Repository navigation
Adjoint field dual is silently zero for a field that enters through a Dirichlet datum #787
Description
Activity
Not reproducible on
development(7c9cbbfd) — the adjoint API is not there:uw.adjoint present: False Stokes adjoint API: [] files with 'adjoint' in the path: 0Only scattered references remain (one line in
petsc_generic_snes_solvers.pyx, five inmcp/__init__.py, eight intranscript_query.py), not the implementation. So this defect lives in unmerged work —feature/discrete-adjoint(#744, currently conflicting and red),feature/adjoint-rotated-bc(#751, based on #744), orscratch/adjoint-merge.That is why it has sat untouched: there is nothing on the main line to fix. It is not
fixed-in-PR— the opposite. It has to be fixed on the branch before that branch lands, or it lands with the defect.Checked in the untouched-issue triage. Blocked behind #744, which needs its conflicts resolved and a CI re-run (its red check is
test_1063_constrained_traction, whose timeout #814 addressed when it merged on 2026-10-06).
solver.gradient(misfit, fields=[f])returns the dual on a field the residual reads. A field that enters only through an essential datum,u|Γ = f(a prescribed velocity field on a wall, an observed temperature used as a boundary value), is not seen: the field route scans the residual, not the boundary data, so the dual comes back zero and nothing says so. The scalar-parameter case was #762; this is its field counterpart.The term is the same reaction,
[(Kᵀμ)_Γ + (∂J/∂u)_Γ]restricted to the constrained rows and mapped onto the field's own nodes (a projection when the spaces differ). Until it is built, a field found in an essential datum should refuse rather than return zero.Branch
feature/discrete-adjoint, PR #744.