proteus
proteus copied to clipboard
Published
20 hours ago •
erdc
erdc
Dambreak_solver_options test redesign
Problem mentioned in issue #1056.
The TwoPhaseFlow/dambreak_solver_options test fails via the hashdist build but not via the conda build.
=================================== FAILURES ===================================
[31m[1m________________ TestTwoPhaseFlow.test_damBreak_solver_options _________________[0m
[gw0] linux -- Python 3.7.2 /home/travis/build/erdc/proteus/linux/bin/python
self = <proteus.tests.TwoPhaseFlow.test_TwoPhaseFlow.TestTwoPhaseFlow object at 0x7ccec18>
def test_damBreak_solver_options(self):
os.system("parun --TwoPhaseFlow --path " + self.path + " "
"damBreak_solver_options.py -l5 -v -C 'final_time=0.1 dt_output=0.1 he=0.1'")
> self.compare_vs_saved_files("damBreak_solver_options")
proteus/tests/TwoPhaseFlow/test_TwoPhaseFlow.py:67:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
self = <proteus.tests.TwoPhaseFlow.test_TwoPhaseFlow.TestTwoPhaseFlow object at 0x7ccec18>
name = 'damBreak_solver_options'
def compare_vs_saved_files(self,name):
actual = tables.open_file(name+'.h5','r')
expected_path = 'comparison_files/' + 'comparison_' + name + '_phi_t2.csv'
#write comparison file
#np.array(actual.root.phi_t2).tofile(os.path.join(self._scriptdir, expected_path),sep=",")
> np.testing.assert_almost_equal(np.fromfile(os.path.join(self._scriptdir, expected_path),sep=","),np.array(actual.root.phi_t2).flatten(),decimal=10)
E AssertionError:
E Arrays are not almost equal to 10 decimals
E
E Mismatch: 100%
E Max absolute difference: 0.12644174
E Max relative difference: 1.27583147
E x: array([-0.6281821969, 1.9642545046, 2.5372557862, 1.1639619602,
E 0.7474668495, 1.3850260196, 2.1816673879, 1.729152002 ,
E 0.9487286845, 1.2566472109, 1.5379914368, 1.3029162311,...
E y: array([-0.6391312677, 2.0310413596, 2.5577650745, 1.1762542886,
E 0.8369607758, 1.3825964545, 2.2229598418, 1.7347065252,
E 0.9678189718, 1.2609615221, 1.5385438642, 1.3023196779,...
proteus/tests/TwoPhaseFlow/test_TwoPhaseFlow.py:44: AssertionError
In fact, the numerics fail fairly early on.
[ 0] ==============================================================
[ 0] Solving over interval [ 0.00000e+00, 1.00000e-03]
[ 0] ==============================================================
[ 0] Min_dt_controller setSubsteps tList=[0.001] self.t_model=0.001 self.substeps= [0.001]
[ 0] Min_dt_controller setSubsteps tList=[0.001] self.t_model=0.001 self.substeps= [0.001]
[ 0] System time step t= 1.00000e-03, dt= 1.00000e-03
[ 0] saving previous velocity dofs 0
[ 0] Split operator iteration 0
[ 0] Model: rans2p
[ 0] Fractional step 1.00000e-03 for model rans2p
[ 0] Model step t= 1.00000e-03, dt= 1.00000e-03 for model rans2p
[ 0] Model substep t= 1.00000e-03 for model rans2p
[ 0] Model substep t= 1.00000e-03 for model rans2p model.timeIntegration.t= 1.00000e-03
[ 0] Maximum CFL = 0.0
[ 0] Newton it 0 norm(r) = 5.81811e+00 norm(r)/(rtol*norm(r0)+atol) = 5.81811e+09 test=r
[ 0] before ksp.rtol= 1e-09 ksp.atol= 1e-20 ksp.converged= False ksp.its= 0 ksp.norm= 0.0
[ 0] after ksp.rtol= 1e-09 ksp.atol= 1e-20 ksp.converged= True ksp.its= 22 ksp.norm= 2.7198399351813935e-09 reason = 2
[ 0] Maximum CFL = 4.2185181435780866e-05
[ 0] Newton it 1 norm(r) = 5.06826e-03 norm(r)/(rtol*norm(r0)+atol) = 5.06826e+06 test=r
[ 0] before ksp.rtol= 1e-09 ksp.atol= 1e-20 ksp.converged= False ksp.its= 22 ksp.norm= 2.7198399351813935e-09
[ 0] after ksp.rtol= 1e-09 ksp.atol= 1e-20 ksp.converged= True ksp.its= 22 ksp.norm= 1.0294966186689725e-12 reason = 2
[ 0] Maximum CFL = 4.227390254218761e-05
[ 0] Newton it 2 norm(r) = 1.35207e-05 norm(r)/(rtol*norm(r0)+atol) = 13520.7 test=r
[ 0] before ksp.rtol= 1e-09 ksp.atol= 1e-20 ksp.converged= False ksp.its= 22 ksp.norm= 1.0294966186689725e-12
[ 0] after ksp.rtol= 1e-09 ksp.atol= 1e-20 ksp.converged= True ksp.its= 22 ksp.norm= 6.193521028209155e-15 reason = 2
[ 0] Maximum CFL = 4.227385204190255e-05
[ 0] Newton it 3 norm(r) = 1.17514e-08 norm(r)/(rtol*norm(r0)+atol) = 11.7514 test=r
[ 0] before ksp.rtol= 1e-09 ksp.atol= 1e-20 ksp.converged= False ksp.its= 22 ksp.norm= 6.193521028209155e-15
[ 0] after ksp.rtol= 1e-09 ksp.atol= 1e-20 ksp.converged= True ksp.its= 22 ksp.norm= 5.940964936203353e-18 reason = 2
[ 0] Maximum CFL = 4.2273852020229116e-05
[ 0] Newton it 4 norm(r) = 1.98236e-11 norm(r)/(rtol*norm(r0)+atol) = 1.98236e-02
[ 0] Step Taken, t_stepSequence= 0.001 Model step t= 1.00000e-03, dt= 1.00000e-03 for model rans2p
[ 0] Model: clsvof
[ 0] Fractional step 1.00000e-03 for model clsvof
[ 0] Model step t= 1.00000e-03, dt= 1.00000e-03 for model clsvof
[ 0] Model substep t= 1.00000e-03 for model clsvof
[ 0] Model substep t= 1.00000e-03 for model clsvof model.timeIntegration.t= 1.00000e-03
[ 0] +++++ Spin up to start with disc ICs +++++
[ 0] ... Normal reconstruction via weighted lumped L2-projection ...
[ 0] Newton it 0 norm(r) = 5.58950e-01 norm(r)/(rtol*norm(r0)+atol) = 539.09 test=r
[ 0] before ksp.rtol= 0.001 ksp.atol= 0.00010368400000000005 ksp.converged= False ksp.its= 0 ksp.norm= 0.0
[ 0] after ksp.rtol= 0.001 ksp.atol= 0.00010368400000000005 ksp.converged= True ksp.its= 10 ksp.norm= 0.0030050066645684026 reason = 2
[ 0] ls #1 norm_r_cur=0.919566224827264 atol=0.00103684 rtol=0
[ 0] ls #2 norm_r_cur=0.2800604074327376 atol=0.00103684 rtol=0
[ 0] Linesearches = 2
[ 0] Newton it 1 == maxIts FAILED convergenceTest = r
[ 0] Newton it 1 norm(r) = 2.80060e-01 norm(r)/(rtol*norm(r0)+atol) = 2.70110e+02
[ 0] ... Normal reconstruction via weighted lumped L2-projection ...
[ 0] Newton it 0 norm(r) = 3.82157e-01 norm(r)/(rtol*norm(r0)+atol) = 368.578 test=r
[ 0] before ksp.rtol= 0.001 ksp.atol= 0.00010368400000000005 ksp.converged= False ksp.its= 10 ksp.norm= 0.0030050066645684026
[ 0] after ksp.rtol= 0.001 ksp.atol= 0.00010368400000000005 ksp.converged= True ksp.its= 6 ksp.norm= 0.0013542895798390295 reason = 2
[ 0] Newton it 1 == maxIts FAILED convergenceTest = r
[ 0] Newton it 1 norm(r) = 3.81313e-02 norm(r)/(rtol*norm(r0)+atol) = 3.67764e+01
[ 0] ... Normal reconstruction via weighted lumped L2-projection ...
[ 0] Newton it 0 norm(r) = 7.78754e-02 norm(r)/(rtol*norm(r0)+atol) = 75.1084 test=r
[ 0] before ksp.rtol= 0.001 ksp.atol= 0.00010368400000000005 ksp.converged= False ksp.its= 6 ksp.norm= 0.0013542895798390295
[ 0] after ksp.rtol= 0.001 ksp.atol= 0.00010368400000000005 ksp.converged= True ksp.its= 6 ksp.norm= 8.474135011046035e-05 reason = 3
[ 0] Newton it 1 == maxIts FAILED convergenceTest = r
[ 0] Newton it 1 norm(r) = 3.12001e-03 norm(r)/(rtol*norm(r0)+atol) = 3.00916e+00
[ 0] ... Normal reconstruction via weighted lumped L2-projection ...
[ 0] Newton it 0 norm(r) = 2.07533e-02 norm(r)/(rtol*norm(r0)+atol) = 20.0159 test=r
[ 0] before ksp.rtol= 0.001 ksp.atol= 0.00010368400000000005 ksp.converged= False ksp.its= 6 ksp.norm= 8.474135011046035e-05
[ 0] after ksp.rtol= 0.001 ksp.atol= 0.00010368400000000005 ksp.converged= True ksp.its= 6 ksp.norm= 3.143365220490628e-05 reason = 3
[ 0] Newton it 1 norm(r) = 3.85318e-04 norm(r)/(rtol*norm(r0)+atol) = 3.71627e-01
[ 0] ... Normal reconstruction via weighted lumped L2-projection ...
[ 0] Newton it 0 norm(r) = 4.70787e-01 norm(r)/(rtol*norm(r0)+atol) = 454.059 test=r
[ 0] before ksp.rtol= 0.001 ksp.atol= 0.00010368400000000005 ksp.converged= False ksp.its= 6 ksp.norm= 3.143365220490628e-05
[ 0] after ksp.rtol= 0.001 ksp.atol= 0.00010368400000000005 ksp.converged= True ksp.its= 4 ksp.norm= 9.323844517860051e-06 reason = 3
[ 0] Newton it 1 == maxIts FAILED convergenceTest = r
[ 0] Newton it 1 norm(r) = 1.14745e-02 norm(r)/(rtol*norm(r0)+atol) = 1.10668e+01
[ 0] ... Normal reconstruction via weighted lumped L2-projection ...
[ 0] Newton it 0 norm(r) = 1.99105e-02 norm(r)/(rtol*norm(r0)+atol) = 19.203 test=r
[ 0] before ksp.rtol= 0.001 ksp.atol= 0.00010368400000000005 ksp.converged= False ksp.its= 4 ksp.norm= 9.323844517860051e-06
[ 0] after ksp.rtol= 0.001 ksp.atol= 0.00010368400000000005 ksp.converged= True ksp.its= 3 ksp.norm= 1.7797814134492084e-05 reason = 3
[ 0] Newton it 1 norm(r) = 9.77294e-05 norm(r)/(rtol*norm(r0)+atol) = 9.42569e-02
[ 0] ... Normal reconstruction via weighted lumped L2-projection ...
[ 0] Newton it 0 norm(r) = 1.64986e-01 norm(r)/(rtol*norm(r0)+atol) = 159.124 test=r
[ 0] before ksp.rtol= 0.001 ksp.atol= 0.00010368400000000005 ksp.converged= False ksp.its= 3 ksp.norm= 1.7797814134492084e-05
[ 0] after ksp.rtol= 0.001 ksp.atol= 0.00010368400000000005 ksp.converged= True ksp.its= 7 ksp.norm= 0.0001335953675773493 reason = 2
[ 0] Newton it 1 == maxIts FAILED convergenceTest = r
[ 0] Newton it 1 norm(r) = 6.49833e-03 norm(r)/(rtol*norm(r0)+atol) = 6.26744e+00
[ 0] ... Normal reconstruction via weighted lumped L2-projection ...
[ 0] Newton it 0 norm(r) = 1.18863e-01 norm(r)/(rtol*norm(r0)+atol) = 114.64 test=r
[ 0] before ksp.rtol= 0.001 ksp.atol= 0.00010368400000000005 ksp.converged= False ksp.its= 7 ksp.norm= 0.0001335953675773493
[ 0] after ksp.rtol= 0.001 ksp.atol= 0.00010368400000000005 ksp.converged= True ksp.its= 6 ksp.norm= 6.911448004381769e-05 reason = 3
[ 0] Newton it 1 norm(r) = 8.04089e-04 norm(r)/(rtol*norm(r0)+atol) = 7.75519e-01
[ 0] +++++ Pre re-distancing +++++
[ 0] ... Normal reconstruction via weighted lumped L2-projection ...
[ 0] Newton it 0 norm(r) = 9.63670e-03 norm(r)/(rtol*norm(r0)+atol) = 9.2943 test=r
[ 0] before ksp.rtol= 0.001 ksp.atol= 0.00010368400000000005 ksp.converged= False ksp.its= 6 ksp.norm= 6.911448004381769e-05
[ 0] after ksp.rtol= 0.001 ksp.atol= 0.00010368400000000005 ksp.converged= True ksp.its= 6 ksp.norm= 4.617002923309927e-05 reason = 3
[ 0] Newton it 1 norm(r) = 2.13174e-04 norm(r)/(rtol*norm(r0)+atol) = 2.05600e-01
[ 0] +++++ Nonlinear CLSVOF +++++
[ 0] ... Normal reconstruction via weighted lumped L2-projection ...
[ 0] Newton it 0 norm(r) = 1.77596e+00 norm(r)/(rtol*norm(r0)+atol) = 1712.86 test=r
[ 0] before ksp.rtol= 0.001 ksp.atol= 0.00010368400000000005 ksp.converged= False ksp.its= 6 ksp.norm= 4.617002923309927e-05
[ 0] after ksp.rtol= 0.001 ksp.atol= 0.00010368400000000005 ksp.converged= True ksp.its= 9 ksp.norm= 6.919419918451551e-05 reason = 3
[ 0] Newton it 1 norm(r) = 6.82798e-02 norm(r)/(rtol*norm(r0)+atol) = 65.8537 test=r
[ 0] before ksp.rtol= 0.001 ksp.atol= 0.00010368400000000005 ksp.converged= False ksp.its= 9 ksp.norm= 6.919419918451551e-05
[ 0] after ksp.rtol= 0.001 ksp.atol= 0.00010368400000000005 ksp.converged= True ksp.its= 2 ksp.norm= 3.63898829685863e-05 reason = 3
[ 0] Newton it 2 norm(r) = 1.97502e-04 norm(r)/(rtol*norm(r0)+atol) = 1.90484e-01
[ 0] Step Taken, t_stepSequence= 0.001 Model step t= 1.00000e-03, dt= 1.00000e-03 for model clsvof
[ 0] Step Taken, Model step t= 1.00000e-03, dt= 1.00000e-03 for model clsvof
[ 0] RANS2P.SubgridError: switched to lagged subgrid error
[ 0] RANS2P.ShockCapturing: switched to lagged shock capturing
[ 0] RANS2P: max numDiff_1 3.423463e-01 numDiff_2 3.423463e-01 numDiff_3 3.423463e-01
[ 0] Min_dt_controller setSubsteps tList=[0.003] self.t_model=0.003 self.substeps= [0.003]
[ 0] Min_dt_controller setSubsteps tList=[0.003] self.t_model=0.003 self.substeps= [0.003]
[ 0] Step Taken, System time step t= 1.00000e-03, dt= 1.00000e-03
[ 0] Potential System time step t= 3.00000e-03, dt= 2.00000e-03 for next step
[ 0] Calculate called at time 0.001
[ 0] Calculate called at time 0.001
[ 0] Writing mesh header for model = rans2p at time t=0.001
[ 0] Writing solution for model = rans2p
[ 0] Writing mesh header for model = clsvof at time t=0.001
[ 0] Writing solution for model = clsvof
[ 0] Writing initial vof from clsvof at time t=0.001
[ 0] Writing quantity of interest at DOFs for = clsvof at time t=0.001
[ 0] Syncing Archive
[ 0] Gathering Archive Time step
[ 0] Done Gathering Archive Time Step
[ 0] Done Syncing Archive
My experience was that refining the mesh did not resolve this issue, which suggests that the test should be redesigned.
