Combine ad non-AD mat diffusion #29888

tophmatthews · 2025-02-15T15:10:03Z

tophmatthews · 2025-02-15T15:11:52Z

I'm trying to consolidate ADMatDiffusion and MatDiffusion and got close, but am now stuck...I'm afraid it may be impossible becuase ADKernel retains a templated feature, while Kernel doesn't. This cascades all the way down through KernelGrad, and MatDiffusionBase. A little help, maybe @dschwen since you did some of the early work on KernelGrad?

tophmatthews · 2025-02-24T17:03:16Z

grrr this is close, but gets a dyld[37997]: symbol not found in flat namespace '__ZN21MatDiffusionBaseTemplIN7libMesh11TensorValueIdEELb0EE12initialSetupEv' due to something with MatAnisoDiffusionTempl in the phase_field module...

dschwen · 2025-02-25T06:09:02Z

Check your explicit instantiations.

framework/src/kernels/MatDiffusionBase.C

tophmatthews · 2025-02-25T19:56:28Z

How do I deal with this @dschwen:

/tmp/build/marmot/build/header_symlinks/SwellingMatDiffusionBase.h:32:60: error: 'MatDiffusionBase' was not declared in this scope; did you mean 'MatDiffusionBaseTempl'?
   32 | class SwellingMatDiffusionBase : public SwellingGradKernel<MatDiffusionBase<T>>
      |                                                            ^~~~~~~~~~~~~~~~
      |                                                            MatDiffusionBaseTempl

I can't really figure out a way to do partial specialization with templates...Do I have to go modify Marmot directly after this is merged?

tophmatthews · 2025-02-25T19:59:08Z

oh maybe:

template <typename T>
using MatDiffusionBase = MatDiffusionBaseTempl<T, false>;

framework/src/kernels/MatDiffusionBase.C

moosebuild · 2025-02-25T23:02:06Z

Job Coverage, step Generate coverage on 7d3d0b5 wanted to post the following:

Framework coverage

	236997	#29888 7d3d0b
	Total	Total	+/-	New
Rate	85.93%	85.94%	+0.01%	100.00%
Hits	122919	122911	-8	50
Misses	20122	20108	-14	0

Diff coverage report

Full coverage report

Modules coverage

Phase field

	236997	#29888 7d3d0b
	Total	Total	+/-	New
Rate	86.10%	86.10%	+0.00%	100.00%
Hits	14018	14012	-6	6
Misses	2264	2263	-1	0

Diff coverage report

Full coverage report

Full coverage reports

Reports

This comment will be updated on new commits.

moosebuild · 2025-02-25T23:05:01Z

Job Documentation, step Docs: sync website on 7d3d0b5 wanted to post the following:

View the site here

This comment will be updated on new commits.

moosebuild · 2025-02-26T01:11:58Z

Job Precheck, step Clang format on dc4ab13 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/29888/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 12e3b024e9909d2df7bc9252a156be46f77c76ea

tophmatthews · 2025-02-27T13:59:12Z

_coupled_moose_vars is broken now...wtf...

no that's not it...

tophmatthews · 2025-02-27T16:04:32Z

I'm missing something @dschwen , the overwritten non-AD precomputeQpJacobian in the new class MatDiffusionBase is not being called at all. It's going through the qp and the j's in KernelGrad, but virtual RealGradient precomputeQpJacobian() override; is not respected. Help!

tophmatthews · 2025-02-27T16:06:14Z

Oh wait, it's because MatDiffusion inherits from MatDiffusionBaseTempl<Real, is_ad> not MatDiffusionBase...

tophmatthews · 2025-02-27T17:16:39Z

~~Cheese and crackers, I'm doubting this being a worthwhile exercise, I'm out of ideas...~~

dyld[57678]: symbol not found in flat namespace '__ZN21MatDiffusionBaseTemplIN7libMesh11TensorValueIdEELb0EE20precomputeQpResidualEv'

framework/include/kernels/GenericKernelGrad.h

tophmatthews · 2025-02-28T20:08:29Z

Holy crapoli this PR will be the end of me...

First issue: There is a intermittent failure in the optimization module:

optimizationreporter/function_misfit.point_loads/volume: Running csvdiff: /opt/moose/share/moose/python/mooseutils/csvdiff.py /tmp/moosetest.TCpJSo/optimization/moose/optimization/optimizationreporter/function_misfit/gold/main_auto_out_OptimizationReporter_0001.csv /tmp/moosetest.TCpJSo/optimization/moose/optimization/optimizationreporter/function_misfit/main_auto_out_OptimizationReporter_0001.csv --relative-tolerance 0.0001 --abs-zero 0.0001
optimizationreporter/function_misfit.point_loads/volume: ERROR: In file main_auto_out_OptimizationReporter_0001.csv: The values in column "parameter_results" don't match. 
optimizationreporter/function_misfit.point_loads/volume: 	relative diff:   -8.997e+02 ~ -8.987e+02 = 1.105e-03 (1.105e-03)

I can't reproduce it locally, and you can see in the civet history it hits often, but somewhat randomly. I'm not sure if has something to do with #29951 or not.

Second issue:
There are Marmot failures due to

1. renaming of member variables (like _D to _diffusivity), and
1. redefinition of return type of MatDiffusionBase<T>::computeQpCJacobian() due to moving to precomputeQpJacobian from KernelGrad.

For (1), I can just use the less (personally) favorable variables names. For (2), I can make a temporary precomputeQpCJacobian, and cleanup Marmot after the MOOSE fixes. OR I can just cleanup MARMOT after. What do you think @dschwen

tophmatthews · 2025-03-24T18:47:40Z

Hi @lynnmunday @maxnezdyur , @dschwen recommended reaching out to you guys on this:
There is a intermittent failure in the optimization module:

optimizationreporter/function_misfit.point_loads/volume: Running csvdiff: /opt/moose/share/moose/python/mooseutils/csvdiff.py /tmp/moosetest.TCpJSo/optimization/moose/optimization/optimizationreporter/function_misfit/gold/main_auto_out_OptimizationReporter_0001.csv /tmp/moosetest.TCpJSo/optimization/moose/optimization/optimizationreporter/function_misfit/main_auto_out_OptimizationReporter_0001.csv --relative-tolerance 0.0001 --abs-zero 0.0001
optimizationreporter/function_misfit.point_loads/volume: ERROR: In file main_auto_out_OptimizationReporter_0001.csv: The values in column "parameter_results" don't match. 
optimizationreporter/function_misfit.point_loads/volume: 	relative diff:   -8.997e+02 ~ -8.987e+02 = 1.105e-03 (1.105e-03)

I can't reproduce it locally, and you can see in the civet history it hits often, but somewhat randomly. I'm not sure if has something to do with #29951 or not.

This happens in moose/modules/optimization/test/tests/optimizationreporter/function_misfit/forward_and_adjoint.i which is run from main_auto.i

Any thoughts?

maxnezdyur · 2025-03-24T18:54:47Z

In the test file, for the test failing you can add cli args that the test will be run with. Can you add cli_args = "-pc_type lu", if we still see intermittent failures then it's not the issue you referenced.

tophmatthews · 2025-10-03T16:27:22Z

Any luck @travismui ?

tophmatthews · 2025-10-09T23:32:12Z

Looks like it was due to SAM using computeQpJacobian, but the new KernelGrad version for MatDiffusion uses precomputeQpJacobian. I'll add a final declaration to computeQpResidual and computeQpJacobian, which exists for ADKernelGrad, but not KernelGrad. That will prevent this slipping through the cracks. That means that Marmot and SAM will require modifications with this new PR. I'm not sure the next steps there, but I have a fix for Marmot, and @travismui has a fix for SAM ready to go.

travismui · 2025-10-10T02:25:52Z

Looks like it was due to SAM using computeQpJacobian, but the new KernelGrad version for MatDiffusion uses precomputeQpJacobian. I'll add a final declaration to computeQpResidual and computeQpJacobian, which exists for ADKernelGrad, but not KernelGrad. That will prevent this slipping through the cracks. That means that Marmot and SAM will require modifications with this new PR. I'm not sure the next steps there, but I have a fix for Marmot, and @travismui has a fix for SAM ready to go.

SAM patch prepared at SAM/SAM!1124, will be merged after this goes into next (I think that's how this works?)

tophmatthews · 2025-10-13T14:01:09Z

Marmot patch is ready at https://github.inl.gov/ncrc/marmot/pull/948.

@dschwen this is ready for re-review!

framework/doc/content/source/kernels/ADMatDiffusion.md

framework/include/kernels/GenericKernelGrad.h

framework/src/kernels/KernelGrad.C

moosebuild · 2025-10-14T22:58:12Z

Job Documentation public apps on 7d3d0b5 : invalidated by @tophmatthews

Updated TMAP8 to remove ADMatDiffusion.md reference

tophmatthews · 2025-10-15T12:32:42Z

Ready to go, with patches for MARMOT and SAM ready.

tophmatthews · 2025-10-24T14:41:51Z

This is ready to go in right? What else can I do to help?

GiudGiud · 2025-10-27T01:39:42Z

Unfortunately not. It needs both a marmot and a SAM patch

EDIT: ok you made those

tophmatthews · 2025-10-27T21:19:54Z

Unfortunately not. It needs both a marmot and a SAM patch

soooo.........

loganharbour · 2025-10-27T22:47:04Z

SAM devel is broken at the moment

tophmatthews · 2025-10-31T16:33:36Z

Is sam broken still? This PR should still be ready....

GiudGiud · 2025-10-31T16:56:02Z

SAM should be fine now.

However this patch is breaking right, there is no point running tests again?

tophmatthews · 2025-10-31T17:38:33Z

However this patch is breaking right, there is no point running tests again?

Yeah. This PR breaks MARMOT and SAM. There are patches for both of those codes ready once this is merged that will unbreak them. Although it's been so long now, something else could be breaking. That's why I'm anxious to get this done with.

tophmatthews · 2025-10-31T17:51:49Z

Thanks @GiudGiud ! For the MARMOT patch, do I update the moose submodule to next or devel or not at all?

GiudGiud · 2025-10-31T18:08:04Z

to next please

tophmatthews · 2025-10-31T18:58:35Z

@travismui can you do the sam patching? Thanks!

loganharbour force-pushed the next branch from cfc8970 to 86b9411 Compare February 19, 2025 14:51

GiudGiud assigned dschwen Feb 22, 2025

tophmatthews force-pushed the combine_ad_nonad_mat_diffusion_15915 branch 2 times, most recently from a66462f to 113bc53 Compare February 24, 2025 16:31

dschwen reviewed Feb 25, 2025

View reviewed changes

framework/src/kernels/MatDiffusionBase.C Outdated Show resolved Hide resolved

tophmatthews force-pushed the combine_ad_nonad_mat_diffusion_15915 branch from e29ee32 to e668f01 Compare February 25, 2025 17:49

dschwen changed the title ~~Combine ad nonad mat diffusion 15915~~ Combine ad non-AD mat diffusion Feb 25, 2025

dschwen reviewed Feb 25, 2025

View reviewed changes

framework/src/kernels/MatDiffusionBase.C Outdated Show resolved Hide resolved

tophmatthews force-pushed the combine_ad_nonad_mat_diffusion_15915 branch from dc4ab13 to 3591419 Compare February 26, 2025 21:23

tophmatthews force-pushed the combine_ad_nonad_mat_diffusion_15915 branch from 3591419 to c5dda9f Compare February 27, 2025 16:03

tophmatthews force-pushed the combine_ad_nonad_mat_diffusion_15915 branch from 9a4c958 to 6be38a9 Compare February 27, 2025 17:54

dschwen reviewed Feb 27, 2025

View reviewed changes

framework/include/kernels/GenericKernelGrad.h Outdated Show resolved Hide resolved

tophmatthews force-pushed the combine_ad_nonad_mat_diffusion_15915 branch 2 times, most recently from 72f5964 to a1a7578 Compare March 7, 2025 13:45

tophmatthews force-pushed the combine_ad_nonad_mat_diffusion_15915 branch from a1a7578 to 90050aa Compare March 21, 2025 14:42

tophmatthews marked this pull request as draft October 3, 2025 16:27

add checks for computeQp methods

e70544b

tophmatthews force-pushed the combine_ad_nonad_mat_diffusion_15915 branch from 3298154 to e70544b Compare October 10, 2025 00:50

tophmatthews marked this pull request as ready for review October 13, 2025 14:53

dschwen reviewed Oct 14, 2025

View reviewed changes

framework/doc/content/source/kernels/ADMatDiffusion.md Outdated Show resolved Hide resolved

dschwen reviewed Oct 14, 2025

View reviewed changes

framework/include/kernels/GenericKernelGrad.h Outdated Show resolved Hide resolved

dschwen reviewed Oct 14, 2025

View reviewed changes

framework/src/kernels/KernelGrad.C Outdated Show resolved Hide resolved

tophmatthews added 2 commits October 14, 2025 10:23

review comments

873251d

remove ADMatDiffusion.md

7d3d0b5

tophmatthews mentioned this pull request Oct 14, 2025

Remove ADMatDiffusion.md references idaholab/TMAP8#319

Closed

dschwen approved these changes Oct 15, 2025

View reviewed changes

tophmatthews mentioned this pull request Oct 15, 2025

Cleanup and adify Reaction and MatReaction #31728

Open

GiudGiud added the PR: Ready for review/merge label Oct 20, 2025

GiudGiud merged commit 4e5abd1 into idaholab:next Oct 31, 2025
65 of 69 checks passed

Combine ad non-AD mat diffusion #29888

Combine ad non-AD mat diffusion #29888

Uh oh!

Conversation

tophmatthews commented Feb 15, 2025

Uh oh!

tophmatthews commented Feb 15, 2025

Uh oh!

tophmatthews commented Feb 24, 2025

Uh oh!

dschwen commented Feb 25, 2025

Uh oh!

Uh oh!

tophmatthews commented Feb 25, 2025

Uh oh!

tophmatthews commented Feb 25, 2025

Uh oh!

Uh oh!

moosebuild commented Feb 25, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Framework coverage

Modules coverage

Phase field

Full coverage reports

Uh oh!

moosebuild commented Feb 25, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

moosebuild commented Feb 26, 2025

Uh oh!

tophmatthews commented Feb 27, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

tophmatthews commented Feb 27, 2025

Uh oh!

tophmatthews commented Feb 27, 2025

Uh oh!

tophmatthews commented Feb 27, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

tophmatthews commented Feb 28, 2025

Uh oh!

tophmatthews commented Mar 24, 2025

Uh oh!

maxnezdyur commented Mar 24, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

tophmatthews commented Oct 3, 2025

Uh oh!

tophmatthews commented Oct 9, 2025

Uh oh!

travismui commented Oct 10, 2025

Uh oh!

tophmatthews commented Oct 13, 2025

Uh oh!

Uh oh!

Uh oh!

Uh oh!

moosebuild commented Oct 14, 2025

Uh oh!

tophmatthews commented Oct 15, 2025

Uh oh!

tophmatthews commented Oct 24, 2025

Uh oh!

GiudGiud commented Oct 27, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

tophmatthews commented Oct 27, 2025

Uh oh!

loganharbour commented Oct 27, 2025

Uh oh!

tophmatthews commented Oct 31, 2025

Uh oh!

GiudGiud commented Oct 31, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

tophmatthews commented Oct 31, 2025

Uh oh!

Uh oh!

moosebuild commented Feb 25, 2025 •

edited

Loading

moosebuild commented Feb 25, 2025 •

edited

Loading

tophmatthews commented Feb 27, 2025 •

edited

Loading

tophmatthews commented Feb 27, 2025 •

edited

Loading

maxnezdyur commented Mar 24, 2025 •

edited

Loading

GiudGiud commented Oct 27, 2025 •

edited

Loading

GiudGiud commented Oct 31, 2025 •

edited

Loading