Initialization Module by michaelbynum · Pull Request #3912 · Pyomo/pyomo

michaelbynum · 2026-04-15T12:02:45Z

Fixes #3878

Summary/Motivation:

This PR adds a module to devel called initialization. The goal of the module is to provide methods to help initialize nonconvex nonlinear programming problems.

Changes proposed in this PR:

Add initialization module
Add an initialization method that uses a global optimization solver to find a feasible solution
Add an initialization method that builds a linear programming approximation using linear least squares
Add an initialization method that builds a piecewise linear approximation and iteratively refines it

Legal Acknowledgement

By contributing to this software project, I have read the contribution guide and agree to the following terms and conditions for my contribution:

I agree my contributions are submitted under the BSD license.
I represent I am authorized to make the contributions and grant the license. If my employer has rights to intellectual property that includes these contributions, I represent that I have received permission to make contributions and grant the required license on behalf of that employer.

…tion

Co-authored-by: Bethany Nicholson <blnicho@users.noreply.github.com>

michaelbynum · 2026-06-02T14:38:02Z

I think I am done making changes. This should be ready for a re-review. If I forgot something or did not address a comment appropriately, please let me know.

…alization

michaelbynum · 2026-06-02T14:43:02Z

And I already lied. I forgot one change. Should be good for a re-review now.

mrmundt

I do have somewhat stronger feelings about model vs nlp on the public-facing functions. Otherwise, this is looking great.

mrmundt · 2026-06-02T15:13:51Z

+    # stat, x = lp_init_ex()
+    # stat, x = pwl_init_ex()


Intentionally commented out?

Yes. This was intentional. Just to make it easy to run the example with different methods.

mrmundt · 2026-06-02T15:16:38Z

+    logger.info(
+        f'solved NLP with {nlp_solver.name}: {res.solution_status}, {res.termination_condition}'
+    )
+    if res.solution_status in {SolutionStatus.feasible, SolutionStatus.optimal}:


I agree with Michael - this can be in a separate PR.

mrmundt · 2026-06-02T15:18:17Z

+
+
+def initialize_with_piecewise_linear_approximation(
+    nlp: BlockData,


John asked a similar question on I think a test file earlier ("why nlp instead of model?"). For tests, I really don't think it matters, but here, I think it should be model. It'll make more sense to our users if they look at the docs to see model (the Pyomo style standard) rather than nlp.

I didn't realize there was a pyomo style standard here. I don't think I agree that model is any more descriptive than nlp. Also, we are not consistent at all. In fact, I think we use m more than model:

find ./ -name "*.py" | xargs grep "(m," | wc -l: 1685
find ./ -name "*.py" | xargs grep "(model," | wc -l: 1025
find ./ -name "*.py" | xargs grep "(instance," | wc -l: 104
find ./ -name "*.py" | xargs grep "(block," | wc -l: 126
find ./ -name "*.py" | xargs grep "(b," | wc -l: 246

There are also many other examples, such as working_blk, working_model, separation_model, subproblem, root, blk, scenario_instance, ef, etc. Some of these argument names have meanings that are significantly more useful than m or model.

mrmundt · 2026-06-02T15:19:08Z

+
+
+def initialize_with_LP_approximation(
+    nlp: BlockData,


Same as above (model instead of nlp).

mrmundt · 2026-06-02T15:19:46Z

+
+
+def initialize_with_global_opt(
+    nlp: BlockData,


Same as above (model instead of nlp).

Co-authored-by: Miranda Mundt <55767766+mrmundt@users.noreply.github.com>

…alization

jsiirola

Two minor changes / questions (and some notes for later development)

jsiirola · 2026-06-03T15:02:00Z

+    return opt
+
+
+def _setup(nlp):


Should setup also relax any discrete variables to continuous domains?

I added an exception for now. I want some time to think about what it would mean to allow models with discrete variables.

jsiirola · 2026-06-03T15:07:32Z

+    )
+
+
+def shallow_clone(m1):


In the future, we should promote this to a standard transformation. Not only does it make a fast shallow copy, but it collapses the entire block structure, which could be useful in a lot of contexts. It could generate more of a "flattened view" into a model than a "shallow copy"?

That is a great suggestion. I don't think "view" is the right term. What you get is a valid model. Flattened is very accurate, though...

Would this fit into the normal transformation workflow? apply_to wouldn't really make sense, and create_using might also not do what you expect. Maybe I'm thinkin about it wrong...

My initial thought would be to have apply_to raise an exception and only support create_using (similar to the dual transform). I would argue that "view" is correct, as when you solve the flattened view, the solution "appears" in the original structured model (because they share variables). If we go down the transformation route, there are questions around other components, though (e.g., should we map results for suffixes back?).

That is a good point. I thought that there was an expectation of create_using that a full clone would get created?

I guess that is where naming and documentation comes in.

The idea was apply_to is "in-place" and create_using is "out-of-place". There was concern that "in place" and "out of place" were less user friendly, and apply_to and create_using (short for "create a new model using data from model m") were more clear.

"Cloning" only came into play because many transformations (like GDP) are most efficiently written in-place, so the default implementation of create_using is simply to clone and then run apply_to...

Got it. That makes sense.

blnicho · 2026-06-03T17:55:33Z

One of the GHA jobs is randomly getting stuck so I'm going to merge this without waiting for the last check.

michaelbynum added 30 commits January 29, 2026 15:19

merge observer_fix into intialization

00b3c27

Merge branch 'pwl_approx' into initialization

d2aae9f

Merge branch 'pwl_factorable_aggressive_substitution' into initializa…

b2fa0a5

…tion

Merge branch 'observer_gurobi_refactor' into initialization

fd06964

starting a module for initialization

d29a2cf

starting a module for initialization

8914a63

initialization

5ebfdd4

pwl initialization is almost working

488a539

debugging

e993623

initialization

aa450e1

Merge branch 'scip_port' into initialization

4945388

Merge branch 'pwl_approx' into initialization

ab9a46c

Merge branch 'pwl_factorable' into initialization

6f30fac

some updates to initialization module

043f8c8

Merge branch 'scip_port' into initialization

83c5e72

Merge branch 'pwl_factorable' into initialization

8bf25d1

Merge branch 'scip_port' into initialization

4e3d913

initialize with global solvers

c8a0730

initialization work

e019bff

lp init should bound all variables

9568bb0

move initialization module from contrib to devel

387a646

update imports

24dbd64

initialization example

2616502

initialization: starting tests

54d354e

initialization: add copyright statements

056cc99

docstring

d70e677

initialization: docs

823641d

Merge branch 'pwl_approx' into initialization

6a29fba

Merge branch 'scip_port' into initialization

5eb638f

Merge branch 'pwl_factorable' into initialization

3ddc25d

michaelbynum and others added 12 commits June 1, 2026 16:38

update tests

be2516c

run black

cf411c5

dont do extra work if it is not needed

b990763

dont do extra work if it is not needed

6c79912

dont reinvent the wheel

bd89375

run black

a9eba73

dont hard-code things

02c20f0

dont reinvent the wheel

370da89

dont hard-code tolerances

825b322

better error message

895bd6c

run black

17e2a39

Update pyomo/devel/initialization/tests/test_initialization.py

ad49d34

Co-authored-by: Bethany Nicholson <blnicho@users.noreply.github.com>

michaelbynum added 2 commits June 2, 2026 08:41

update imports

6d50ff1

Merge remote-tracking branch 'michaelbynum/initialization' into initi…

c5af457

…alization

michaelbynum added 2 commits June 2, 2026 09:06

revert imports

8c2533b

update imports

e3cab9c

mrmundt requested changes Jun 2, 2026

View reviewed changes

michaelbynum and others added 3 commits June 2, 2026 16:49

Update pyomo/devel/initialization/pwl_init.py

5c69315

Co-authored-by: Miranda Mundt <55767766+mrmundt@users.noreply.github.com>

fix imports

418f7cf

Merge remote-tracking branch 'michaelbynum/initialization' into initi…

361d91f

…alization

blnicho approved these changes Jun 3, 2026

View reviewed changes

jsiirola requested changes Jun 3, 2026

View reviewed changes

be more careful

7c5d61f

jsiirola approved these changes Jun 3, 2026

View reviewed changes

mrmundt self-requested a review June 3, 2026 16:17

mrmundt approved these changes Jun 3, 2026

View reviewed changes

blnicho merged commit e404b97 into Pyomo:main Jun 3, 2026
55 of 58 checks passed



		def initialize_with_piecewise_linear_approximation(
		nlp: BlockData,

Uh oh!

Conversation

michaelbynum commented Apr 15, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Fixes #3878

Summary/Motivation:

Changes proposed in this PR:

Legal Acknowledgement

Uh oh!

michaelbynum commented Jun 2, 2026

Uh oh!

michaelbynum commented Jun 2, 2026

Uh oh!

mrmundt 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!

michaelbynum Jun 2, 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!

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!

jsiirola left a comment

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!

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!

Choose a reason for hiding this comment

Uh oh!

blnicho commented Jun 3, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

michaelbynum commented Apr 15, 2026 •

edited

Loading

michaelbynum Jun 2, 2026 •

edited

Loading