Codestin Search App

lhy11009 · 2026-09-03T20:15:46Z

Pull Request Checklist. Please read and check each box with an X. Delete any part not applicable. Ask on the forum if you need help with any step.

This is a PR that modified the usage of reaction progress in the Visco Plastic model, with the new reaction chain interface from @buchanankerswell

Before your first pull request:

I have read the guidelines in our CONTRIBUTING.md document.

For all pull requests:

I have followed the instructions for indenting my code.
I followed the AI Policy:
- Significant parts of this PR were written by AI (please add a sentence below).
- AI has not been used in significant ways.

Codex does the actual implementation. I assure that I read all the lines and performed tests myself.

For new features/models or changes of existing features:

I have tested my new feature locally to ensure it is correct.
I have created a testcase for the new feature/benchmark in the tests/ directory.
I have added a changelog entry in the doc/modules/changes directory that will inform other users of my change.

gassmoeller

I like it, we should make sure to use the new interface.

One question though: As far as I understand, this for now uses ReactionChain to clamp the reaction progress values, but it does not yet use ReactionChain to actually progress the reaction, do I understand this correctly?

gassmoeller · 2026-09-04T12:35:30Z

+          const unsigned int n_reaction_progress_fields =
+            this->introspection().get_number_of_fields_of_type(CompositionalFieldDescription::reaction_progress);
+
+          if (n_reaction_progress_fields > 0)
+            {
+              reaction_chain.initialize_simulator(this->get_simulator());
+              reaction_chain.parse_parameters(prm);
+
+              AssertThrow(reaction_chain.n_reactions() == n_reaction_progress_fields,
+                          ExcMessage("The number of reactions configured in 'Reaction chain/Kinetic models' must match "
+                                     "the number of reaction progress compositional fields."));
+              AssertThrow(reaction_progress_mapping.size() == n_reaction_progress_fields,
+                          ExcMessage("The number of entries in 'Reaction progress mapping' must match the number of "
+                                     "reaction progress compositional fields."));
+            }


I would say unconditionally always initialize and parse the reaction chain parameters, and move the asserts into the parse_parameters function of reaction_chain. This way you will always have a fully functional reaction_chain object, that may just happen to manage 0 reactions. But that is safer than the current state, in which reaction_chain may or may not be initialized.

Changed. But there are questions. See my other comments

buchanankerswell · 2026-09-04T14:12:29Z

I like it, we should make sure to use the new interface.

One question though: As far as I understand, this for now uses ReactionChain to clamp the reaction progress values, but it does not yet use ReactionChain to actually progress the reaction, do I understand this correctly?

Just wanted to flag something for your attention @gassmoeller and @lhy11009.

@lhy11009 may or may not have already noticed issue #7287 in this updated visco plastic model. I have two open PRs that ensure:

the ReactionChain doesn't crash the simulation when reaction_progress fields under/overshoot [0, 1] Relax reaction progress tolerance by default (1e20) #7319
that the material model can write the clamped progress back to the solution Add no-op hook to material model interface to clamp reaction progress… buchanankerswell/aspect#1 (e.g., see https://github.com/buchanankerswell/aspect/blob/7978d4d2fbbd310de76b37f72ff8b559afaf00c6/cookbooks/mtz_kinetics/mtz-kinetics.cc#L328-L345)

Together, these PRs solve #7287. I have tested this in cookbook #7119.

I would suggest we merge #7319 and buchanankerswell#1 before going too much further. Otherwise all material models with ReactionChain will run into issue #7287, including this one.

gassmoeller · 2026-09-04T14:38:01Z

Good point. I wanted to get to your PRs next week. My comment above still applies, but we should figure out the other issues before merging this PR.

lhy11009 · 2026-09-05T00:38:26Z

Thanks to both of you for the heads-up.

I left my comments #7287

This is indeed an important issue to solve. Let me know how I can contribute. @buchanankerswell, would you be able to join the user meeting next Mon to discuss?

On the other hand, it wouldn't change much what we need to do in this PR.

lhy11009 · 2026-09-05T00:45:02Z

I like it, we should make sure to use the new interface.

One question though: As far as I understand, this for now uses ReactionChain to clamp the reaction progress values, but it does not yet use ReactionChain to actually progress the reaction, do I understand this correctly?

Exactly, that is the case.

In the existing test, I basically assign an initial value to the transition progress field.

My intention is to first connect all the interfaces before using ReactionChain to actually progress the reaction.

lhy11009 · 2026-09-05T18:02:23Z

    void
    ViscoPlastic<dim>::initialize()
    {
+      reaction_chain.initialize();


I have put the initialize function here, but the initial_simulator call is in the parse_parameters. Feel free to suggest otherwise.

lhy11009 · 2026-09-05T18:03:23Z

          reaction_progress_mapping = Utilities::string_to_unsigned_int
                                      (Utilities::split_string_list(prm.get ("Reaction progress mapping")));
+
+          AssertThrow(reaction_progress_mapping.size() == n_reaction_progress_fields,


I leave this assert here, because this is checking the number of reaction_progress_mapping that is parsed in here.

lhy11009 · 2026-09-05T18:04:39Z

-                            Patterns::List(Patterns::Selection(ReactionModelPluginList<dim>::get_pattern_of_names()),
-                                           1, Patterns::List::max_int_value, "|"),
+                            "",
+                            Patterns::List(Patterns::Anything()),


I have changed the patterns here, to allow 0 kinetic models by default, similar to the "name of fields" entry for the "Composition".

lhy11009 · 2026-09-05T18:05:02Z

-                                           1, Patterns::List::max_int_value, "|"),
+                            "",
+                            Patterns::List(Patterns::Anything()),
                            "'|'-separated list of registered reaction kinetics model names, one per reaction in the chain. The number of entries "


In addition, should we change "|" to ","?

lhy11009 · 2026-09-05T18:06:57Z

+          const unsigned int n_reaction_progress_fields =
+            this->introspection().get_number_of_fields_of_type(CompositionalFieldDescription::reaction_progress);
+
+          AssertThrow(n_reactions == n_reaction_progress_fields,


I move one of the asserts here. For these assertions to work, things should be parsed in this order: compositions -> reaction chains -> the assigned "mapping" in the material model. Tests passed, so this should be the case.

lhy11009 · 2026-09-05T18:09:00Z

@gassmoeller, please take another look when available. And check my posts above.

lhy11009 commented Sep 3, 2026

View reviewed changes

Comment thread tests/phase_kinetics_composition_metastable/statistics

gassmoeller reviewed Sep 4, 2026

View reviewed changes

lhy11009 mentioned this pull request Sep 5, 2026

Operator-splitting reaction progress fields (xi_*) can overshoot [0, 1] and break ReactionChain monotonicity #7287

Open

Use reaction chain in visco plastic model

2ded8e4

lhy11009 force-pushed the visco_plastic_reaction_chain branch from 22dcdc2 to 2ded8e4 Compare September 5, 2026 18:00

lhy11009 commented Sep 5, 2026

View reviewed changes

Uh oh!

Conversation

lhy11009 commented Sep 3, 2026

Before your first pull request:

For all pull requests:

For new features/models or changes of existing features:

Uh oh!

Uh oh!

gassmoeller left a comment

Choose a reason for hiding this comment

Uh oh!

gassmoeller Sep 4, 2026

Choose a reason for hiding this comment

Uh oh!

lhy11009 Sep 5, 2026

Choose a reason for hiding this comment

Uh oh!

buchanankerswell commented Sep 4, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

gassmoeller commented Sep 4, 2026

Uh oh!

lhy11009 commented Sep 5, 2026

Uh oh!

lhy11009 commented Sep 5, 2026

Uh oh!

lhy11009 Sep 5, 2026

Choose a reason for hiding this comment

Uh oh!

lhy11009 Sep 5, 2026

Choose a reason for hiding this comment

Uh oh!

lhy11009 Sep 5, 2026

Choose a reason for hiding this comment

Uh oh!

lhy11009 Sep 5, 2026

Choose a reason for hiding this comment

Uh oh!

lhy11009 Sep 5, 2026

Choose a reason for hiding this comment

Uh oh!

lhy11009 commented Sep 5, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

buchanankerswell commented Sep 4, 2026 •

edited

Loading