Mesh intersection routines #209

abhiyanpaudel · 2025-10-22T19:04:50Z

This pull request adds routines for performing Galerkin projection using mesh intersections. The implementation supports linear, 2D triangular elements and enables conservative field transfer between non-matching meshes.

…on_svd_solver

Copilot

Pull Request Overview

This pull request adds conservative field projection capabilities via mesh intersection to the PCMS interpolator library. The implementation enables Galerkin-based projection of scalar fields between non-conforming 2D triangular meshes using supermesh techniques.

Key changes include:

Implementation of mesh intersection algorithms with adjacency-based breadth-first search
Load vector and mass matrix integrators for conservative field transfer
PETSc-based linear system solver for projection
New optional dependencies: MeshFields and PETSc
Enhanced RBF kernel documentation and new kernel types (multiquadric, inverse multiquadric, thin plate spline, cubic)

Reviewed Changes

Copilot reviewed 16 out of 16 changed files in this pull request and generated 11 comments.

Show a summary per file

File	Description
test/test_load_vector.cpp	Test suite for load vector computation and mesh intersection
test/CMakeLists.txt	Build configuration for new mesh intersection tests with conditional dependencies
src/pcms/interpolator/mesh_intersection/*.hpp	Core mesh intersection, conservative projection, and integration routines
src/pcms/interpolator/mls_interpolation.hpp	Expanded RBF kernel documentation and new kernel definitions
src/pcms/interpolator/mls_interpolation.cpp	Implementation of new RBF kernels (multiquadric, inverse multiquadric, thin plate spline, cubic)
src/pcms/interpolator/mls_interpolation_impl.hpp	Code formatting updates and scratch size fix
src/pcms/interpolator/pcms_interpolator_view_utils.hpp	Documentation for square root helper function
src/pcms/interpolator/adj_search.hpp	Fix for handling points outside mesh boundaries
CMakeLists.txt	Build system support for optional MeshFields and PETSc dependencies

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2025-10-31T19:46:56Z

test/test_load_vector.cpp

+    0.0, 1.0  // v3
+  });
+
+  // Target Mesh with two traingles


Corrected spelling of 'traingles' to 'triangles'.

Copilot · 2025-10-31T19:46:57Z

test/test_load_vector.cpp

+  Omega_h::build_from_elems_and_coords(&target_mesh, OMEGA_H_SIMPLEX, 2,
+                                       ev2v_source, coords);
+
+  // Source Mesh with two traingles


Corrected spelling of 'traingles' to 'triangles'.

Suggested change

// Source Mesh with two traingles

// Source Mesh with two triangles

Copilot · 2025-10-31T19:46:57Z

test/test_load_vector.cpp

+    points_h(0, 0) = 0.0;
+    points_h(0, 1) = 0.0;
+    points_h(1, 0) = 1.0;
+    points_h(0, 1) = 0.0;


Index mismatch: line 59 sets points_h(0, 1) but should set points_h(1, 1) to define the second point's y-coordinate. This overwrites the first point's y-coordinate and leaves the second point's y-coordinate uninitialized.

Suggested change

points_h(0, 1) = 0.0;

points_h(1, 1) = 0.0;

Copilot · 2025-10-31T19:46:57Z

test/test_load_vector.cpp

+          "source elements")
+  {
+    // the source elements are triangle1 (0,0), (1,0) & (1,1) and triangle2
+    // (0,0), (1,1) & (0,1) the target elements are traingle1 (0,0), (1,0) &


Corrected spelling of 'traingle1' to 'triangle1'.

Suggested change

// (0,0), (1,1) & (0,1) the target elements are traingle1 (0,0), (1,0) &

// (0,0), (1,1) & (0,1) the target elements are triangle1 (0,0), (1,0) &

Copilot · 2025-10-31T19:46:58Z

test/CMakeLists.txt

  endif()
+
+  if (PCMS_ENABLE_MESHFIELDS)
+	list(APPEND PCMS_UNIT_TEST_SOURCES test_load_vector.cpp)


Inconsistent indentation: use spaces instead of tabs for consistency with the surrounding code style. Lines 358, 364, 368-369 use tabs while the rest of the file uses spaces.

Copilot · 2025-10-31T19:46:58Z

src/pcms/interpolator/mesh_intersection/conservative_projection_solver.hpp

+ */
+
+#ifndef PCMS_INTERPOLATOR_GALERKIN_PROJECTION_SOLVER_HPP
+#ifndef PCMS_INTERPOLATOR_GALERKIN_PROJECTION_SOLVER_HPP


Duplicate header guard declaration. The second #ifndef on line 19 should be #define PCMS_INTERPOLATOR_GALERKIN_PROJECTION_SOLVER_HPP.

Suggested change

#ifndef PCMS_INTERPOLATOR_GALERKIN_PROJECTION_SOLVER_HPP

#define PCMS_INTERPOLATOR_GALERKIN_PROJECTION_SOLVER_HPP

Copilot · 2025-10-31T19:46:59Z

src/pcms/interpolator/mesh_intersection/calculate_load_vector.hpp

+#include <petscvec_kokkos.hpp>
+#include <sstream> //ostringstream
+
+#include <pcms/interpolator/mesh_interscetion/load_vector_integrator.hpp>


Corrected spelling in include path: 'mesh_interscetion' should be 'mesh_intersection'.

Suggested change

#include <pcms/interpolator/mesh_interscetion/load_vector_integrator.hpp>

#include <pcms/interpolator/mesh_intersection/load_vector_integrator.hpp>

Copilot · 2025-10-31T19:46:59Z

src/pcms/interpolator/mls_interpolation_impl.hpp

  // calculates the interpolated values
  Kokkos::parallel_for(
-    "MLS coefficients", tp.set_scratch_size(1, Kokkos::PerTeam(scratch_size)),
+    "MLS coefficients", tp.set_scratch_size(1, Kokkos::PerTeam(shared_size)),


Changed from scratch_size to shared_size without verification that shared_size is the correct value. Line 474 asserts scratch_size > shared_size, so using shared_size may lead to insufficient scratch memory allocation. This should use scratch_size or the assertion logic should be updated.

Suggested change

"MLS coefficients", tp.set_scratch_size(1, Kokkos::PerTeam(shared_size)),

"MLS coefficients", tp.set_scratch_size(1, Kokkos::PerTeam(scratch_size)),

Copilot · 2025-10-31T19:46:59Z

CMakeLists.txt

+	find_library(PETSC_LIBRARY petsc HINTS ${PETSC_LIB_DIR})
+
+	if (NOT PETSC_LIBRARY)
+		message(FATAL_ERROR "Couldnot find libpetsc in ${PETSC_LIB_DIR}")


Corrected spacing: 'Couldnot' should be 'Could not'.

Suggested change

message(FATAL_ERROR "Couldnot find libpetsc in ${PETSC_LIB_DIR}")

message(FATAL_ERROR "Could not find libpetsc in ${PETSC_LIB_DIR}")

Copilot · 2025-10-31T19:47:00Z

src/pcms/interpolator/mls_interpolation.cpp

+      phi = 0.0;
+    } else {
+      phi =
+        a * a * ratio * ratio * Kokkos::log(a * ratio); // a shouldnot be zero


Corrected spacing in comment: 'shouldnot' should be 'should not'.

Suggested change

a * a * ratio * ratio * Kokkos::log(a * ratio); // a shouldnot be zero

a * a * ratio * ratio * Kokkos::log(a * ratio); // a should not be zero

jacobmerson · 2025-11-03T05:03:00Z

@abhiyanpaudel can you share the petsc configuration you are using so we can add that to CI?

abhiyanpaudel · 2025-11-03T18:22:55Z

./configure \
    PETSC_ARCH=linux-gnu-gpu-kokkos \
    --with-kokkos-dir="$BUILD_DIR/${DEVICE_ARCH}/kokkos/install/" \
    --with-kokkos-kernels-dir="$BUILD_DIR/${DEVICE_ARCH}/kokkoskernels/install/" \
    --with-cuda=1 \
    --with-openblas-dir="${OPENBLAS_RHEL9_ROOT}"

jacobmerson · 2025-11-03T05:01:16Z

src/pcms/interpolator/mesh_intersection/calculate_load_vector.hpp

+ * meshes.
+ *
+ * @created by Abhiyan Paudel
+ * @date August 2025


I would take out @created and @date from the header. That info is captured in the git history.

Sichao25

Overall looks good to me. Added a few minor questions and suggestions for consideration. A test for the main behavior (e.g., Galerkin projection) would be helpful.

Sichao25 · 2025-11-06T00:00:06Z

src/pcms/interpolator/mesh_intersection/calculate_load_vector.hpp

+#include <pcms/interpolator/mesh_interscetion/load_vector_integrator.hpp>
+#include <MeshField.hpp>
+
+#include <petscmat.h>


Correct me if I'm wrong, but from the CMakeLists it looks like the mesh_intersection source and headers will always be compiled, but the petsc library can be disabled. I'm wondering do these headers still exist if we disable the petsc library?

Sichao25 · 2025-11-06T00:02:28Z

src/pcms/interpolator/mesh_intersection/calculate_mass_matrix.hpp

+
+  auto elmMassMatrix = buildMassMatrix(mesh, coordFe);
+
+  auto host_elmMassMatrix = Kokkos::create_mirror_view(elmMassMatrix);


Do we need this view now or in the future?

Sichao25 · 2025-11-06T00:06:19Z

src/pcms/interpolator/mesh_intersection/conservative_projection_solver.hpp

+  CHKERRABORT(PETSC_COMM_WORLD, ierr);
+
+  KSP ksp;
+  ierr = KSPCreate(PETSC_COMM_SELF, &ksp);


I'm curious why we need PETSC_COMM_SELF here. Based on the documentation, PETSC_COMM_WORLD and MPI_COMM_WORLD are identical by default unless you want a subset of MPI_COMM_WORLD.

Sichao25 · 2025-11-07T00:29:46Z

src/pcms/interpolator/mesh_intersection/CMakeLists.txt

+	target_link_libraries(pcms_mesh_intersection PUBLIC ${PCMS_PETSC_LIBRARIES})
+endif()
+
+target_compile_definitions(pcms_mesh_intersection PUBLIC R3D_USE_CUDA)


Is there any behavior that depends specifically on the CUDA-only backend in Omega_h? If so, it may be removed in the future.

jacobmerson · 2025-11-12T05:13:42Z

@abhiyanpaudel #217 should be merged soon. This brings in petsc to the CI pipeline.

abhiyan123 added 30 commits August 17, 2025 21:16

add new rbf functions

96012d6

Merge branch 'develop' of github.com:SCOREC/pcms into mls_interpolati…

1f7cc8c

…on_svd_solver

minor changes

3a14936

Merge branch 'develop' into mls_interpolation_svd_solver

088e322

add doxygen comment

9e92528

add mesh intersection files

219c35e

add doxygen comments

e3e7fc1

add doxygen comments

53b6498

add doxygen comments

3dea928

add doxygen comments

6b5c25e

minor fixes

2ef3545

update doxygen comments

bdf44eb

update comments

93a08d5

update CMakeLists

aabe402

update CMakeLists

a3330ef

update topmost CMakeLists

7ab665b

update CMakeLists

7a7ea7b

add pcms namespace

7cce586

add load vector test

9984c7b

add namespace

e67922e

remove main

a9a2c57

correct the name of petsc variables

1da8be9

correct the petsc variables name

7c0fa98

add PCMS_ENABLE_PETSC and PCMS_ENABLE_MESHFIELDS macros

e72a6aa

fix memory space issue

f90f795

fix petsc variable name

cba1703

fix header name

7bfe489

fix header name

dd105ff

add more rbfs and their use cases

92dd379

fix constructor

e0cf357

remove semicolon

ba5d9ab

jacobmerson requested review from Sichao25, Copilot and jacobmerson and removed request for Copilot October 31, 2025 19:42

Copilot AI reviewed Oct 31, 2025

View reviewed changes

Merge branch 'develop' into mesh_intersection

f48c2f9

jacobmerson mentioned this pull request Nov 3, 2025

Add PETSc with Kokkos support to CI/CD #213

Closed

jacobmerson reviewed Nov 3, 2025

View reviewed changes

Sichao25 reviewed Nov 6, 2025

View reviewed changes

Sichao25 reviewed Nov 7, 2025

View reviewed changes

jacobmerson mentioned this pull request Nov 7, 2025

dropping direct cuda support SCOREC/omega_h#175

Closed

jacobmerson mentioned this pull request Dec 9, 2025

Cmake issues #219

Merged

	// Source Mesh with two traingles
	// Source Mesh with two triangles

	// (0,0), (1,1) & (0,1) the target elements are traingle1 (0,0), (1,0) &
	// (0,0), (1,1) & (0,1) the target elements are triangle1 (0,0), (1,0) &

	#ifndef PCMS_INTERPOLATOR_GALERKIN_PROJECTION_SOLVER_HPP
	#define PCMS_INTERPOLATOR_GALERKIN_PROJECTION_SOLVER_HPP

	#include <pcms/interpolator/mesh_interscetion/load_vector_integrator.hpp>
	#include <pcms/interpolator/mesh_intersection/load_vector_integrator.hpp>

	"MLS coefficients", tp.set_scratch_size(1, Kokkos::PerTeam(shared_size)),
	"MLS coefficients", tp.set_scratch_size(1, Kokkos::PerTeam(scratch_size)),

	message(FATAL_ERROR "Couldnot find libpetsc in ${PETSC_LIB_DIR}")
	message(FATAL_ERROR "Could not find libpetsc in ${PETSC_LIB_DIR}")

	a * a * ratio * ratio * Kokkos::log(a * ratio); // a shouldnot be zero
	a * a * ratio * ratio * Kokkos::log(a * ratio); // a should not be zero


		auto elmMassMatrix = buildMassMatrix(mesh, coordFe);

		auto host_elmMassMatrix = Kokkos::create_mirror_view(elmMassMatrix);

Mesh intersection routines #209

Are you sure you want to change the base?

Mesh intersection routines #209

Uh oh!

Conversation

abhiyanpaudel commented Oct 22, 2025

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull Request Overview

Reviewed Changes

Uh oh!

Copilot AI Oct 31, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Oct 31, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Oct 31, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Oct 31, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Oct 31, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Oct 31, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Oct 31, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Oct 31, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Oct 31, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Oct 31, 2025

Choose a reason for hiding this comment

Uh oh!

jacobmerson commented Nov 3, 2025

Uh oh!

abhiyanpaudel commented Nov 3, 2025

Uh oh!

jacobmerson Nov 3, 2025

Choose a reason for hiding this comment

Uh oh!

Sichao25 left a comment

Choose a reason for hiding this comment

Uh oh!

Sichao25 Nov 6, 2025

Choose a reason for hiding this comment

Uh oh!

Sichao25 Nov 6, 2025

Choose a reason for hiding this comment

Uh oh!

Sichao25 Nov 6, 2025

Choose a reason for hiding this comment

Uh oh!

Sichao25 Nov 7, 2025

Choose a reason for hiding this comment

Uh oh!

jacobmerson commented Nov 12, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants