Le_beullenst test has an inconsistent behavior

[Dioprz]

When executing one of the tests runned by docker-pytest, the command

sage tests/test_le_beullens.sage

Has a very inconsistent behaviour:

root@d541a72bc243:/home/cryptographic_estimators# sage tests/test_le_beullens.sage
Traceback (most recent call last):
  File "/home/cryptographic_estimators/tests/test_le_beullens.sage.py", line 55, in <module>
    test_beullens2()
  File "/home/cryptographic_estimators/tests/test_le_beullens.sage.py", line 50, in test_beullens2
    assert t2 - RANGES < t1 < t2 + RANGES
AssertionError
root@d541a72bc243:/home/cryptographic_estimators# sage tests/test_le_beullens.sage
Traceback (most recent call last):
  File "/home/cryptographic_estimators/tests/test_le_beullens.sage.py", line 55, in <module>
    test_beullens2()
  File "/home/cryptographic_estimators/tests/test_le_beullens.sage.py", line 50, in test_beullens2
    assert t2 - RANGES < t1 < t2 + RANGES
AssertionError
root@d541a72bc243:/home/cryptographic_estimators# sage tests/test_le_beullens.sage # No error from here
root@d541a72bc243:/home/cryptographic_estimators# sage tests/test_le_beullens.sage
root@d541a72bc243:/home/cryptographic_estimators# sage tests/test_le_beullens.sage
root@d541a72bc243:/home/cryptographic_estimators# sage tests/test_le_beullens.sage

This issue must be addressed in the next phase of the SageMath migration.

[FloydZ]

@Dioprz is this still an issue? And if so on which branch are you working. The problem is that the script by Beullens is faulty, hence the Estimator is computing something slightly different. Thats why we added this ranges parameter. To compensate up to small factor for the faulty computation.

[Dioprz]

Got it. Just let me ask:

1. Do you know why errors stop after some executions? It feels like if something is being cacheed, but I'm not sure.
2. Anything we can do to fix/improve the reliability of the test?
3. Should/could the range be increased, so it works even from the first execution?

EDIT: About your questions, I will check if this is still an issue while making the test directory refactor. On the branch, it could be the nmonomials series refactor, but I'm not sure. Give me some days so I can check things and report back.