Finish support for AD time derivatives of dofs #28857
Conversation
GiudGiud
force-pushed
the
PR_lumped
branch
2 times, most recently
from
1bda2b7
to
f5a8d37
Compare
GiudGiud
changed the title
Add a test material to test those refs idaholab#18178
|
Job Documentation, step Docs: sync website on e71c341 wanted to post the following: View the site here This comment will be updated on new commits. |
GiudGiud
changed the title
… nonAD version) Add support for nodal AD time derivatives, with errors for unsupported cases closes idaholab#28859 refs idaholab#28867
|
Job Coverage, step Generate coverage on e71c341 wanted to post the following: Framework coverage
Modules coverageCoverage did not change Full coverage reportsReports
Warnings
This comment will be updated on new commits. |
||||||||||||||||||||||||||
| @@ -428,6 +428,11 @@ class MooseVariableData : public MooseVariableDataBase<OutputType> | |||
| */ | |||
| const MooseArray<ADReal> & adDofValues() const; | |||
|
|
|||
| /** | |||
| * Return the AD time derivative values | |||
There was a problem hiding this comment.
| * Return the AD time derivative values | |
| * Return the AD time derivative values of degrees of freedom |
| else | ||
| { | ||
| if (_c_nodal) | ||
| mooseError("AD neighbor nodal dof dot implemented"); |
There was a problem hiding this comment.
| mooseError("AD neighbor nodal dof dot implemented"); | |
| mooseError("AD neighbor nodal dof dot not implemented"); |
| if (_need_ad_u_dot) | ||
| { | ||
| if (_time_integrator && _time_integrator->dt()) | ||
| { | ||
| _ad_dofs_dot[i] = _ad_dof_values[i]; | ||
| _time_integrator->computeADTimeDerivatives(_ad_dofs_dot[i], | ||
| _dof_indices[i], | ||
| _need_ad_u_dotdot ? _ad_dofs_dotdot[i] | ||
| : _ad_real_dummy); | ||
| } | ||
| // Executing something with a time derivative at initial should not put a NaN | ||
| else if (_time_integrator && !_time_integrator->dt()) | ||
| _ad_dofs_dot[i] = 0; | ||
| else | ||
| mooseError("AD nodal time derivatives not implemented for variables without a time " | ||
| "integrator (auxiliary variables)"); | ||
| } |
There was a problem hiding this comment.
computeNodalValues is the only caller of fetchADDoFValues. I've always thought we should sync the names better. As it is, I think it's a bit confusing
| InputParameters | ||
| CoupledAux::validParams() | ||
| CoupledAuxTempl<is_ad>::validParams() |
There was a problem hiding this comment.
It seems like these changes are really only for testing purposes? Is there any reason in production to use an ADCoupledAux?
There was a problem hiding this comment.
And I see that you are modifying a test material ... so do we really need these aux kernel changes?
There was a problem hiding this comment.
I thought the changes didn't overlap but they might have ended up overlapping. I ll check
No reason to use ADcoupledAux in production. If they don't overlap do you prefer creating a new object in test/ over templating in the framework ?
also includes a bug fix for second order AD dot dots
closes #28859
refs #28867