Allow fully traction-free boundaries for elasticity problem#466
Allow fully traction-free boundaries for elasticity problem#466masterleinad wants to merge 1 commit intoadamantine-sim:masterfrom
Conversation
Summary of ChangesHello, I'm Gemini Code Assist1! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed! This pull request introduces a crucial enhancement to the elasticity problem solver by automatically handling fully traction-free boundary conditions. It ensures the stability and convergence of numerical methods, particularly for multigrid solvers like MueLU/ML, by implicitly restricting rigid body translation modes when no explicit boundary constraints are defined. Highlights
🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console. Using Gemini Code AssistThe full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips. Invoking Gemini You can request assistance from Gemini at any point by creating a comment using either
Customization To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a Limitations & Feedback Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for GitHub and other Google products, sign up here. Footnotes
|
There was a problem hiding this comment.
Code Review
This pull request introduces logic to automatically apply affine constraints to a boundary node when no explicit boundary IDs are provided, executed on MPI rank 0, to prevent rigid body modes. The current method for selecting and constraining a node on the boundary is identified as potentially fragile and difficult to maintain due to its reliance on internal deal.II DoF ordering, and a more robust approach by explicitly selecting a vertex for constraints has been suggested.
| const dealii::FiniteElement<dim> &fe = cell->get_fe(); | ||
|
|
||
| const unsigned int dofs_per_face = fe.n_dofs_per_face(face); | ||
| std::vector<dealii::types::global_dof_index> face_dof_indices( | ||
| dofs_per_face); | ||
| cell->face(face)->get_dof_indices(face_dof_indices, | ||
| cell->active_fe_index()); | ||
|
|
||
| std::array<unsigned int, dim> boundary_dofs; | ||
| for (unsigned int i = 0; i < dofs_per_face; ++i) | ||
| boundary_dofs[fe.face_system_to_component_index(i, face).first] = | ||
| face_dof_indices[i]; | ||
| for (int i = 0; i < dim; ++i) | ||
| _affine_constraints.add_line(boundary_dofs[i]); |
There was a problem hiding this comment.
The logic to select a node on the boundary for constraining is difficult to follow. It relies on implementation details of DoF ordering on a face, which makes it fragile and hard to maintain.
A more direct and robust approach is to explicitly select a vertex on the boundary face and constrain its degrees of freedom. This makes the code's intent clear and is less dependent on internal deal.II orderings.
// Pin the first vertex of the face to remove translational rigid body modes.
const unsigned int vertex_on_face = 0;
const unsigned int vertex_on_cell =
cell->face_vertex_index(face, vertex_on_face);
for (unsigned int c = 0; c < dim; ++c)
_affine_constraints.add_line(
cell->vertex_dof_index(vertex_on_cell, c));
Restricting translation modes seems to be sufficient for MueLU/ML to converge.