Overhaul of `ConvergenceTest` namespace by sebglane · Pull Request #152 · LKM-code-base/RotatingMHD

sebglane · 2021-08-20T16:10:24Z

No description provided.

sebglane · 2021-08-20T16:10:47Z

Tests are missing...

… of tables

sebglane · 2021-08-25T14:51:56Z

Relevant output of fix_convergence_results for TGV.cc

Velocity convergence table
================================================================================
level  dofs  cells refinements cell diameter time step         L2               Linfty               H1         
1     132098 16384 8           1.10e-02      1.00e-01  2.569818e-02 -     3.885917e-02 -     3.094638e-01 -     
2     132098 16384 8           1.10e-02      5.00e-02  7.593713e-03 -3.52 1.083262e-02 -3.69 8.404851e-02 -3.76 
3     132098 16384 8           1.10e-02      2.50e-02  2.049094e-03 -3.78 2.862242e-03 -3.84 2.201931e-02 -3.86 
4     132098 16384 8           1.10e-02      1.25e-02  5.317996e-04 -3.89 7.367094e-04 -3.92 5.656499e-03 -3.92 
5     132098 16384 8           1.10e-02      6.25e-03  1.354414e-04 -3.95 1.870005e-04 -3.96 1.458189e-03 -3.91 


Pressure convergence table
================================================================================
level dofs  cells refinements cell diameter time step         L2               Linfty               H1         
1     16641 16384 8           1.10e-02      1.00e-01  8.840995e-03 -     2.528316e-02 -     1.228554e-01 -     
2     16641 16384 8           1.10e-02      5.00e-02  2.520948e-03 -3.62 6.771660e-03 -3.80 3.849286e-02 -3.35 
3     16641 16384 8           1.10e-02      2.50e-02  6.758555e-04 -3.80 1.832223e-03 -3.77 2.059895e-02 -1.80 
4     16641 16384 8           1.10e-02      1.25e-02  1.751759e-04 -3.90 5.321705e-04 -3.57 1.856231e-02 -0.30 
5     16641 16384 8           1.10e-02      6.25e-03  4.461541e-05 -3.95 1.968393e-04 -2.87 1.838439e-02 -0.03

and that of master

                               Velocity convergence table
==============================================================================================
level    dt    cells  dofs    hmax           L2                 H1               Linfty       
    7 1.00e-01 16384 132098 1.10e-02 2.569818e-02     - 3.094638e-01     - 3.885917e-02     - 
    7 5.00e-02 16384 132098 1.10e-02 7.593713e-03 -1.76 8.404851e-02 -1.88 1.083262e-02 -1.84 
    7 2.50e-02 16384 132098 1.10e-02 2.049094e-03 -1.89 2.201931e-02 -1.93 2.862242e-03 -1.92 
    7 1.25e-02 16384 132098 1.10e-02 5.317996e-04 -1.95 5.656499e-03 -1.96 7.367094e-04 -1.96 
    7 6.25e-03 16384 132098 1.10e-02 1.354414e-04 -1.97 1.458189e-03 -1.96 1.870005e-04 -1.98 

                               Pressure convergence table
==============================================================================================
level    dt    cells dofs    hmax           L2                 H1               Linfty       
    7 1.00e-01 16384 16641 1.10e-02 8.840995e-03     - 1.228554e-01     - 2.528316e-02     - 
    7 5.00e-02 16384 16641 1.10e-02 2.520948e-03 -1.81 3.849286e-02 -1.67 6.771660e-03 -1.90 
    7 2.50e-02 16384 16641 1.10e-02 6.758555e-04 -1.90 2.059895e-02 -0.90 1.832223e-03 -1.89 
    7 1.25e-02 16384 16641 1.10e-02 1.751759e-04 -1.95 1.856231e-02 -0.15 5.321705e-04 -1.78 
    7 6.25e-03 16384 16641 1.10e-02 4.461541e-05 -1.97 1.838439e-02 -0.01 1.968393e-04 -1.43

The values of the error are the same, rates are different.

j507 · 2021-08-25T15:18:34Z

In fix_convergence_results the rates are computed with evaluate_all_convergence_rates. I had been using evaluate_convergence_rates where I can manually set dim = 1. Maybe this is the culprit.

sebglane · 2021-08-25T15:18:53Z

Relevant output of fix_convergence_results for Guermond.cc

Velocity convergence table
================================================================================
level  dofs  cells refinements cell diameter time step         L2               Linfty               H1         
1     132098 16384 8           1.10e-02      1.00e-01  8.553226e-03 -     2.984499e-02 -     1.802801e-01 -     
2     132098 16384 8           1.10e-02      5.00e-02  2.106748e-03 -4.04 3.850347e-03 -5.91 2.427610e-02 -5.79 
3     132098 16384 8           1.10e-02      2.50e-02  5.270936e-04 -4.00 6.671879e-04 -5.06 5.897830e-03 -4.08 
4     132098 16384 8           1.10e-02      1.25e-02  1.322852e-04 -3.99 1.668826e-04 -4.00 1.636175e-03 -3.70 
5     132098 16384 8           1.10e-02      6.25e-03  3.319652e-05 -3.99 4.178635e-05 -4.00 4.617877e-04 -3.65 


Pressure convergence table
================================================================================
level dofs  cells refinements cell diameter time step         L2               Linfty               H1         
1     16641 16384 8           1.10e-02      1.00e-01  8.728770e-03 -     4.231686e-02 -     1.034468e-01 -     
2     16641 16384 8           1.10e-02      5.00e-02  1.888946e-03 -4.42 1.387980e-02 -3.22 1.816496e-02 -5.02 
3     16641 16384 8           1.10e-02      2.50e-02  4.692749e-04 -4.02 4.332481e-03 -3.36 5.317010e-03 -3.54 
4     16641 16384 8           1.10e-02      1.25e-02  1.190634e-04 -3.96 1.337364e-03 -3.39 3.035792e-03 -1.62 
5     16641 16384 8           1.10e-02      6.25e-03  3.683274e-05 -3.39 4.172894e-04 -3.36 2.755555e-03 -0.28

and that of master

                               Velocity convergence table
==============================================================================================
level    dt    cells  dofs    hmax           L2                 H1               Linfty       
    7 1.00e-01 16384 132098 1.10e-02 8.553226e-03     - 1.802801e-01     - 2.984499e-02     - 
    7 5.00e-02 16384 132098 1.10e-02 2.106748e-03 -2.02 2.427610e-02 -2.89 3.850347e-03 -2.95 
    7 2.50e-02 16384 132098 1.10e-02 5.270936e-04 -2.00 5.897830e-03 -2.04 6.671879e-04 -2.53 
    7 1.25e-02 16384 132098 1.10e-02 1.322852e-04 -1.99 1.636175e-03 -1.85 1.668826e-04 -2.00 
    7 6.25e-03 16384 132098 1.10e-02 3.319652e-05 -1.99 4.617877e-04 -1.83 4.178635e-05 -2.00 

                               Pressure convergence table
==============================================================================================
level    dt    cells dofs    hmax           L2                 H1               Linfty       
    7 1.00e-01 16384 16641 1.10e-02 8.728745e-03     - 1.034468e-01     - 4.230299e-02     - 
    7 5.00e-02 16384 16641 1.10e-02 1.888833e-03 -2.21 1.816494e-02 -2.51 1.386592e-02 -1.61 
    7 2.50e-02 16384 16641 1.10e-02 4.688182e-04 -2.01 5.316969e-03 -1.77 4.318606e-03 -1.68 
    7 1.25e-02 16384 16641 1.10e-02 1.172501e-04 -2.00 3.035722e-03 -0.81 1.323489e-03 -1.71 
    7 6.25e-03 16384 16641 1.10e-02 3.046558e-05 -1.94 2.755478e-03 -0.14 4.034146e-04 -1.71

sebglane · 2021-08-25T16:01:29Z

In fix_convergence_results the rates are computed with evaluate_all_convergence_rates. I had been using evaluate_convergence_rates where I can manually set dim = 1. Maybe this is the culprit.

Fixed in ecb7d4f

sebglane · 2021-08-26T07:00:57Z

New convergence tables for ecb7d4f

TGV.cc

Velocity convergence table
================================================================================
level  dofs  cells refinements    h     time step         L2               Linfty               H1         
1     132098 16384 8           1.10e-02 1.00e-01  2.569818e-02 -     3.885917e-02 -     3.094638e-01 -     
2     132098 16384 8           1.10e-02 5.00e-02  7.593713e-03 -1.76 1.083262e-02 -1.84 8.404851e-02 -1.88 
3     132098 16384 8           1.10e-02 2.50e-02  2.049094e-03 -1.89 2.862242e-03 -1.92 2.201931e-02 -1.93 
4     132098 16384 8           1.10e-02 1.25e-02  5.317996e-04 -1.95 7.367094e-04 -1.96 5.656499e-03 -1.96 
5     132098 16384 8           1.10e-02 6.25e-03  1.354414e-04 -1.97 1.870005e-04 -1.98 1.458189e-03 -1.96 


Pressure convergence table
================================================================================
level dofs  cells refinements    h     time step         L2               Linfty               H1         
1     16641 16384 8           1.10e-02 1.00e-01  8.840995e-03 -     2.528316e-02 -     1.228554e-01 -     
2     16641 16384 8           1.10e-02 5.00e-02  2.520948e-03 -1.81 6.771660e-03 -1.90 3.849286e-02 -1.67 
3     16641 16384 8           1.10e-02 2.50e-02  6.758555e-04 -1.90 1.832223e-03 -1.89 2.059895e-02 -0.90 
4     16641 16384 8           1.10e-02 1.25e-02  1.751759e-04 -1.95 5.321705e-04 -1.78 1.856231e-02 -0.15 
5     16641 16384 8           1.10e-02 6.25e-03  4.461541e-05 -1.97 1.968393e-04 -1.43 1.838439e-02 -0.01

Guermond.cc

Velocity convergence table
================================================================================
level  dofs  cells refinements    h     time step         L2               Linfty               H1         
1     132098 16384 8           1.10e-02 1.00e-01  8.553226e-03 -     2.984499e-02 -     1.802801e-01 -     
2     132098 16384 8           1.10e-02 5.00e-02  2.106748e-03 -2.02 3.850347e-03 -2.95 2.427610e-02 -2.89 
3     132098 16384 8           1.10e-02 2.50e-02  5.270936e-04 -2.00 6.671879e-04 -2.53 5.897830e-03 -2.04 
4     132098 16384 8           1.10e-02 1.25e-02  1.322852e-04 -1.99 1.668826e-04 -2.00 1.636175e-03 -1.85 
5     132098 16384 8           1.10e-02 6.25e-03  3.319652e-05 -1.99 4.178635e-05 -2.00 4.617877e-04 -1.83 


Pressure convergence table
================================================================================
level dofs  cells refinements    h     time step         L2               Linfty               H1         
1     16641 16384 8           1.10e-02 1.00e-01  8.728770e-03 -     4.231686e-02 -     1.034468e-01 -     
2     16641 16384 8           1.10e-02 5.00e-02  1.888946e-03 -2.21 1.387980e-02 -1.61 1.816496e-02 -2.51 
3     16641 16384 8           1.10e-02 2.50e-02  4.692749e-04 -2.01 4.332481e-03 -1.68 5.317010e-03 -1.77 
4     16641 16384 8           1.10e-02 1.25e-02  1.190634e-04 -1.98 1.337364e-03 -1.70 3.035792e-03 -0.81 
5     16641 16384 8           1.10e-02 6.25e-03  3.683274e-05 -1.69 4.172894e-04 -1.68 2.755555e-03 -0.14

sebglane · 2021-08-26T15:16:09Z

@j507 I think that the implementation works. However, there is still the issue discussed in #118. We should check whether this PR is a problem or whether the current master branch also has the same issues. The changes made to ThermalTGV.cc should not make a difference.

j507 · 2021-08-31T14:27:54Z

 * @brief The source file solving the Couette flow problem.
 *
 */


We do not really use this program anymore. We can just delete it.

I would prefer to keep this program as a convergence test.

j507 · 2021-08-31T14:32:29Z

+    if (!std::filesystem::exists(this->prm.graphical_output_directory))
+    {
+      try
+      {
+        std::filesystem::create_directories(this->prm.graphical_output_directory);


If I am not mistaken this is already done in ProblemClass

j507 · 2021-08-31T14:45:48Z

+| Start time                               | 0                    |
+| Final time                               | 2                    |
+| Verbose                                  | false                |
+------------------------------------------+----------------------++------------------------------------------+----------------------+


Double printed?

sebglane · 2021-12-20T16:13:25Z

This PR should fix

To be checked with the current version.

Initial overhaul (tests missing)

c58bca6

sebglane marked this pull request as draft August 20, 2021 16:10

Sebastian Glane added 3 commits August 25, 2021 15:22

Merge branch 'master' into fix_convergence_results

1eb4553

Omitted column from computation of convergence rates, adjusted output…

01fcd1c

… of tables

Added tests

d0b86ce

sebglane mentioned this pull request Aug 25, 2021

[master] Hyperconvergence in the pure diffusion and advection-diffusion tests #147

Closed

Sebastian Glane added 2 commits August 25, 2021 17:37

Added parameter files of additional tests, fixed CMakeLists.txt

8d38096

Fixed computation of convergence rates, adjusted test output

ecb7d4f

This was linked to issues Aug 25, 2021

[master] Hyperconvergence in the pure diffusion and advection-diffusion tests #147

Closed

[master] Convergence tests are not executable when explicit time stepping is used #161

Closed

sebglane linked an issue Aug 26, 2021 that may be closed by this pull request

[master] Error increases in the thermal convergence test for very small timesteps #118

Open

Fixed computation of convergence rates in tests and application

56ea811

sebglane marked this pull request as ready for review August 26, 2021 07:09

Sebastian Glane added 2 commits August 26, 2021 09:49

Fixed convergence table in ThermalTGV.cc

f1a96f4

Fixed explicit test

b164467

sebglane added do not merge ☠ and removed do not merge ☠ labels Aug 26, 2021

sebglane requested a review from j507 August 31, 2021 10:43

j507 approved these changes Aug 31, 2021

View reviewed changes

Merge branch 'master' into fix_convergence_results

ae444e5

sebglane linked an issue Dec 20, 2021 that may be closed by this pull request

[master] Other time stepping schemes do not yield convergence #162

Open

Merge branch 'master' into fix_convergence_results

da0d8f4

Merge branch 'master' into fix_convergence_results

73b85c0

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Overhaul of `ConvergenceTest` namespace - #152

Overhaul of `ConvergenceTest` namespace#152
sebglane wants to merge 12 commits into
masterfrom
fix_convergence_results

sebglane commented Aug 20, 2021

Uh oh!

sebglane commented Aug 20, 2021

Uh oh!

sebglane commented Aug 25, 2021

Uh oh!

j507 commented Aug 25, 2021

Uh oh!

sebglane commented Aug 25, 2021

Uh oh!

sebglane commented Aug 25, 2021

Uh oh!

sebglane commented Aug 26, 2021

Uh oh!

sebglane commented Aug 26, 2021

Uh oh!

j507 Aug 31, 2021

Uh oh!

sebglane Dec 20, 2021

Uh oh!

j507 Aug 31, 2021

Uh oh!

j507 Aug 31, 2021

Uh oh!

sebglane commented Dec 20, 2021 •

edited

Loading

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

sebglane commented Aug 20, 2021

Uh oh!

sebglane commented Aug 20, 2021

Uh oh!

sebglane commented Aug 25, 2021

Uh oh!

j507 commented Aug 25, 2021

Uh oh!

sebglane commented Aug 25, 2021

Uh oh!

sebglane commented Aug 25, 2021

Uh oh!

sebglane commented Aug 26, 2021

Uh oh!

sebglane commented Aug 26, 2021

Uh oh!

j507 Aug 31, 2021

Choose a reason for hiding this comment

Uh oh!

sebglane Dec 20, 2021

Choose a reason for hiding this comment

Uh oh!

j507 Aug 31, 2021

Choose a reason for hiding this comment

Uh oh!

j507 Aug 31, 2021

Choose a reason for hiding this comment

Uh oh!

sebglane commented Dec 20, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

sebglane commented Dec 20, 2021 •

edited

Loading