Tupek/system solver extended by tupek2 · Pull Request #1609 · llnl/smith

tupek2 · 2026-05-20T21:17:26Z

This introduces a new interface for constructing multiphysics systems of equations, their boundary conditions, time integration rules, and solving them using potentially customized staggered or coupled algorithms.

…em).

…d dynamic steps.

… we go along. workout some stress output details.

…eanup cycle zero handling.

…terface.

…gain.

lihanyu97 · 2026-06-18T19:37:59Z

+    weak_form_->AddDomainIntegral(
+        Dimension<spatial_dim>{}, DependsOn<all_params...>{},
+        [dt, cycle, mode, integrand](double time, auto X, auto... inputs) {
+          return integrand(TimeInfo(time, *dt, *cycle, *mode), X, inputs...);


I think reconstructing time_info here will also result in double adding dt, as we observed in time_discretized_weak_form. Since we are already storing dt_, cycle_, and mode_, can we store time_info_ = time_info when we invoke residual() and jacobian()?

Then the lambda injected for these add integral functions would be

[time_info, integrand] (double /* time */, auto X, auto... inputs) { return integrand(time_info, X, inputs...); }

lihanyu97 · 2026-07-01T16:16:15Z

+  return applyTimeRuleToPrefix(
+      rule, t_info,
+      [&](auto... self_states_and_tail) {
+        constexpr std::size_t n_self = sizeof...(self_states_and_tail) - tail_count;
+        auto all = std::forward_as_tuple(self_states_and_tail...);
+        return [&]<std::size_t... Si, std::size_t... Ti>(std::index_sequence<Si...>, std::index_sequence<Ti...>) {
+          return applyCouplingTimeRules(
+              coupling, t_info,
+              [&](auto... interpolated_coupling) {
+                return std::forward<Callback>(callback)(std::get<Si>(all)..., interpolated_coupling...);
+              },
+              std::get<n_self + Ti>(all)...);
+        }(std::make_index_sequence<n_self>{}, std::make_index_sequence<tail_count>{});
+      },
+      raw_args...);
+}


The nested return here is not very human-readable. Can we simplify this method?

white238 · 2026-07-13T22:14:54Z

+                                         smith::TimeInfo(physics->time(), dt, static_cast<size_t>(physics->cycle())));
+  }
+
+  std::cout << "reaction norm: " << physics->getReactionStates().front().get()->Norml2() << '\n';


I'd prefer SLIC_INFO + std::format, easier to read imo.

white238 · 2026-07-13T22:16:20Z

+int main(int argc, char* argv[])
+{
+  // _init_start
+  smith::ApplicationManager application_manager(argc, argv);


These examples are user facing so should have some inline comments

white238 · 2026-07-13T22:23:21Z

+#include "smith/differentiable_numerics/nonlinear_solve.hpp"
+#include "smith/differentiable_numerics/make_time_info_material.hpp"
+#include "smith/physics/functional_objective.hpp"
+#include "gretl/wang_checkpoint_strategy.hpp"


I think this is a larger problem than this PR but we have gretl as an optional TPL guarded by the CMake option SMITH_ENABLE_GRETL. This will fail if someone turns that off, which I don't think anyone has tried.

That's a good point. Gretl really isn't optional, core functionality (differentiability) depends on it. Unless I'm mistaken? If not, the CMake option should be removed IMO.

Issue created. we can handle it outside of this PR #1630

btalamini

Creating multi-physics is very nice with this!

btalamini · 2026-06-06T14:55:33Z


  /// @brief evaluate value of the ode state as used by the integration rule
  template <typename T1>
+  /// @brief Return the static field value.


Looks like the agent added a second set of doxygen comments, (one before the template, the other after. Delete one set. The new set seems more specific then the old, maybe delete the old ones.

btalamini · 2026-07-13T21:26:30Z

+  ///        @c h = 1e-4 * max(1,|t|).
+  ///
+  /// Rebuilt on each call from current value-level essentials, so late additions are reflected.
+  const smith::BoundaryConditionManager& getSecondDerivativeManager() const;


What is the purpose of this? It doesn't seem to be used. Is this to enable acceleration BCs?

btalamini · 2026-07-13T22:01:34Z

+  Strumpack,  /**< Strumpack MPI-enabled direct frontal solver*/
+  PetscCG,    /**< PETSc MPI-enabled conjugate gradient solver */
+  PetscGMRES, /**< PETSc MPI-enabled generalize minimal residual solver */
+  None        /**< Preconditioner application only, No linear solver Krylov iterations */


Maybe the intent would be clearer to call this one PrecondOnly or something like that?

chapman39

It would be good to track this PRs new feature(s) in the changelog's unreleased section under "added"

https://github.com/llnl/smith/blob/develop/RELEASE-NOTES.md?plain=1#L22

lihanyu97 · 2026-07-15T17:21:56Z

+    addBodyIntegralImpl(body_name, integrand, std::integer_sequence<int, active_parameters...>{});
+  }
+
+  /// @brief Add a body integral to the weak form.


Suggested change

/// @brief Add a body integral to the weak form.

/// @brief Add a body integral depending on all input fields to the weak form.

lihanyu97 · 2026-07-15T17:23:48Z

+    addBoundaryIntegralImpl(boundary_name, integrand, std::integer_sequence<int, active_parameters...>{});
+  }
+
+  /// @brief Add a boundary integral to the weak form.


Suggested change

/// @brief Add a boundary integral to the weak form.

/// @brief Add a boundary integral depending on all input fields to the weak form.

lihanyu97 · 2026-07-15T17:24:19Z

+    addInteriorBoundaryIntegralImpl(interior_name, integrand, std::integer_sequence<int, active_parameters...>{});
+  }
+
+  /// @brief Add an interior boundary integral to the weak form.


Suggested change

/// @brief Add an interior boundary integral to the weak form.

/// @brief Add an interior boundary integral depending on all input fields to the weak form.

lihanyu97 · 2026-07-15T17:27:40Z

-  /// @brief Accessor to get a reference to the underlying ShapeAwareFunctional in case more direct access is needed.
-  /// @return Reference to ShapeAwareFunctional instance.
-  ShapeAwareFunctional<ShapeDispSpace, OutputSpace(InputSpaces...)>& getShapeAwareResidual() { return *weak_form_; }
+ protected:


Are these methods still necessary? Overloaded functions to add integrals with all inputs are defined previously already.

lihanyu97 · 2026-07-15T17:29:07Z

+    addBodyIntegralImpl(body_name, qfunction, std::integer_sequence<int, active_parameters...>{});
+  }
+
+  /// @brief Add a body integral to the objective function.


Suggested change

/// @brief Add a body integral to the objective function.

/// @brief Add a body integral depending on all input fields.

lihanyu97 · 2026-07-15T17:29:34Z

+    addBoundaryIntegralImpl(boundary_name, qfunction, std::integer_sequence<int, active_parameters...>{});
+  }
+
+  /// @brief Add a boundary integral to the objective function.


Suggested change

/// @brief Add a boundary integral to the objective function.

/// @brief Add a boundary integral depending on all input fields.

lihanyu97 · 2026-07-15T17:36:20Z

  {
-    dt_ = time_info.dt();
-    cycle_ = time_info.cycle();
+    current_time_info_ = &time_info;


Why do we have a guard to prevent current_time_info_ from being nullified prematurally in functional_weak_form but not for functional_objective

lihanyu97

The PR looks great and allows flexibility of constructing multiphysics modules from previously defined single physics modules, which can significantly speedup future developments.

Considering the substaintial changes, can you summarize a list of interface breaking changes (eg. removal of time_discretized_weak_form)? It will be easier for other PRs created before this one to adjust to the new API.

mrtupek2 and others added 30 commits April 11, 2026 15:46

Trying to establish a more specific design for SystemOfEquation (Syst…

dc6cf98

…em).

Fix multiphysics_time_integrator loose ends

2e1778f

Decouple StateAdvancer from SystemBase, add helper functions.

513972b

Try to cleanup construction of systems.

d870691

Document and update gretl tracking capabilities

7ec5a46

Update gretl_notracking.md with the Graph-Resident but Ephemeral plan

2564572

Update gretl_notracking.md with the finalized Stop-Gradient approach

ed004cd

Update gretl_notracking.md reflecting removal of ScopedGraphDisable

35a8df8

Update gretl_notracking.md with the no-recompute optimization

533b1e2

We are trying to usable stress output as an option at the end of soli…

6797361

…d dynamic steps.

Fix style.:

9f9d51c

Get solid mechanics problems passing, with style.

6371ba5

Trying to implement system combine functionality.

0e9ddbd

Increment on the path to coupled systems.

e7456fc

Refactor system interface to make it easy to combine systems.

5a53f4b

Implement state variable system.

1b46d4c

Support a None preconditioner.

4eecd96

Working toward our dream interface.

b91c70a

Working on simpler build methods.

dbfc11c

trying to test 3 systems, also fix up some solve issues identifies as…

cad8ebb

… we go along. workout some stress output details.

Style.

2cbf8df

Simplify field and system registration.

91eeab6

Remove duplicated test.

1220878

style.

f7b90f6

Clean up some interface issues, rename cz to be clearer.

8231f28

Fix style.

cbc7ed7

Update submodules.

396392f

Merge branch 'tupek/system_solver' into tupek/system_solver_extended

16cd1d8

Remove files that were not indended to be saved

b461e9c

Try to make sure the solver and system responsibilities are clear, cl…

54a04a0

…eanup cycle zero handling.

mrtupek2 and others added 7 commits May 21, 2026 12:14

Try to be better about using existing material models with the new in…

0d4ebeb

…terface.

Try to cleanup material wrappers even more.

3b0c7e9

fix docs.

a2ab27a

Merge branch 'develop' into tupek/system_solver_extended

d5ebe10

Merge branch 'develop' into tupek/system_solver_extended

fbabc2a

Address a comment, simplify example wording a bit, enable DependsOn a…

1c0bb03

…gain.

Fix docs.

4f97908

lihanyu97 reviewed Jun 18, 2026

View reviewed changes

lihanyu97 reviewed Jul 1, 2026

View reviewed changes

tupek2 added 5 commits July 1, 2026 15:03

Merge branch 'develop' into tupek/system_solver_extended

b20e65d

Trying to simplify coupling params.

54e432b

Simplify some templating.

65f2655

Try to simplify TimeInfo usage, avoid double counting dt bug.

47cb01a

Merge branch 'develop' into tupek/system_solver_extended

8d44762

white238 reviewed Jul 13, 2026

View reviewed changes

btalamini approved these changes Jul 13, 2026

View reviewed changes

tupek2 added 2 commits July 14, 2026 14:43

Fix doxegen.

e2f399f

Fix comments, formatting, preconditioner naming.

224b89c

chapman39 reviewed Jul 14, 2026

View reviewed changes

fix some potential dangerous copies in functional objective.

c6556d8

lihanyu97 reviewed Jul 15, 2026

View reviewed changes

lihanyu97 reviewed Jul 16, 2026

View reviewed changes

	/// @brief Add a body integral to the weak form.
	/// @brief Add a body integral depending on all input fields to the weak form.

	/// @brief Add a boundary integral to the weak form.
	/// @brief Add a boundary integral depending on all input fields to the weak form.

	/// @brief Add an interior boundary integral to the weak form.
	/// @brief Add an interior boundary integral depending on all input fields to the weak form.

	/// @brief Add a body integral to the objective function.
	/// @brief Add a body integral depending on all input fields.

Uh oh!

Conversation

tupek2 commented May 20, 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!

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!

btalamini 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!

chapman39 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!

Choose a reason for hiding this comment

Uh oh!

lihanyu97 Jul 15, 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!

lihanyu97 left a comment

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

lihanyu97 Jul 15, 2026 •

edited

Loading