CSG Transformations by kkiesling · Pull Request #32403 · idaholab/moose

kkiesling · 2026-02-27T17:46:01Z

refs / closes #32308

Reason

Design

Impact

…ake up space

…er object types idaholab#32308

kkiesling · 2026-02-27T19:53:10Z

@shikhar413

moosebuild · 2026-02-27T20:37:38Z

Job Documentation, step Docs: sync website on 6210b19 wanted to post the following:

View the site here

This comment will be updated on new commits.

moosebuild · 2026-03-02T03:56:55Z

Job Coverage, step Generate coverage on 6210b19 wanted to post the following:

Framework coverage

	aec0a3	#32403 6210b1
	Total	Total	+/-	New
Rate	85.79%	85.80%	+0.01%	100.00%
Hits	128628	128751	+123	150
Misses	21303	21311	+8	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.

moosebuild · 2026-03-02T03:56:55Z

Job Coverage, step Verify coverage on c07e899 wanted to post the following:

The following coverage requirement(s) failed:

Failed to generate chemical_reactions coverage rate (required: 92.0%)
Failed to generate combined coverage rate (required: 37.0%)
Failed to generate contact coverage rate (required: 87.0%)
Failed to generate electromagnetics coverage rate (required: 94.0%)
Failed to generate external_petsc_solver coverage rate (required: 84.0%)
Failed to generate fluid_properties coverage rate (required: 84.0%)
Failed to generate fsi coverage rate (required: 85.0%)
Failed to generate functional_expansion_tools coverage rate (required: 81.0%)
Failed to generate geochemistry coverage rate (required: 96.0%)
Failed to generate heat_transfer coverage rate (required: 87.0%)
Failed to generate level_set coverage rate (required: 85.0%)
Failed to generate misc coverage rate (required: 32.0%)
Failed to generate navier_stokes coverage rate (required: 77.0%)
Failed to generate optimization coverage rate (required: 86.0%)
Failed to generate peridynamics coverage rate (required: 77.0%)
Failed to generate phase_field coverage rate (required: 85.0%)
Failed to generate porous_flow coverage rate (required: 95.0%)
Failed to generate ray_tracing coverage rate (required: 93.0%)
Failed to generate rdg coverage rate (required: 63.0%)
Failed to generate reactor coverage rate (required: 90.0%)
Failed to generate richards coverage rate (required: 93.0%)
Failed to generate scalar_transport coverage rate (required: 81.0%)
Failed to generate solid_mechanics coverage rate (required: 84.0%)
Failed to generate solid_properties coverage rate (required: 83.0%)
Failed to generate stochastic_tools coverage rate (required: 88.0%)
Failed to generate subchannel coverage rate (required: 87.0%)
Failed to generate thermal_hydraulics coverage rate (required: 88.0%)
Failed to generate xfem coverage rate (required: 80.0%)

GiudGiud

do we need to apply transformations to surfaces?

and do you envision that transformations applied to a universe should be propagated to all cells they contain?
To know the real location of the cell we need to get or gather the full list of transformations

GiudGiud · 2026-03-02T06:01:04Z

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

+|---------------------|----------------------------------------------|--------------------|
+| `TRANSLATION`       | $\{$ x-distance, y-distance, z-distance $\}$ | none               |
+| `ROTATION`          | $\{ \phi, \theta, \psi \}$                   | none               |
+| `SCALE`             | $\{$ x-scale, y-scale, z-scale $\}$          | non-zero           |


positive no?

So I had this as a positive, non-zero restriction initially, but @shikhar413 pointed out that negative values were allowed in the TransformGenerator. I did look the source code there and did not see a restriction on any of the values for scaling, including the fact that the values could be zero. We decided to allow negative values to be consistent with that generator, but restrict zero values because that wouldn't make any sense from a CSG/MC standpoint. I am fine with going back to restricting it to positive values only if we think it makes most sense even though the TransformGenerator allows negative.

ok seems fine.
TransformGenerator negative scaling needs an extra step to flip the elements back to a positive volume.

I guess it's a way to do a central symmetry

GiudGiud · 2026-03-02T06:07:54Z

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

+
+!alert! note title=CSGRegion Transformations
+
+If applying a transformation to a `CSGRegion` object, the `CSGRegion` object will not store the transformation information.Rather, the transformation will be applied to each `CSGSurface` associated with that `CSGRegion`. This means that if one `CSGSurface` object is used in multiple `CSGRegion` objects, all regions with that surface may be affected by the transformation. If the intention is to apply a transformation to just one use case of the shared `CSGSurface`, then a clone of that surface should be made prior to applying the transformation.


Suggested change

If applying a transformation to a `CSGRegion` object, the `CSGRegion` object will not store the transformation information.Rather, the transformation will be applied to each `CSGSurface` associated with that `CSGRegion`. This means that if one `CSGSurface` object is used in multiple `CSGRegion` objects, all regions with that surface may be affected by the transformation. If the intention is to apply a transformation to just one use case of the shared `CSGSurface`, then a clone of that surface should be made prior to applying the transformation.

If applying a transformation to a `CSGRegion` object, the `CSGRegion` object will not store the transformation information. Rather, the transformation will be applied to each `CSGSurface` associated with that `CSGRegion`. This means that if one `CSGSurface` object is used in multiple `CSGRegion` objects, all regions with that surface may be affected by the transformation. If the intention is to apply a transformation to just one use case of the shared `CSGSurface`, then a clone of that surface should be made prior to applying the transformation.

seems like something that is easy to miss. Could check in dbg mode? (in between #ifndef NDEBUG)

What exactly are you recommending to check here with debug mode?

framework/src/csg/CSGSurface.C

framework/include/csg/CSGTransformation.h

GiudGiud · 2026-03-02T14:59:37Z

unit/src/CSGTransformationTest.C

+  std::vector<std::pair<TransformationType, std::tuple<Real, Real, Real>>> transformations = {
+      {TransformationType::TRANSLATION, {1.0, 2.0, 3.0}},
+      {TransformationType::ROTATION, {45.0, 30.0, 60.0}},
+      {TransformationType::SCALE, {-2.0, 2.0, 2.0}}};


maybe use a valid one for the test in case the checking moves around

assuming you are referring to the negative value in the SCALE one, that was intentional given the fact that negatives are valid there, so I will leave this.

kkiesling · 2026-03-02T15:20:35Z

do we need to apply transformations to surfaces?

We decided against this because it seems like most of the MC codes actually have methods in place to apply transformations to surfaces (or other entities) and so it would be more reliable for the codes to use their own methods.

and do you envision that transformations applied to a universe should be propagated to all cells they contain? To know the real location of the cell we need to get or gather the full list of transformations

yes I envision this would be iteratively applied to each cell within a universe

GiudGiud · 2026-03-02T15:24:04Z

framework/include/csg/CSGSurface.h

+  getTransformations() const
+  {
+    return _transformations;
+  }


right my question is more that transformations are not usually applied to surfaces in MC codes?
At least in OpenMC if I recall correctly

Surfaces are the primary entities that can have transformations applied to them across all MC codes, various codes allow for cells and universes too under different circumstances (these are less consistent across codes). But pretty much all codes support surface transformations directly with built-in methods.

At the end of the day, it is really all the surfaces that need to be transformed to alter the geometry with a transformation.

kkiesling · 2026-03-03T22:32:44Z

Just realized that I need to add a check for transformations in the equality operator overloads for each of these object types. Making note here so I don't forget to do that. EDIT: done

shikhar413 · 2026-03-10T15:17:21Z

@kkiesling now that we have transformations defined here, do we want to update the documentation for hexagonal lattices to mention that x-type lattices should be implemented by applying a rotation to a y-type lattice?

kkiesling · 2026-03-10T17:25:19Z

now that we have transformations defined here, do we want to update the documentation for hexagonal lattices to mention that x-type lattices should be implemented by applying a rotation to a y-type lattice?

good idea. Just added a note in the docs.

moosebuild · 2026-03-10T18:32:42Z

Job Test, step Results summary on 6210b19 wanted to post the following:

Framework test summary

Compared against aec0a32 in job civet.inl.gov/job/3640883.

No added tests

Run time changes

Test	Base (s)	Head (s)	+/-	Base (MB)	Head (MB)
`executioners/multisys-ss-detection.multi-sys-ss`	4.99	7.51	+50.60%	136.29	136.47

Modules test summary

Compared against aec0a32 in job civet.inl.gov/job/3640883.

No added tests

Run time changes

Test	Base (s)	Head (s)	+/-	Base (MB)	Head (MB)
`solid_mechanics/test:ad_anisotropic_creep.jac_ad`	19.06	30.29	+58.96%	145.60	178.71
`stochastic_tools/test:transfers/sampler_reporter.transfer/distributed`	3.42	5.43	+58.62%	476.23	475.47
`solid_mechanics/test:rate_independent_cyclic_hardening.nonlin_isoharden_symmetric_strain_controlled`	3.86	5.86	+51.83%	126.80	129.98
`solid_mechanics/test:smeared_cracking.rz_exponential`	12.39	18.72	+51.10%	134.13	124.25
`solid_mechanics/test:crystal_plasticity/cp_eigenstrains.thermal_eigenstrain`	4.66	7.04	+51.01%	119.34	122.35

kkiesling added 19 commits February 27, 2026 11:32

separate the checking of surface presence in base

e129cd2

allow for transformations to be applied to any CSG Object type

76789a5

include transformation lists in the json output using string notation

b77d2c2

add some convenience transform methods

3891b2c

getTransformation methods can and should be public

c66af2e

add convenience methods for translation

8ed5e55

unit tests for transformation methods

2d2b032

monodirectional convenience functions don't add much value and just t…

2b8e7b2

…ake up space

csg translation documentation

daa3fd2

add rotation to CSG MG test for example purposes

ae942e8

add examples to docs

90622da

minor doc fixes

9f11c18

simplified check

229c104

negative scales are valid, but 0 is not

97bdc17

use tuple of reals instead of vector to avoid having to check length

0bf54e6

failsafe assert to make sure objs are correct

14862c9

use setup function for less redundancy

b2751c1

remove unnecessary includes

e36ef7c

move transfromations for surfaces to private for consistency with oth…

e2e24ac

…er object types idaholab#32308

kkiesling requested review from GiudGiud and lindsayad as code owners February 27, 2026 17:46

kkiesling added 4 commits February 27, 2026 11:52

clang format

847d014

include not actually needed

a38a4d9

fix scaling restrictions in docs

ce517fe

typo

c07e899

GiudGiud self-assigned this Mar 2, 2026

GiudGiud reviewed Mar 2, 2026

View reviewed changes

kkiesling added 4 commits March 6, 2026 13:18

correct documentation wording

6c341a1

rename apply->add for more accurate nomenclature

735736a

check transformations in equality too

4e2558e

typo

9b7461f

kkiesling requested a review from GiudGiud March 6, 2026 21:18

add note about using rotations for y-oriented hex lattices

6210b19


		!alert! note title=CSGRegion Transformations

		If applying a transformation to a `CSGRegion` object, the `CSGRegion` object will not store the transformation information.Rather, the transformation will be applied to each `CSGSurface` associated with that `CSGRegion`. This means that if one `CSGSurface` object is used in multiple `CSGRegion` objects, all regions with that surface may be affected by the transformation. If the intention is to apply a transformation to just one use case of the shared `CSGSurface`, then a clone of that surface should be made prior to applying the transformation.

Conversation

kkiesling commented Feb 27, 2026

Reason

Design

Impact

Uh oh!

kkiesling commented Feb 27, 2026

Uh oh!

moosebuild commented Feb 27, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

moosebuild commented Mar 2, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Framework coverage

Modules coverage

Full coverage reports

Uh oh!

moosebuild commented Mar 2, 2026

Uh oh!

GiudGiud left a comment

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!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

kkiesling commented Mar 2, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

kkiesling commented Mar 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

shikhar413 commented Mar 10, 2026

Uh oh!

kkiesling commented Mar 10, 2026

Uh oh!

moosebuild commented Mar 10, 2026

Framework test summary

No added tests

Run time changes

Modules test summary

No added tests

Run time changes

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

moosebuild commented Feb 27, 2026 •

edited

Loading

moosebuild commented Mar 2, 2026 •

edited

Loading

kkiesling commented Mar 3, 2026 •

edited

Loading