aug 29, 2024

decide_du_val_singularity: Ideal does not describe a single K-point #2531

decide_du_val_singularity: Ideal does not describe a single K-point #2531

Comments

benlorenz commented Jul 7, 2023 • edited Loading

afkafkafk13 commented Jul 7, 2023

afkafkafk13 commented Jul 8, 2023 • edited Loading

fingolfin commented Jul 19, 2023

fingolfin commented Jul 19, 2023

fingolfin commented Jul 19, 2023

fingolfin commented Jul 19, 2023

afkafkafk13 commented Jul 19, 2023 • edited Loading

wdecker commented Jul 19, 2023 • edited by fingolfin Loading

wdecker commented Jul 19, 2023

fingolfin commented Jul 19, 2023

afkafkafk13 commented Jul 19, 2023

thofma commented Aug 1, 2023

afkafkafk13 commented Aug 1, 2023

fieker commented Aug 9, 2023 • edited by lgoettgens Loading

benlorenz commented Feb 7, 2024 • edited Loading

afkafkafk13 commented Feb 7, 2024 • edited Loading

afkafkafk13 commented Feb 7, 2024

hannes14 commented Feb 8, 2024 • edited by lgoettgens Loading

afkafkafk13 commented Feb 8, 2024

lgoettgens commented Aug 29, 2024

benlorenz commented Jul 7, 2023 •

edited

Loading

afkafkafk13 commented Jul 8, 2023 •

edited

Loading

afkafkafk13 commented Jul 19, 2023 •

edited

Loading

wdecker commented Jul 19, 2023 •

edited by fingolfin

Loading

fieker commented Aug 9, 2023 •

edited by lgoettgens

Loading

benlorenz commented Feb 7, 2024 •

edited

Loading

afkafkafk13 commented Feb 7, 2024 •

edited

Loading

hannes14 commented Feb 8, 2024 •

edited by lgoettgens

Loading