Some test tolerances not met on NCI system

[Rhodri Davies]

Dear all,

Some tests in the short and medium test suites are failing here on the NCI system from what appear to be tolerance issues:

The test that continually fails is Darcy_p1dgp2_pressBCinlet. My output from this test is below. Could the test owner (btollit) please either loosen the test tolerances somewhat so that they pass or let the devs know if this is pointing at a more serious issue?

I also have a test that fails intermittently: divergence_free_velocity_press_cg_test_cty_cv.xml. This sometimes fails it's first test (Running MaxControlVolumeDivergence_p0p1_3d lower than tolerance 1.0e-10:) - can this tolerance also be loosened somewhat (again test owner btollit)?

Best wishes,

Rhod

darcy_p1dgp2_pressBCinlet: Assigning variables:
darcy_p1dgp2_pressBCinlet: Assigning pressure_1d = (0.0, 1500000000.0)
darcy_p1dgp2_pressBCinlet: Assigning inter_vel_1d = 10.0
darcy_p1dgp2_pressBCinlet: Assigning pressure_2d = (0.0, 1500000000.0)
darcy_p1dgp2_pressBCinlet: Assigning inter_vel_2d = 10.0000000056
darcy_p1dgp2_pressBCinlet: Assigning pressure_3d = (0.0, 1500000000.0)
darcy_p1dgp2_pressBCinlet: Assigning inter_vel_3d = 10.0000000121
darcy_p1dgp2_pressBCinlet: Assigning poisson_pressure_2d = (0.0, 1500000000.0)
darcy_p1dgp2_pressBCinlet: Assigning poisson_inter_vel_2d = 10.0000000056
darcy_p1dgp2_pressBCinlet: Assigning poisson_pressure_3d = (0.0, 1500000000.0)
darcy_p1dgp2_pressBCinlet: Assigning poisson_inter_vel_3d = 10.0000000121
darcy_p1dgp2_pressBCinlet: Running failure tests:
darcy_p1dgp2_pressBCinlet: Running change_P for 1d should equal domain_length*visc*darcy_vel_BC/perm, check relative difference to be under tolerance 1.0e-09:
Relative error of pressure difference: 1.58945719401e-16
darcy_p1dgp2_pressBCinlet: success.
darcy_p1dgp2_pressBCinlet: Running interstitial velocity 1d should equal darcy_vel/porosity, check relative difference to be under tolerance 1.0e-09:
Relative error of interstitial velocity: 1.42108547152e-15
darcy_p1dgp2_pressBCinlet: success.
darcy_p1dgp2_pressBCinlet: Running change_P for 2d should equal domain_length*visc*darcy_vel_BC/perm, check relative difference to be under tolerance 1.0e-09:
Relative error of pressure difference: 1.58945719401e-16
darcy_p1dgp2_pressBCinlet: success.
darcy_p1dgp2_pressBCinlet: Running interstitial velocity 2d should equal darcy_vel/porosity, check relative difference to be under tolerance 1.0e-09:
Relative error of interstitial velocity: 5.62437740825e-10
darcy_p1dgp2_pressBCinlet: success.
darcy_p1dgp2_pressBCinlet: Running change_P for 3d should equal domain_length*visc*darcy_vel_BC/perm, check relative difference to be under tolerance 1.0e-09:
Relative error of pressure difference: 1.58945719401e-16
darcy_p1dgp2_pressBCinlet: success.
darcy_p1dgp2_pressBCinlet: Running interstitial velocity 3d should equal darcy_vel/porosity, check relative difference to be under tolerance 1.0e-09:
Relative error of interstitial velocity: 1.20875185416e-09
darcy_p1dgp2_pressBCinlet: failure.
darcy_p1dgp2_pressBCinlet: Running poisson related change_P for 2d should equal domain_length*visc*darcy_vel_BC/perm, check relative difference to be under tolerance 1.0e-09:
Relative error of pressure difference: 1.58945719401e-16
darcy_p1dgp2_pressBCinlet: success.
darcy_p1dgp2_pressBCinlet: Running poisson related interstitial velocity 2d should equal darcy_vel/porosity, check relative difference to be under tolerance 1.0e-09:
Relative error of interstitial velocity: 5.6243631974e-10
darcy_p1dgp2_pressBCinlet: success.
darcy_p1dgp2_pressBCinlet: Running poisson related change_P for 3d should equal domain_length*visc*darcy_vel_BC/perm, check relative difference to be under tolerance 1.0e-09:
Relative error of pressure difference: 1.58945719401e-16
darcy_p1dgp2_pressBCinlet: success.
darcy_p1dgp2_pressBCinlet: Running poisson related interstitial velocity 3d should equal darcy_vel/porosity, check relative difference to be under tolerance 1.0e-09:
Relative error of interstitial velocity: 1.20875274234e-09
darcy_p1dgp2_pressBCinlet: failure.
darcy_p1dgp2_pressBCinlet: PPPPPFPPPF

Summary of test problems with failures or warnings:
darcy_p1dgp2_pressBCinlet.xml: PPPPPFPPPF

[Tim Greaves (tim-greaves)]

Two orphan tests, I fear - Brendan has moved on to other employment now. Anyone interested in picking up these tests?

[Stephan Kramer (s-kramer)]

I nominate Rhodri to fix this. If noone apparently cares enough about these tests, can you change the tolerances and commit? (to github obviously!)