Rotated free-slip: the custom-FMG velocity sub-KSP must converge, not just apply - #465

Open

lmoresi wants to merge 12 commits into

developmentfrom

bugfix/rotated-bc-preconditioner-anisotropy

Commits on Jul 27, 2026

feat(rotated-bc): prescribed-normal datum through the nonlinear SNES rotated path (#438)
lmoresi
committed
refactor(rotated-bc): datum-activity reduction via PETSc Vec norm, not bare MPI
lmoresi
committed

Commits on Jul 28, 2026

refactor(rotated-bc): one Newton loop for linear and nonlinear rotated solves; datum via conds (#438, #403)
lmoresi
committed
fix(rotated-bc): honest step-norm exit; operator-verified rotation null modes
lmoresi
committed
feat(free-surface): warm-start the consistent datum solve; reference-scale convergence for warm starts
lmoresi
committed
docs(constitutive): TODO(DESIGN) - verify the TI consistent tangent with the PETSc Jacobian checker
lmoresi
committed
docs(constitutive): TI consistent-tangent TODO upgraded to TODO(BUG) -> #457
lmoresi
committed
docs+test: rotated-freeslip subsystem page, changelog entry, np2 nonlinear datum test
lmoresi
committed
Merge origin/development: unified rotated dispatch + wall-clock guard refusal
lmoresi
committed

Commits on Jul 29, 2026