Project

General

Profile

Slug #1390

CoCoA-5 test: radical too slow

Added by John Abbott over 4 years ago. Updated about 4 years ago.

Status:
Closed
Priority:
Normal
Assignee:
John Abbott
Category:
enhancing/improving
Target version:
CoCoA-5.3.0
Start date:
09 Jan 2020
Due date:
% Done:

100%

Estimated time:
2.11 h
Spent time:
2.10 h

Description

The CoCoA-5 test radical.cocoa5 is too slow: it takes longer than all the other CoCoA-5 tests combined.

The culprit seems to be the last example (over ZZ/(32003) which takes almost 90% of the time.

Can we make this test faster?

Related issues

Related to CoCoALib - Feature #1780: radical for ideals in SparsePolyRing: code in C++ Closed2024-02-06

History

#1 Updated by John Abbott over 4 years ago

On my computer with debugging the test radical.cocoa5 takes more than 60s. It is just about tolerably fast when I compile without debugging.

Is the last example in the test file really so important as to justify taking as much time as all the other tests combined?

NOTE If we resolve this, I should probably remove the ulimit call in the RunTests.sh script.

#2 Updated by John Abbott over 4 years ago

  • Status changed from New to In Progress
  • % Done changed from 0 to 20

The radical.cocoa5 test took 160s (2 min 40s) on the netbook.

I propose simply commenting out that last slow case in the file.
When eventually a new impl of radical becomes available we can perhaps reactivate it.

#3 Updated by John Abbott over 4 years ago

  • Assignee set to John Abbott
  • % Done changed from 20 to 50
  • Estimated time set to 1.33 h

I have disabled the slow test, and checked in.

Now all CoCoA-5 tests complete in 12s instead of 23s

#4 Updated by Anna Maria Bigatti over 4 years ago

  • Status changed from In Progress to Resolved
  • % Done changed from 50 to 80

I separated the file into two files
radical0dim.cocoa5 and radical.cocoa5.
Even though the function is the same the underlying algorithms are drastically different.

This split will make it easier to test them separately.

Moreover, there was another slow test which was already commented out.

M:=6;  use R::=QQ[x[1..M,1..M]], DegLex;

Now it's fast, so I activated it.

I made it easier to activate the slow test when one wants to actually check it.

#5 Updated by John Abbott over 4 years ago

  • Status changed from Resolved to Feedback
  • % Done changed from 80 to 90
  • Estimated time changed from 1.33 h to 2.11 h

I have removed ulimit from RunTests.sh.

All tests pass, even with debugging active.
Now all CoCoA-5 tests together take 66s with debugging active.
Without debugging active the tests take about 13s.

#6 Updated by John Abbott about 4 years ago

  • Status changed from Feedback to Closed
  • % Done changed from 90 to 100

#7 Updated by Anna Maria Bigatti 3 months ago

  • Related to Feature #1780: radical for ideals in SparsePolyRing: code in C++ added

Also available in: Atom PDF