bugGNU Scientific Library - Bugs: bug #35367, ode-initval2 testsuite failure

 
 

You are not allowed to post comments on this tracker with your current authentication level.

bug #35367: ode-initval2 testsuite failure

Submitter:  Andreas Schwab <schwab>
Submitted:  Wed 25 Jan 2012 02:09:37 PM UTC
   
 
Category:  Accuracy problem Severity:  3 - Normal
Operating System:  Status:  Fixed
Assigned to:  tkeskita Open/Closed:  Closed
Release:  1.15

Jump to the original submission

Sun 03 Feb 2013 06:36:33 PM UTC, comment #6: 

Bug against ode-initval2 corrected, so I'm closing this ticket.  Please open new bug reports against those other problems you have found.

Tuomo Keskitalo <tkeskita>
Group Member
Sun 27 Jan 2013 03:47:29 PM UTC, comment #5: 

The ode-initval2 test no longer fails, but others remain.

specfun:
FAIL: gsl_sf_hyperg_2F1_e(3.5, -0.5, 5.0, 0.9, &r) [2104]
  expected: 5.9239812843706530e-01
  obtained: 4.5788485734926559e-01 +/- 1.1491551919116466e-14 (rel=2.5097e-14)
  fracdiff: 1.2807336014025938e-01
 tolerance: 5.6843418860808015e-14
  value/expected not consistent within reported error
  value not within tolerance of expected value
  4.578848573492655949e-01  1.149155191911646631e-14
FAIL: gsl_sf_hyperg_2F1_e(-1.0, -10.0, 1.0, 0.5, &r) [2105]
  expected: 6.0000000000000000e+00
  obtained: 0.0000000000000000e+00 +/- 0.0000000000000000e+00 (rel=nan)
  fracdiff: 1.0000000000000000e+00
 tolerance: 4.4408920985006262e-16
  value/expected not consistent within reported error
  value not within tolerance of expected value
  unexpected return code: 24
  0.000000000000000000e+00  0.000000000000000000e+00
FAIL: Hypergeometric Functions [2134]
FAIL: gsl_sf_laguerre_n_e(1e6+1, 2.5, 2.5, &r) [2165]
  expected: -4.8017961545391276e+05
  obtained: -4.8017961546181823e+05 +/- 3.6891004271829821e-02 (rel=7.68275e-08)
  fracdiff: 8.2317878690294596e-12
 tolerance: 3.6379788070917130e-12
  value not within tolerance of expected value
  -4.801796154618182336e+05  3.689100427182982062e-02
FAIL: Laguerre Polynomials [2204]

poly:
FAIL: z11.real, 15th-order polynomial (1.00000003899845002 observed vs 2 expected) [179]
FAIL: z14.real, 15th-order polynomial (2.00000002744542504 observed vs 1 expected) [185]

cdf:
gsl: ../../gsl/cdf/gammainv.c:111: ERROR: inverse failed to converge

When compiling with -ffp-contract=off, the following failures remain:

specfun:
FAIL: gsl_sf_hyperg_2F1_e(3.5, -0.5, 5.0, 0.9, &r) [2104]
  expected: 5.9239812843706530e-01
  obtained: 4.5788485734926559e-01 +/- 1.1491551919116466e-14 (rel=2.5097e-14)
  fracdiff: 1.2807336014025938e-01
 tolerance: 5.6843418860808015e-14
  value/expected not consistent within reported error
  value not within tolerance of expected value
  4.578848573492655949e-01  1.149155191911646631e-14
FAIL: gsl_sf_hyperg_2F1_e(-1.0, -10.0, 1.0, 0.5, &r) [2105]
  expected: 6.0000000000000000e+00
  obtained: 0.0000000000000000e+00 +/- 0.0000000000000000e+00 (rel=nan)
  fracdiff: 1.0000000000000000e+00
 tolerance: 4.4408920985006262e-16
  value/expected not consistent within reported error
  value not within tolerance of expected value
  unexpected return code: 24
  0.000000000000000000e+00  0.000000000000000000e+00
FAIL: Hypergeometric Functions [2134]

poly:
gsl: ../../gsl/poly/zsolve.c:78: ERROR: root solving qr method failed to converge

cdf:
gsl: ../../gsl/cdf/gammainv.c:111: ERROR: inverse failed to converge

Andreas Schwab <schwab>
Sun 27 Jan 2013 02:43:02 PM UTC, comment #4: 

I've modified msbdf, available in GSL BZR trunk, rev 4788. Could you please test if you still can reproduce problem with it? --Tuomo

Tuomo Keskitalo <tkeskita>
Group Member
Sun 02 Dec 2012 07:07:00 AM UTC, comment #3: 
Tuomo Keskitalo <tkeskita>
Group Member
Sun 02 Dec 2012 06:52:37 AM UTC, comment #2: 
Tuomo Keskitalo <tkeskita>
Group Member
Sun 02 Dec 2012 06:50:56 AM UTC, comment #1: 
Tuomo Keskitalo <tkeskita>
Group Member
Wed 25 Jan 2012 02:09:37 PM UTC, original submission:  

gsl: ../../gsl/ode-initval2/msbdf.c:1336: ERROR: msbdf_apply too large order change
-- order change -2
Default GSL error handler invoked.

That only happens when the compiler is allowed to generate FMA instructions, which suggests instable algorithms.

Andreas Schwab <schwab>

 

(Note: upload size limit is set to 16384 kB, after insertion of the required escape characters.)

No files currently attached

 

Depends on the following items: None found

Items that depend on this one: None found

 

Carbon-Copy List
  • -email is unavailable- added by tkeskita (Posted a comment)
  • -email is unavailable- added by schwab (Submitted the item)
  •  

    Follow 4 latest changes.

    Date Changed by Updated Field Previous Value => Replaced by
    2013-02-03 tkeskita StatusReady For Test Fixed
        Open/ClosedOpen Closed
    2013-01-27 tkeskita StatusNone Ready For Test
        Assigned toNone tkeskita

    Back to the top

    Powered by Savane 3.13-758e.
    Corresponding source code