Repository navigation
experimental/adrh: Richards' equation (van Genuchten-Mualem, layered materials) against proteus's richards model (stacked on #115) - #117
Merged
Conversation
… model; segments (1D); bindElement hook richards.h (van Genuchten-Mualem, ported from pskRelations.h; several materials through bindElement), 1D segment spaces, weak Dirichlet with the interior advective flux (Richards_IIPG_exterior), and richards_regression.py: test/richards/test_1 (Celia) matches proteus's saved Galerkin profiles to 2.5e-13 with proteus's Newton stopping rule, on the host and Kokkos modules. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
…EADME section for the Richards case The engine's two-material steady column converges to the shooting solution of the constant-flux ODE at exactly the P1 rate (2.00, 11 to 321 nodes), host and Kokkos modules identical. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
… inline fixed, CSV path from the environment richards.h declared vgmWetting 'ADRH_FN inline'; ADRH_FN already includes inline, which gcc and nvcc reject (clang accepted it). RichardsMaterials with no table assigns materials by element number modulo the count, so benchmarks exercise bindElement on a device without a data pointer. The Celia column matches proteus's saved profiles to 3.3e-13 on the GPU module. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
…d as measured Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
cekees
added a commit
that referenced
this pull request
Oct 27, 2026
experimental/adrh: algebraic stabilization, lumped mass and discrete upwinding (stacked on #117)
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Stacked on #115 (base
adrh-kokkos-faces, itself on #114 and #113); merge in that order, then retarget to main.What
Richards' equation in pressure-head form, the first nonlinear, strongly heterogeneous problem for the ADRH engine.
richards.h: the Mualem-van Genuchten law ported fromproteus/pskRelations.hand the coefficients ofproteus/richards/Richards.h;RichardsMaterials(up to 8 materials, a material number per element).bindElement(eN), in the element and face kernels (engine.h,boundary.h) and the Kokkos launches: a functor that defines it is asked for the functor to use on that element; others are untouched.spaces.h), andFaceBC::advInteriorfor the weak Dirichlet variant of proteus'sRichards_IIPG_exterior.richards_regression.pyandrichards_layers.py; module entriesrichards_vgmandrichards_vgm_faceson the host and Kokkos modules.Checks
test/richards/test_1) against proteus's saved Galerkin profiles: 2.5e-13 at 12 h and 24 h with the weak Dirichlet treatment and proteus's Newton stopping rule (2-norm < 1e-8), host and Kokkos modules identical. With the Dirichlet rows replaced instead the fronts differ by 0.06 to 0.08, which is how the deck's weak boundary condition was found. Converged tighter, the engine differs from the saved profiles by 8.7e-8 and 4.2e-6: proteus's Newton tolerance, not the discretization.Not done
Stabilization: proteus's HYDRUS and Szymkiewicz decks run with entropy viscosity or FCT because plain Galerkin does not run them, so they are not compared yet. Faces through
bindElementare implemented but not exercised by a test.CI note: every branch currently fails the same 31 archive-format tests until the proteus writer for ymf archive format 2 (#110) merges; that is unrelated to this PR.
🤖 Generated with Claude Code