Skip to content

Symmetric tangential constraints - #248

Open
pbrubeck wants to merge 8 commits into
mainfrom
pbrubeck/gn-moments
Open

pbrubeck wants to merge 8 commits into
mainfrom
pbrubeck/gn-moments

Conversation

@pbrubeck

@pbrubeck pbrubeck commented May 12, 2026

Copy link
Copy Markdown

Moments that enforce degree constraints are now against symmetric functions obtained by projecting a bubble into the set of homogeneous degree-k polynomials Pk \ P_{k-1}.

(1 - proj_{k-1}) bubble_k = sum_j c_j q_j for q_j in Pk \ P_{k-1}

This is merely a cleanup and does not affect the nodal bases for the elements affected.

pbrubeck and others added 4 commits May 12, 2026 15:56
Add make_projected_bubble_moment, which projects the bubble onto
polynomials of the given degree and drops the components of lower
degree.  What remains is orthogonal to P_{degree-1} and invariant under
the symmetry group of the cell, which is what a constraint functional
annihilating that lower-degree space needs of its weight.

Bernardi-Raugel builds its tangential constraint weight this way
already; move that construction into the helper.  Its own scaling is
invariant under the helper's normalization, so the weight is unchanged.

Walkington's face constraint was a cubic interpolating [0, 0, 0, 1] at
four quadrature points, with the rule then truncated to the points where
the result happened to be nonzero.  That weight is invariant under the
three rotations of the face but not under the reflections.  The
projected bubble is the unique symmetric cubic orthogonal to P_2, so the
truncation is no longer available: the face rule goes from 6 points to
12.  The constraint dofs are discarded before the FInAT dual basis is
built, so this is paid once at element construction.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Comment thread FIAT/bubble.py Outdated

@rckirby rckirby left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It would help me reviewing if you would put in a description to the PR -- is this a clean-up exercise, fixing something that was broken, or adding a new feature? I thought we already had BR and Walkington working?

Comment thread FIAT/bubble.py Outdated
Comment thread FIAT/polynomial_set.py Outdated
@pbrubeck

pbrubeck commented Sep 23, 2026

Copy link
Copy Markdown
Author

It would help me reviewing if you would put in a description to the PR -- is this a clean-up exercise, fixing something that was broken, or adding a new feature? I thought we already had BR and Walkington working?

Description updated. This is just a cleanup for BR and Walkington, and facilitates imposing higher degree face constraints or getting hierarchical face bubbles for future elements.

@rckirby
rckirby self-requested a review September 23, 2026 21:36

@rckirby rckirby left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for clarifying in the comments, etc.

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants