Opened 10 years ago

Last modified 10 years ago

#11897 closed defect

Fix eigenmatrix doctest to work across all platforms — at Version 6

Reported by: rbeezer Owned by: jason, was
Priority: minor Milestone: sage-4.7.2
Component: linear algebra Keywords:
Cc: kcrisman, jdemeyer, leif Merged in:
Authors: Rob Beezer Reviewers: Karl-Dieter Crisman, Leif Leonhardy
Report Upstream: N/A Work issues:
Branch: Commit:
Dependencies: Stopgaps:

Status badges

Description (last modified by leif)

Doctest for RDF matrix fails on some platforms by returning the negatives of the more commonly returned eigenvectors.

See sage-release discussion: http://groups.google.com/group/sage-release/msg/e885e53316c6880f


Apply only trac_11897-doctest-RDF-eigenmatrix.rebased.patch to the Sage library.

Change History (8)

Changed 10 years ago by rbeezer

comment:1 follow-up: Changed 10 years ago by rbeezer

  • Authors set to Rob Beezer
  • Status changed from new to needs_review
  1. Patch marks failing doctests as "not tested"

Rationale: this was just meant to show how to get results for inexact matrices with real or complex entries. It is in the middle of a docstring for an exact routine. So this preserves the "doc" part and abandons the "test" part.

  1. Patch repeats, and fixes, doctest in the TEST section of the eigenmatrix routines.

Rational: doctest has not been abandoned. Due to the need to adjust the sign of the eigenvectors, this is relegated to a test section.

This was built on an alpha3 prerelease, I trust it will be OK on a real alpha3 (which I am about to build right now).

comment:2 in reply to: ↑ 1 ; follow-up: Changed 10 years ago by leif

  • Status changed from needs_review to needs_work

Replying to rbeezer:

This was built on an alpha3 prerelease, I trust it will be OK on a real alpha3 (which I am about to build right now).

Two hunks do not apply because of (the late) ticket:11595:trac_11595-fix_noisy_zero_doctest_errors.reviewer.patch.

Otherwise looks fine, but perhaps Karl-Dieter should rerun the tests on his famous favorite machine.

comment:3 follow-up: Changed 10 years ago by leif

Do you also open a follup-up ticket to #7852?

ticket:7852:trac_7852-adjust_noisy_zero_term_threshold_for_polys.reviewer.patch wasn't enough for his machine. (I later increased epsilon from 1.0e-15 to 2.5e-15 in one example to make bsd.math happy, but he gets 2.6645352591e-15.)

comment:4 in reply to: ↑ 2 Changed 10 years ago by kcrisman

Two hunks do not apply because of (the late) ticket:11595:trac_11595-fix_noisy_zero_doctest_errors.reviewer.patch.

Correct. Let me know when you have a new one. :(

Otherwise looks fine, but perhaps Karl-Dieter should rerun the tests on his famous favorite machine.

Now, now! No disparaging remarks. At least I'm not using Windows or FreeBSD! Those pose more drastic problems :)

comment:5 in reply to: ↑ 3 Changed 10 years ago by rbeezer

Replying to leif:

Do you also open a follup-up ticket to #7852?

No, did not want to venture into polynomials - just took responsibility for matrices.

Forgot to actually start a build last night, so it will be maybe 12 hours at the soonest before I can build a proper patch (long story). But will do.

Rob

Changed 10 years ago by leif

Rob's patch rebased on Sage 4.7.2.alpha3.

comment:6 Changed 10 years ago by leif

  • Description modified (diff)
  • Reviewers set to Karl-Dieter Crisman, Leif Leonhardy
  • Status changed from needs_work to needs_review

Sorry, I could easily have done that yesterday, but I was too tired to even look at the rejects.

The rebased patch also passes tests and the documentation builds fine, so positive review from my side.

Karl-Dieter (or Dasher / student?), please finalize!

Note: See TracTickets for help on using tickets.