update inclusion benchmark #788
Conversation
bjs2
commented
Mar 9, 2016
bjs2
commented
- change output directories for global and adaptive
- include velocity and pressure dofs in error output line
- include option to turn on averaging
- change output directories for global and adaptive - include velocity and pressure dofs in error output line - include option to turn on averaging
|
@tjhei Is this work in progress, or is this open for review? |
|
I thing this is good to go. /run-tests |
| @@ -513,13 +513,16 @@ namespace aspect | |||
| VectorTools::L2_norm, | |||
| &comp_p); | |||
|
|
|||
| unsigned int n = this->get_solution().block(0).size() + | |||
| this->get_solution().block(1).size(); | |||
There was a problem hiding this comment.
It would be good to use introspection.block_indices.pressure (and the same for the velocity) instead of just 0 and 1 here. Also, if the direct solver is used, pressure and velocity are in the same block, and then the numbers 0 and 1 are suddenly not correct any more.
- Modify inclusion.cc to output correct number of DoF's if the direct solver is used - Modify run.sh to loop through different averaging schemes (none, arithmetic, geometric)
|
Thanks for updating the pull request! |
|
There should be an adaptive run for each type of averaging, and this was the case when I ran the shell script. I believe the run.sh has a nested for-loop. These are the results from when I ran it: https://docs.google.com/spreadsheets/d/1GYPWl4UOe-Jc5AkVqDX6PDQmL73YU-0S9rwBi7S-Ce0/edit?usp=sharing |
|
Ah, I get it, because you use the option to run the postprocessors on the initial refinement you get the output for every refinement step. |
