Added an option for multiple initial conditions in IDAKLU solver #4981

Rishab87 · 2025-04-17T17:55:23Z

Description

Added an option for multiple initial conditions in IDAKLU solver

Fixes #3713

Type of change

Please add a line in the relevant section of CHANGELOG.md to document the change (include PR #)

Important checks:

Please confirm the following before marking the PR as ready for review:

No style issues: nox -s pre-commit
All tests pass: nox -s tests
The documentation builds: nox -s doctests
Code is commented for hard-to-understand areas
Tests added that prove fix is effective or that feature works

…multiple-initial-conditions

kratman · 2025-04-17T19:03:55Z

CHANGELOG.md

+- Added "use lumped thermal capacity" option in lumped thermal model ([#4968](https://github.com/pybamm-team/PyBaMM/pull/4968))
+- Added an option for multiple initial conditions in IDAKLU solver ([#4981](https://github.com/pybamm-team/PyBaMM/pull/4981))


- Added "use lumped thermal capacity" option in lumped thermal model ([#4968](https://github.com/pybamm-team/PyBaMM/pull/4968)) - Added an option for multiple initial conditions in IDAKLU solver ([#4981](https://github.com/pybamm-team/PyBaMM/pull/4981))

Wrong section, this needs to go into unreleased.

I also fixed the lumped thermal capacity option in the changelog for the last bugfix release, so just move the new one up to unreleased

Thanks for the review moved them

codecov · 2025-04-17T22:24:33Z

Codecov Report

Attention: Patch coverage is 97.43590% with 1 line in your changes missing coverage. Please review.

Project coverage is 98.57%. Comparing base (2e9f431) to head (32242e1).
Report is 3 commits behind head on develop.

Files with missing lines	Patch %	Lines
src/pybamm/solvers/idaklu_solver.py	97.36%	1 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #4981      +/-   ##
===========================================
- Coverage    98.57%   98.57%   -0.01%     
===========================================
  Files          304      304              
  Lines        23652    23687      +35     
===========================================
+ Hits         23316    23350      +34     
- Misses         336      337       +1

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

…multiple-initial-conditions

martinjrobins

thanks @Rishab87, this is a great start and on the right track. I've made a few comments and suggestions below, happy to discuss any of them just reply to this or message me on the slack.

martinjrobins · 2025-04-24T17:11:31Z

src/pybamm/solvers/idaklu_solver.py

+                y0_list = []
+
+                for ic in initial_conditions:
+                    model_copy = model.new_copy()


you don't want to copy the model for each initial_condition, as models are quite big

martinjrobins · 2025-04-24T17:20:58Z

src/pybamm/solvers/base_solver.py

@@ -739,7 +740,11 @@ def solve(
        t_interp : None, list or ndarray, optional
            The times (in seconds) at which to interpolate the solution. Defaults to None.
            Only valid for solvers that support intra-solve interpolation (`IDAKLUSolver`).
-
+        initial_conditions : dict, numpy.ndarray, or list, optional


put somewhere here that this is only for the idaklu solver

and jax perhaps?

yeah right,added that

martinjrobins · 2025-04-24T17:22:31Z

src/pybamm/solvers/base_solver.py

@@ -911,6 +916,7 @@ def solve(
                    t_eval[start_index:end_index],
                    model_inputs_list,
                    t_interp,
+                    initial_conditions,


does this arg work for the jax solver?

Not now , I'll add support for jax in a seperate PR in future otherwise this would become too long for now

martinjrobins · 2025-04-24T17:27:40Z

src/pybamm/solvers/idaklu_solver.py

@@ -822,7 +822,44 @@ def supports_parallel_solve(self):
    def requires_explicit_sensitivities(self):
        return False

-    def _integrate(self, model, t_eval, inputs_list=None, t_interp=None):
+    def _apply_initial_conditions(self, model, initial_conditions):


this function overlaps a bit with BaseModel.set_initial_conditions_from, have you seen this function?

yeah I saw it but it works only when we have a solution object. In multiple IC scenario, we need to spin up copies of the discretised and built model before we call the IDA integrator so I think we should have a separate _apply_initial_conditions function?

They serve different purpose also, they live in different layers and they use different data to work with so I think they should be kept separate

happy to keep them separate, just wanted to point it out as this function does a lot of special handling for concatenated symbols which you might need for your _apply_initial_conditions (or might not?)

by the time we get to _apply_solver_initial_conditions, the model has already been discretized and flattened into a single y0 vector, so I dont think we need special handling for concatenated symbols

martinjrobins · 2025-04-24T17:30:47Z

tests/unit/test_solvers/test_idaklu_solver.py

+        n_sims = 3
+        initial_conditions = [{"u": i + 1} for i in range(n_sims)]
+        inputs = [{} for _ in range(n_sims)]
+        t_eval = np.linspace(0, 1, 10)


use:

t_eval = [0, 1] t_interp = np.linspace(0, 1, 10)

then

solutions = solver.solve( model, t_eval, t_interp=t_interp, inputs=inputs, initial_conditions=initial_conditions )

martinjrobins · 2025-04-24T17:31:45Z

tests/unit/test_solvers/test_idaklu_solver.py

+        disc = pybamm.Discretisation()
+        disc.process_model(model)
+
+        solver = pybamm.IDAKLUSolver(rtol=1e-8, atol=1e-10, options={"num_threads": 1})


I don't think you need these tolerances for such a simple problem, just use the defaults

without these getting assertion errors for these problems, alternatively we can add these tolerances while comparing if not here?

when comparing the solution? I would just use the default tolerances, then use these also (perhaps multipled by 10 or so) for your comparison. I notice you don't specify an absolute tolerance for your comparison, the default is 0 which is not suitable for a solution tending to 0.

martinjrobins · 2025-04-24T17:34:31Z

src/pybamm/solvers/jax_solver.py

-    def _integrate(self, model, t_eval, inputs=None, t_interp=None):
+    def _integrate(
+        self, model, t_eval, inputs=None, t_interp=None, intial_conditions=None
+    ):


If you are not planning on supporting the jax solver, you need to raise an error if initial_conditions is not None

martinjrobins · 2025-04-24T17:35:00Z

tests/unit/test_solvers/test_idaklu_solver.py

+        ):
+            solver.solve(
+                model, t_eval, inputs=inputs, initial_conditions=initial_conditions
+            )


also need to test for pde models that need to be discretised

martinjrobins · 2025-04-24T17:35:57Z

tests/integration/test_solvers/test_idaklu.py

@@ -208,3 +208,39 @@ def test_with_experiments(self):
            sols[0].cycles[-1]["Current [A]"].data,
            sols[1].cycles[-1]["Current [A]"].data,
        )
+
+    def test_multiple_initial_conditions(self):


this is just the same as the unit test I think?
for the integration tests, you should use the inbuilt models (at least SPM and DFN) and make sure the feature works with these.

…multiple-initial-conditions

Rishab87 · 2025-04-27T11:02:30Z

@martinjrobins Thanks for the review, I've made few changes accordingly and replied to all the comments

martinjrobins

thanks @Rishab87, test_multiple_initial_conditions_spm and test_multiple_initial_conditions_dfn look very similar, can we combine this into a single test that is parameterised by the model class? Also, it would be good to check the solution against two separate solves using those initial conditions, so we know the solution is correct, not just the initial condition

kratman · 2025-04-29T20:36:47Z

CHANGELOG.md

@@ -25,6 +25,8 @@
 ## Features

 - Added "use lumped thermal capacity" option in lumped thermal model ([#4968](https://github.com/pybamm-team/PyBaMM/pull/4968))
+- Added an option for multiple initial conditions in IDAKLU solver ([#4981](https://github.com/pybamm-team/PyBaMM/pull/4981))


Move this to unreleased above

Rishab87 · 2025-05-01T09:11:41Z

@martinjrobins thanks for the review again, I've made few changes in the tests and replied to the comments

Rishab87 added 3 commits April 17, 2025 23:16

initial conditions options and few tests

23e50f9

added to changelog

fe86f2b

Merge branch 'develop' of https://github.com/pybamm-team/PyBaMM into …

8b07bbd

…multiple-initial-conditions

kratman reviewed Apr 17, 2025

View reviewed changes

Rishab87 added 3 commits April 19, 2025 23:51

fixed tests, coverage, changelog and other minor changes

1b3e547

Merge branch 'develop' of https://github.com/pybamm-team/PyBaMM into …

b09df51

…multiple-initial-conditions

changes in changelog

320197e

martinjrobins requested changes Apr 24, 2025

View reviewed changes

Rishab87 added 2 commits April 27, 2025 15:07

made changes according to comments

2c4833c

Merge branch 'develop' of https://github.com/pybamm-team/PyBaMM into …

e86adc1

…multiple-initial-conditions

Rishab87 marked this pull request as ready for review April 27, 2025 11:01

Rishab87 requested a review from a team as a code owner April 27, 2025 11:01

Rishab87 requested a review from martinjrobins April 27, 2025 11:02

minor changes

32242e1

martinjrobins requested changes Apr 29, 2025

View reviewed changes

kratman reviewed Apr 29, 2025

View reviewed changes

fixed tests and tolerances

d261930

Rishab87 requested a review from martinjrobins May 1, 2025 09:11

minor change

72ef8a4

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Added an option for multiple initial conditions in IDAKLU solver #4981

Added an option for multiple initial conditions in IDAKLU solver #4981

Rishab87 commented Apr 17, 2025

kratman Apr 17, 2025 •

edited

Loading

kratman Apr 17, 2025

Rishab87 Apr 19, 2025

codecov bot commented Apr 17, 2025 •

edited

Loading

martinjrobins left a comment

martinjrobins Apr 24, 2025

martinjrobins Apr 24, 2025

martinjrobins Apr 24, 2025

Rishab87 Apr 27, 2025

martinjrobins Apr 24, 2025

Rishab87 Apr 27, 2025

martinjrobins Apr 24, 2025

Rishab87 Apr 27, 2025 •

edited

Loading

martinjrobins Apr 29, 2025

Rishab87 May 1, 2025 •

edited

Loading

martinjrobins Apr 24, 2025

martinjrobins Apr 24, 2025

Rishab87 Apr 27, 2025

martinjrobins Apr 29, 2025

martinjrobins Apr 24, 2025

martinjrobins Apr 24, 2025

martinjrobins Apr 24, 2025

Rishab87 Apr 27, 2025

Rishab87 commented Apr 27, 2025

martinjrobins left a comment

kratman Apr 29, 2025

Rishab87 commented May 1, 2025

		- Added "use lumped thermal capacity" option in lumped thermal model ([#4968](https://github.com/pybamm-team/PyBaMM/pull/4968))
		- Added an option for multiple initial conditions in IDAKLU solver ([#4981](https://github.com/pybamm-team/PyBaMM/pull/4981))

Added an option for multiple initial conditions in IDAKLU solver #4981

Are you sure you want to change the base?

Added an option for multiple initial conditions in IDAKLU solver #4981

Conversation

Rishab87 commented Apr 17, 2025

Description

Type of change

Important checks:

kratman Apr 17, 2025 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

codecov bot commented Apr 17, 2025 • edited Loading

Codecov Report

martinjrobins left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Rishab87 Apr 27, 2025 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Rishab87 May 1, 2025 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Rishab87 commented Apr 27, 2025

martinjrobins left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Rishab87 commented May 1, 2025

kratman Apr 17, 2025 •

edited

Loading

codecov bot commented Apr 17, 2025 •

edited

Loading

Rishab87 Apr 27, 2025 •

edited

Loading

Rishab87 May 1, 2025 •

edited

Loading