New Angular Quadrature types added into OpenSn. #899

AndrewNJ03 · 2026-01-18T15:59:03Z

3 new angular quadrature types have been introduced, along with corresponding regression tests. 2D Lebedev, a modification of its 3D variant, as well as 2D and 3D GLC Triangular Sets.

AndrewNJ03 · 2026-01-18T20:58:38Z

@andrsd @wdhawkins @ragusa

Would y'all mind reviewing this code? This PR should also close out the last milestone on Issue #245.

ragusa

making good progress. several things need adjustments, maybe the biggest one is creating a dedicated triangular class. @wdhawkins can you concur or amend this statement?

ragusa · 2026-01-19T16:28:57Z

framework/math/quadratures/angular/lebedev_quadrature.h

+class LebedevQuadrature2DXY : public AngularQuadrature
+{
+public:
+  /**


@andrsd @wdhawkins what are our standards for doxygen tags/doco?

There are no detailed instructions for all doxygen comments, but roughly this.

The use of brief is not enforced, but should be since JAVADOC_AUTOBRIEF = NO in the Doxyfile.

I changed to JAVADOC_AUTOBRIEF = YES in PR 902. \brief and \details usage should be removed in the future.

framework/math/quadratures/angular/triangular_quadrature.cc

python/lib/aquad.cc

test/python/framework/math/quadrature/gold/triangular_quad_2d_xy.py.gold

test/python/framework/math/quadrature/gold/triangular_quad_3d_xyz.py.gold

framework/math/quadratures/angular/triangular_quadrature.h

framework/math/quadratures/angular/triangular_quadrature.cc

ragusa

thanks for using tquad and not pquad in examples :-)

python/lib/aquad.cc

quocdang1998 · 2026-01-21T17:29:30Z

framework/math/quadratures/angular/triangular_quadrature.cc

+  // Calculate number of azimuthal angles per polar level
+  // At equator: Nazimuthal angles
+  // Each level away from equator: 4 less azimuthal angles (1 less per octant)
+  // This maintains octant symmetry in the triangular pattern


Are these kind of comments allowed? @ragusa

is the issue // versus /// or the text itself (too basic)? maybe @wdhawkins can chime in

// is allowed for in-code comments. Even if they were /// they would be ignored by doxygen.

quocdang1998 · 2026-01-21T17:31:06Z

framework/math/quadratures/angular/lebedev_quadrature.h

+class LebedevQuadrature2DXY : public AngularQuadrature
+{
+public:
+  /**


I changed to JAVADOC_AUTOBRIEF = YES in PR 902. \brief and \details usage should be removed in the future.

framework/math/quadratures/angular/lebedev_quadrature.h

quocdang1998 · 2026-01-21T17:35:07Z

framework/math/quadratures/angular/angular_quadrature.h

+   * Return a reference to the precomputed discrete-to-moment operator.
+   *
+   * The operator is accessed as [m][d], where m is the moment index and d is the direction index.


Is this a setter/getter? @andrsd @wdhawkins

If yes, we don't need any documentation here.

Well, the function name is GetDiscreteToMomentOperator. In code, we access it via [d][m], so the comment has it wrong :-) Plus, telling the users how to access the 2D array is useful, so I would fix the comment, at minimum.

I would like to bring this to @wdhawkins 's attention. Andrew's implementation followed the math derivation at first, until I remember that we likely have M and D indexed similarly, like [d][m] whereas D should be [m][d], so Andrew did a transpose operation to abide by how it is implemented in the code. But once D and M are exposed to the user, the user will also be confused about [d][m] versus [m][d], unless we return the object to them in python with indexing that abides with the math notation.

Yeah, I missed these related comment when I went back through and transposed the D2M matrix back to how OpenSn stores it. I can fix the comment to read [d][m].

Well, the function name is GetDiscreteToMomentOperator. In code, we access it via [d][m], so the comment has it wrong :-) Plus, telling the users how to access the 2D array is useful, so I would fix the comment, at minimum.

Sorry, looking back over this, this fix is specific to the next section regarding the D2M operator right? The M2D operator in OpenSn is stored equivalently to its definition in literature, so this comment is correct since the GetMomentToDiscreteOperator is still accessed as [m][d] unless I'm misunderstanding how it is used in the sweeper. The issue is the * Return a reference to the precomputed discrete-to-moment operator. since this is getting the M2D operator, not the D2M operator.

Ignore the above, I somehow read the comments upside down :/

Fixed to be [d][m] for now. Leaving this thread open.

andrsd · 2026-01-21T18:29:36Z

framework/math/quadratures/angular/angular_quadrature.h

-   * direction index.
+   * Return a reference to the precomputed moment-to-discrete operator.
+   *
+   * The operator is accessed as [m][d], where m is the moment index and d is the direction index.


This could be fixed too. The indexing is really [d][m]

pinging @wdhawkins here too

Fixed to be [d][m] for now. I'm going to leave this thread open so that it isn't hidden.

3 new angular quadrature types have been introduced, along with corresponding regression tests. 2D Lebedev, a modification of its 3D variant, as well as 2D and 3D GLC Triangular Sets.

wdhawkins · 2026-01-23T00:19:42Z

Be sure to update doc/source/pyapi/index.rst with the new classes. Also, you'll need to update doc/source/tutorials/aquad/introduction.rst.

andrsd · 2026-01-23T15:50:00Z

framework/math/quadratures/angular/triangular_quadrature.cc

+    Vector3 new_omega;
+    new_omega.x = sin(qpoint.theta) * cos(qpoint.phi);
+    new_omega.y = sin(qpoint.theta) * sin(qpoint.phi);
+    new_omega.z = cos(qpoint.theta);


In lebedev_quadrature.cc‎, you use the C++ versions of math functions, so you may want to do it here as well.

sin -> std::sin

cos -> std::cos

andrsd · 2026-01-23T15:50:23Z

framework/math/quadratures/angular/triangular_quadrature.cc

+  std::vector<double> polar;
+  polar.reserve(Npolar);
+  for (unsigned int j = 0; j < Npolar; ++j)
+    polar.emplace_back(M_PI - acos(gl_polar.qpoints[j][0]));


acos -> std::acos

andrsd · 2026-01-23T15:51:30Z

framework/math/quadratures/angular/triangular_quadrature.cc

+    Vector3 new_omega;
+    new_omega.x = sin(qpoint.theta) * cos(qpoint.phi);
+    new_omega.y = sin(qpoint.theta) * sin(qpoint.phi);
+    new_omega.z = cos(qpoint.theta);


Use std::sin, std::cos here as well

andrsd · 2026-01-23T15:57:12Z

python/lib/aquad.cc

+      {
+        static const std::vector<std::string> required_keys = {"n_polar", "scattering_order"};
+        static const std::vector<std::pair<std::string, py::object>> optional_keys = {{"verbose", py::bool_(false)}};
+        return construct_from_kwargs<GLCTriangularQuadrature3DXYZ, int, int, bool>(params, required_keys, optional_keys);


The types after the class name should match the types of the class constructor arguments. Since we changed that to unsigned int, these need to be changed as well.

Also, update on line 295, 581.

AndrewNJ03 force-pushed the new_quadrature_types branch 5 times, most recently from f967aa2 to a2f3d2a Compare January 18, 2026 20:46

ragusa requested review from andrsd, quocdang1998, ragusa and wdhawkins and removed request for andrsd and quocdang1998 January 19, 2026 16:25

ragusa requested changes Jan 19, 2026

View reviewed changes

andrsd requested changes Jan 19, 2026

View reviewed changes

framework/math/quadratures/angular/triangular_quadrature.h Outdated Show resolved Hide resolved

framework/math/quadratures/angular/triangular_quadrature.h Outdated Show resolved Hide resolved

framework/math/quadratures/angular/triangular_quadrature.cc Outdated Show resolved Hide resolved

AndrewNJ03 force-pushed the new_quadrature_types branch from 6d8b695 to f933ceb Compare January 20, 2026 20:00

AndrewNJ03 requested review from andrsd and ragusa January 20, 2026 20:30

ragusa approved these changes Jan 20, 2026

View reviewed changes

AndrewNJ03 force-pushed the new_quadrature_types branch from b15ff5c to 463c446 Compare January 21, 2026 15:40

quocdang1998 requested changes Jan 21, 2026

View reviewed changes

andrsd reviewed Jan 21, 2026

View reviewed changes

AndrewNJ03 force-pushed the new_quadrature_types branch from 463c446 to f58cebf Compare January 21, 2026 19:07

New Angular Quadrature types added into OpenSn.

2dc948f

3 new angular quadrature types have been introduced, along with corresponding regression tests. 2D Lebedev, a modification of its 3D variant, as well as 2D and 3D GLC Triangular Sets.

AndrewNJ03 force-pushed the new_quadrature_types branch from f58cebf to 2dc948f Compare January 21, 2026 19:14

AndrewNJ03 requested review from andrsd and quocdang1998 January 21, 2026 19:16

andrsd requested changes Jan 23, 2026

View reviewed changes

andrsd mentioned this pull request Jan 23, 2026

Updating python binders for quadratures #907

Merged

New Angular Quadrature types added into OpenSn. #899

Are you sure you want to change the base?

New Angular Quadrature types added into OpenSn. #899

Uh oh!

Conversation

AndrewNJ03 commented Jan 18, 2026

Uh oh!

AndrewNJ03 commented Jan 18, 2026

Uh oh!

ragusa left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

andrsd Jan 19, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

ragusa left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

AndrewNJ03 Jan 21, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

wdhawkins commented Jan 23, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

andrsd Jan 19, 2026 •

edited

Loading

AndrewNJ03 Jan 21, 2026 •

edited

Loading