CSG 2D Lattice Support #31993

kkiesling · 2025-11-26T16:43:06Z

Reason

CSG Lattice support is needed for implementing the generateCSG method for rest of the RGMB mesh generators and other patterned MGs.

Design

Here are a few key design choices not previously discussed + some related questions (in bold):

Hexagonal and Cartesian lattices are directly supported by the CSGBase class, but any other type of custom 2D lattices could be defined by deriving from the CSGLattice class and added to the CSGBase instance (analogous to how CSGSurfaces are handled). The difference between the lattice and surface implementation for this is that I provide multiple methods directly in CSGBase for creating either a CSGHexagonalLattice or CSGCartesianLattice, rather than requiring that those types be created separately and then added to the base.
Per discussion on slack, I decided to implement the lattice "outer" region (ie the space between lattice elements that is undefined by a fill universe) to be either a CSG material (string ID) or a Universe.
For all the lattice type-specific information, I created a method called getAttributes which provides all the "extra" info that isn't inherently known from the universe map (or that is convenient to have easily on hand), ie pitch or nrings. Because the data type of these attributes can be anything technically, I set it up to use the std::any type. However, if accessing this data, it requires specific casting to the expected data type std::any_cast. In particular, the JSON dictionary doesn't like this data type, so I created the extra helper method in the CSGUtils file to convert these data types. This data type conversion isn't really CSG-specific though, so is there a different place in MOOSE that this method should live? Or, if this isn't the desired approach to handling the different types of data, the getAttributes method could just be setup to create a JSON object instead, and this helper method would not be needed. However, depending on the decision on 6 below, we may want to leave the getAttributes as it is now so that this sort of information can be accessed outside of JSON file creation.
For hexagonal lattices, I implemented some methods to help convert between the (row, column) form and (ring, position) form. Per discussion on Slack, it was recommended that these methods be located in the HexagonalLatticeUtils file. However, that file appears to be in the Reactor module, and not the framework. Does it make sense for the framework to rely on a utils file in a module like that? I didn't move the methods due to this, but let me know if you think that is where they should live. Alternatively, there is the CSGUtils file now where I could move these to.
Because of the existence of the aforementioned extra methods in the CSGHexagonalLattice class, in order for those methods to be available after a lattice object is created via CSGBase, those createXLattice methods in CSGBase have to return the specific lattice type, rather than simply CSGLattice. I set this up to use dynamic casting to handle this. But this also required updating the getLatticeByName method to use a template so that the correct type is returned and these methods are accessible. The templated method is setup such that it will default to CSGLattice type if no type is provided, thus releasing the burden of having to do this a bit. Shikhar and I went back and forth on this a bit, and decided to wait for input on the review about whether we should use this template and specify the specific lattice return types, or if we should use the CSGLattice only. If we move the extra CSGHexagonalLattice methods to a utils file, then this becomes less necessary and we can set the methods up to just return CSGLattice in all cases. Methods like getNRows or getPitch on these two lattice classes also wouldn't be available, but those values can also be retrieved via getAttributes (though this requires use of any_cast to correctly gather the value, as noted above). So the question is: does it make sense to leave this as it is with the dynamic casting, or should we use CSGLattice only?

Impact

Adds support for CSGLattices to the CSGBase class so that the generateCSG methods can be completed for various MGs relying on lattice forms.

…attice accordingly

… virtual function

…(needs more tests)

…ated - tests need to be updated

…il populated - tests need to be updated" This reverts commit 3f6e101adea23476b7e093ffc42914a1f59122ae.

… doesn't change anything that needs to be managed by base

…e names and conversion of any data type

Note: updates and confirmation/review for correctness are required; this assumes a (x,y) type of grid indexing scheme, but that may need to be (ring, index) instead. Or conversion methods between the two are needed. Tests are still needed too.

…rows and rings

…ds instead of getDimensions directly

…ons already set

…t with a map instead

…es obsolete; only set pitch

…verse fills

moosebuild · 2025-11-26T17:01:03Z

large_media

@@ -1 +1 @@
-Subproject commit 7a7e52d8ffbb7223b46768ff88885c8c52f71855
+Subproject commit ed12daf37cc1919b018483d879776ef002a94595


Caution! This contains a submodule update

moosebuild · 2025-11-26T18:00:22Z

large_media

@@ -1 +1 @@
-Subproject commit 7a7e52d8ffbb7223b46768ff88885c8c52f71855
+Subproject commit ed12daf37cc1919b018483d879776ef002a94595


Caution! This contains a submodule update

moosebuild · 2025-11-26T18:01:26Z

Job Precheck, step Clang format on d0f6835 wanted to post the following:

Your code requires style changes.

A patch was auto generated and copied here
You can directly apply the patch by running, in the top level of your repository:

curl -s https://mooseframework.inl.gov/docs/PRs/31993/clang_format/style.patch | git apply -v

Alternatively, with your repository up to date and in the top level of your repository:

git clang-format b85414d92b7659931cc90106a67659da1e979214

moosebuild · 2025-11-26T18:01:27Z

large_media

@@ -1 +1 @@
-Subproject commit 7a7e52d8ffbb7223b46768ff88885c8c52f71855
+Subproject commit ed12daf37cc1919b018483d879776ef002a94595


Caution! This contains a submodule update

moosebuild · 2025-11-26T19:16:04Z

large_media

@@ -1 +1 @@
-Subproject commit 7a7e52d8ffbb7223b46768ff88885c8c52f71855
+Subproject commit ed12daf37cc1919b018483d879776ef002a94595


Caution! This contains a submodule update

moosebuild · 2025-11-26T20:21:24Z

large_media

@@ -1 +1 @@
-Subproject commit 7a7e52d8ffbb7223b46768ff88885c8c52f71855
+Subproject commit ed12daf37cc1919b018483d879776ef002a94595


Caution! This contains a submodule update

moosebuild · 2025-11-26T21:03:20Z

large_media

@@ -1 +1 @@
-Subproject commit 7a7e52d8ffbb7223b46768ff88885c8c52f71855
+Subproject commit ed12daf37cc1919b018483d879776ef002a94595


Caution! This contains a submodule update

kkiesling · 2025-11-26T21:34:25Z

This looks like it will pass CI at this point, so it should be ready for review @GiudGiud

kkiesling · 2025-11-26T21:34:41Z

@eshemon FYI

moosebuild · 2025-11-26T22:08:19Z

Job Test, step Results summary on f0a2d2f wanted to post the following:

Framework test summary

Compared against b85414d in job civet.inl.gov/job/3409862.

Added tests

Test	Time (s)
`csg.geometry/csg_only_nested_lattices`	0.83
`csg.geometry/csg_only_hex_lattice`	0.80
`csg.geometry/csg_only_cart_lattice`	0.80

Modules test summary

Compared against b85414d in job civet.inl.gov/job/3409862.

No change

moosebuild · 2025-11-26T23:08:58Z

Job Documentation, step Docs: sync website on f0a2d2f wanted to post the following:

View the site here

This comment will be updated on new commits.

moosebuild · 2025-11-26T23:58:41Z

Job Coverage, step Generate coverage on f0a2d2f wanted to post the following:

Framework coverage

	b85414	#31993 f0a2d2
	Total	Total	+/-	New
Rate	85.98%	85.99%	+0.02%	100.00%
Hits	125642	126230	+588	661
Misses	20494	20560	+66	0

Diff coverage report

Full coverage report

Modules coverage

Coverage did not change

Full coverage reports

Reports

This comment will be updated on new commits.

GiudGiud · 2025-11-26T23:57:24Z

framework/doc/content/source/csg/CSGBase.md

+
+A `CSGLattice` is defined as a patterned arrangement of [`CSGUniverse`](#universes) objects and an "outer" to fill the space around lattice elements.
+The `CSGBase` class provides support for two types of 2D lattice types: Cartesian and regular hexagonal.
+It is assumed that these two types are in the $x-y$ plane (having a $+z$ norm).


Suggested change

It is assumed that these two types are in the $x-y$ plane (having a $+z$ norm).

It is assumed that these two types are in the $x-y$ plane (having a $+z$ normal).

GiudGiud · 2025-12-01T17:35:50Z

framework/doc/content/source/csg/CSGBase.md

+
+!alert! note title=2D vs. 3D Lattices
+
+The `CSGBase` class supports only the creation of 2D lattices. A "3D" lattice can be created by filling multiple 3D cells with 2D lattices and arranging them in layers to form the desired 3D structure.


Suggested change

The `CSGBase` class supports only the creation of 2D lattices. A "3D" lattice can be created by filling multiple 3D cells with 2D lattices and arranging them in layers to form the desired 3D structure.

The `CSGBase` class supports only the creation of 2D lattices. A "3D" lattice can be created by filling multiple 3D cells with 2D lattices and arranging them in layers to form the desired 3D structure, or by modifying the cells into a 2D lattice to using extrusion in the third dimension.

reactor module always does extrusion at the last step. We ll need to be able to turn these 2D cells into 3D cells then.

GiudGiud · 2025-12-01T17:43:20Z

framework/include/utils/CSGUtils.h

+ * @param data std::any object to convert
+ * @return nlohmann::json representation of the data
+ */
+nlohmann::json anyToJson(const std::any & data);


let's make a new file "JSONUtils.h" we wont find this new cool routine otherwise

GiudGiud · 2025-12-01T19:39:00Z

framework/doc/content/source/csg/CSGBase.md

+
+The `CSGLattice` objects can be accessed or updated with the following methods from `CSGBase`:
+
+- `setLatticeUniverses`: sets the vector of vectors of `CSGUniverse` objects as the lattice layout.


this is kind of assuming a regular layout. The FlexiblePattern (widely used) would probably not work here

maybe it would just work with a single vector passed as a 2d vector

GiudGiud · 2025-12-01T19:40:42Z

framework/doc/content/source/csg/CSGBase.md

+- `resetLatticeOuter`: resets the outer fill to void for the lattice.
+- `renameLattice`: change the name of the `CSGLattice` object.
+- `getAllLattices`: retrieve a list of const references to each `CSGLattice` object in the `CSGBase` instance.
+- `getLatticeByName`: retrieve a const reference to the lattice object of the specified name.


as mentioned, this is a little different that the usual source documentation, as it overlaps with the doxygen. I generally would not recommend listing routines, though I wont ask that you change this either

GiudGiud · 2025-12-01T22:48:10Z

framework/src/csg/CSGBase.C

+  auto name = lattice.getName();
+  if (!checkLatticeInBase(lattice))
+    mooseError("Cannot set universe at index for lattice " + name +
+               ". Lattice is different from the lattice of the same name in the "


how do we know it has the same name

GiudGiud · 2025-12-01T22:52:12Z

test/include/csg/NestedLatticeCellUniverseMeshGenerator.h

  const Real _inner_rad;
  /// radius of outer sphere surface
  const Real _outer_rad;
+  /// radius of lattice sphere surface


lattices dont have regions?

GiudGiud · 2025-12-01T23:18:13Z

test/src/csg/ExampleCSGInfiniteSquareMeshGenerator.C

+  // determine fill: either from input fill mesh generator or default material
+  if (_has_fill)
+  {
+    // join the fill CSGBase into the current CSGBase & use the lattice as the fille


Suggested change

// join the fill CSGBase into the current CSGBase & use the lattice as the fille

// join the fill CSGBase into the current CSGBase & use the lattice as the fill

GiudGiud · 2025-12-01T23:30:48Z

test/src/csg/TestCSGLatticeMeshGenerator.C

+  params.addRequiredParam<std::vector<std::vector<unsigned int>>>(
+      "pattern",
+      "A double-indexed array starting with the upper-left corner where the index"
+      "represents the layout of inputs in the lattice.");


Suggested change

"represents the layout of inputs in the lattice.");

"represents the index of the mesh/CSG generator in the 'inputs' vector");

GiudGiud · 2025-12-01T23:34:20Z

framework/src/csg/CSGLatticeList.C

+    const Real pitch,
+    std::vector<std::vector<std::reference_wrapper<const CSGUniverse>>> universes)
+{
+  return addLattice(std::make_unique<CSGCartesianLattice>(name, pitch, universes));


you could just do

template type T addLattice { return addLattice(std::make_unique<T>(...); }

kkiesling added 30 commits November 25, 2025 10:24

abstract CSG lattice class

0fd036e

cartesian lattice csg class

6c93cb1

finish CSGCartesianLattice preliminary implementation and revise CSGL…

43ce9bc

…attice accordingly

cannot set universes in abstract class constructor because it calls a…

43f77cd

… virtual function

cannot use unsigned ints

5acfb87

unit tests for CSGLattice and CSGCartesianLattice

2b0b6e1

initial lattice list implementation

1710fd7

add method for updating a set dimension of the lattice

86bef1b

methods for comparing lattice equality

73673e7

method for getting list of unique universes in a lattice

d9395c7

exposing CSGLattice and CSGCartesianLattice methods to CSGBase class …

96ee340

…(needs more tests)

test and updates for updateDimensions for cart lattices

72fbeef

change structure of universe lattices to all null entries until popul…

0d98f98

…ated - tests need to be updated

Revert "change structure of universe lattices to all null entries unt…

3c198c2

…il populated - tests need to be updated" This reverts commit 3f6e101adea23476b7e093ffc42914a1f59122ae.

tests for set/add universes from base

cbcdbf0

tests for the getLattice methods

6165bcd

remove unnecessary wrapper method and make updateDims method public -…

8548957

… doesn't change anything that needs to be managed by base

join lattice_lists when joining - needs updated tests

0fcf35b

create lattice-filled cells

308cb92

write out lattice info- requires creating list of of lists of univers…

0f1a1d9

…e names and conversion of any data type

add lattice to viable universe linking options in check

b03854a

define hex lattice by rows but give functionality to convert between …

700744f

…rows and rings

add methods for getting values directly for ease

f5ff3cf

add tests for hex lattice construction, update tests to use get metho…

c5dd897

…ds instead of getDimensions directly

change the validity of how full universe map can be set after dimensi…

ea778a9

…ons already set

remove arbitrary setting of dimensions for an empty lattice- to be se…

e5d744a

…t with a map instead

hasValidDimensions is obsolete if just gathering from valid universe map

a50d94d

lattice shape is altered in different ways, so updateDimensions becom…

0afe344

…es obsolete; only set pitch

remove support for converting between ring and index- algo not working

d466fb1

shikhar413 and others added 6 commits November 25, 2025 10:24

Rename NestedCellUniverseMG to NestedLatticeCellUniverseMG

1b045bd

Remove extraneous curly braces in CSGBase

53fe486

Return fill names from lattice and universe names for lattice and uni…

c63d225

…verse fills

use dynamic_cast instead of static to be able to check at compile time

302da5f

update to include CSG lattice images for docs

e78841b

update docs to include links to images - refs idaholab#31889

5d306db

kkiesling requested a review from lindsayad as a code owner November 26, 2025 16:43

fix unicode characters in docstring

a3a2697

moosebuild reviewed Nov 26, 2025

View reviewed changes

clang format

b34fc06

kkiesling force-pushed the csg_lattices branch from d0f6835 to b34fc06 Compare November 26, 2025 19:08

moosebuild reviewed Nov 26, 2025

View reviewed changes

kkiesling added 3 commits November 26, 2025 13:32

fix comment syntax

869f8c9

needs to be int to be able to account for negative number input

7786b03

vector cannot contain const values

bebfb90

moosebuild reviewed Nov 26, 2025

View reviewed changes

kkiesling added 2 commits November 26, 2025 14:53

correct the index type to int for tests

24f61b9

update element->position for ring form for consistency

f0a2d2f

moosebuild reviewed Nov 26, 2025

View reviewed changes

lindsayad requested a review from GiudGiud November 26, 2025 21:20

GiudGiud self-assigned this Nov 26, 2025

GiudGiud reviewed Dec 1, 2025

View reviewed changes

		@@ -1 +1 @@
		Subproject commit 7a7e52d8ffbb7223b46768ff88885c8c52f71855
		Subproject commit ed12daf37cc1919b018483d879776ef002a94595

	It is assumed that these two types are in the $x-y$ plane (having a $+z$ norm).
	It is assumed that these two types are in the $x-y$ plane (having a $+z$ normal).


		!alert! note title=2D vs. 3D Lattices

		The `CSGBase` class supports only the creation of 2D lattices. A "3D" lattice can be created by filling multiple 3D cells with 2D lattices and arranging them in layers to form the desired 3D structure.


		The `CSGLattice` objects can be accessed or updated with the following methods from `CSGBase`:

		- `setLatticeUniverses`: sets the vector of vectors of `CSGUniverse` objects as the lattice layout.

	// join the fill CSGBase into the current CSGBase & use the lattice as the fille
	// join the fill CSGBase into the current CSGBase & use the lattice as the fill

	"represents the layout of inputs in the lattice.");
	"represents the index of the mesh/CSG generator in the 'inputs' vector");

CSG 2D Lattice Support #31993

Are you sure you want to change the base?

CSG 2D Lattice Support #31993

Uh oh!

Conversation

kkiesling commented Nov 26, 2025

Reason

Design

Impact

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

moosebuild commented Nov 26, 2025

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!

kkiesling commented Nov 26, 2025

Uh oh!

kkiesling commented Nov 26, 2025

Uh oh!

moosebuild commented Nov 26, 2025

Framework test summary

Added tests

Modules test summary

Uh oh!

moosebuild commented Nov 26, 2025

Uh oh!

moosebuild commented Nov 26, 2025

Framework coverage

Modules coverage

Full coverage reports

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!

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!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants