Extend Functionality to Use the Presolver Plugin and Add a Tutorial #1076

fvz185 · 2025-10-04T14:23:33Z

To clarify and document the usage of PySCIPOpt's presolver plugin, a working example and a tutorial were added. They required wrappers to be implemented, too. The wrapper functions are supplemented with tests. The example and tutorial replicate the Boundshift presolver.

However, some open issues remain:

The test for the var.isActive() wrapper is incomplete since it does not cover all possible settings. Should tests on inactive variables due to probing or aggregation be included?
Only SCIP functions that are not easily emulatable within Python and required for the example were wrapped. Should any additional SCIP functions be wrapped to make implementations of custom presolvers more straightforward? This would reduce the efforts needed from PySCIPOpt users to inspect SCIP's source code.

…vity

…aggregation

…s in the knapsack model

…s functionality

…for fixed and aggregated variables

…, and example shiftbound.py

codecov · 2025-10-04T14:29:25Z

Codecov Report

❌ Patch coverage is 75.00000% with 2 lines in your changes missing coverage. Please review.
✅ Project coverage is 54.70%. Comparing base (6452ea6) to head (5ce2777).
⚠️ Report is 24 commits behind head on master.

Files with missing lines	Patch %	Lines
src/pyscipopt/scip.pxi	75.00%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1076      +/-   ##
==========================================
+ Coverage   54.13%   54.70%   +0.56%     
==========================================
  Files          22       24       +2     
  Lines        5045     5327     +282     
==========================================
+ Hits         2731     2914     +183     
- Misses       2314     2413      +99

☔ 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.

Copilot

Pull Request Overview

This PR extends PySCIPOpt to support custom presolver plugins by adding wrapper functions, tests, a working example, and documentation. The key enhancement enables users to implement custom presolvers in Python by subclassing the Presol class.

Added isActive() method for variables to check if they are active (not fixed or aggregated)
Added aggregateVars() method to aggregate two variables with custom coefficients
Created a complete Shiftbound presolver example that replicates SCIP's boundshift.c logic

Reviewed Changes

Copilot reviewed 7 out of 7 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
src/pyscipopt/scip.pxi	Added `isActive()` and `aggregateVars()` wrapper methods for variables and models
src/pyscipopt/scip.pxd	Added C function declarations for `SCIPvarIsActive` and `SCIPaggregateVars`
tests/test_vars.py	Added basic test for `isActive()` method on newly created variables
tests/test_aggregate_vars.py	Added comprehensive tests for `aggregateVars()` functionality
examples/finished/shiftbound.py	Complete presolver example implementing boundshift logic
docs/tutorials/presolver.rst	Tutorial documentation explaining presolver plugin usage
CHANGELOG.md	Updated changelog with new features

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}

tests/test_vars.py

examples/finished/shiftbound.py

Co-authored-by: Copilot <[email protected]>

mmghannam

Thank you very much @fvz185 for the useful contribution! I believe this will make it much easier for people to prototype presolving techniques in python.
I've added a few small comments otherwise looks good to merge to me :)

mmghannam · 2025-10-07T07:20:47Z

docs/tutorials/presolver.rst

+               if val > 0 and val >= scip.infinity():
+                   return scip.infinity()
+               if var.vtype() != "CONTINUOUS":
+                   return scip.feasCeil(val)


I might be wrong, but shouldn't this be rounding to the nearest integer?

Yes, you are right. It's actually only adjustedLB() and forgot to include adjustedUB(). I guess that is a good example for why we want to wrap functions instead of copying their functionality.

mmghannam · 2025-10-07T07:22:17Z

docs/tutorials/presolver.rst

+
+           for var in reversed(vars):
+               if var.vtype() == "BINARY":
+                   continue


shouldn't this be an assert? I assume all binary variables are skipped from earlier.

docs/tutorials/presolver.rst

mmghannam · 2025-10-07T07:43:15Z

docs/tutorials/presolver.rst

+               )
+
+               # Aggregate old variable with new variable:
+               #   x  =  y + lb      (no flip), or


how about writing this as x - y = lb to be the same as what is passed to SCIPaggregateVars?

mmghannam · 2025-10-07T07:45:56Z

tests/test_aggregate_vars.py

+    assert not infeasible
+    assert aggregated
+    # model should stay consistent
+    model.optimize()


Could you please add assertion to check the bounds of the variables? also an assertion for the final status and the objective value of the model would be nice :)

mmghannam · 2025-10-07T07:49:52Z

src/pyscipopt/scip.pxi


+    def aggregateVars(self, Variable varx, Variable vary, scalarx=1.0, scalary=-1.0, rhs=0.0):
+        """
+        Aggregate varx and vary: calls SCIPaggregateVars and returns


Could you please add more info on what SCIPaggregateVars does? a formula would be very helpful here so that one wouldn't need to go the C-API documentation.

mmghannam · 2025-10-07T07:55:34Z

To also answer your questions:

The test for the var.isActive() wrapper is incomplete since it does not cover all possible settings. Should tests on inactive variables due to probing or aggregation be included?

I'm happy with the current test, but if you have the time yes, assertions in your presolver test on the aggregated variables would be nice.

Only SCIP functions that are not easily emulatable within Python and required for the example were wrapped. Should any additional SCIP functions be wrapped to make implementations of custom presolvers more straightforward? This would reduce the efforts needed from PySCIPOpt users to inspect SCIP's source code.

I don't have a lot of experience with presolvers but I would say one should wrap the methods when a use-case/feature request pops up. If you see anything that could be useful to "preemptively" wrap I'd be happy to review it in future PRs :D

fvz185 · 2025-10-07T15:15:27Z

I just realised that, while I was working on the documentation, the implementation of presol_boundshift.c had changed. Besides slight modifications to the sequence, it now no longer uses SCIPvarIsActive(). I'll check and align my documentation again but prioritise to make the presolver documentation/example self-contained/-explanatory.

I'll also take a look again on the tests to capture your suggestions.

Lastly, I think I will crawl through the other presolvers to see which SCIP functions are most frequently used and then wrap them (as part of future PRs). My goal is anyway to better grasp SCIP's source code, so it'll be a good exercise. Just let me know about any other feature that I could take a look at (ofc, I'll also take a look at the currently open issues).

Thank you @mmghannam for your feedback :)

Joao-Dionisio · 2025-11-17T11:46:09Z

Hey @fvz185 ! THank you so much for your contribution! We would love to include this in the next release, PySCIPOpt 7.0.0, which coincides with SCIP 10 (should be a couple more weeks, end of the month, start of next month). How are the TODOs looking, is there any way we can help out?

Joao-Dionisio · 2025-11-17T11:49:13Z

docs/tutorials/presolver.rst

+
+Here is a high-level flow:
+
+1. Subclass ``MyPresolver`` and capture any parameters in ``__init__``.


Suggested change

1. Subclass ``MyPresolver`` and capture any parameters in ``__init__``.

1. Create subclass ``MyPresolver`` and capture any parameters in ``__init__``.

(unless Subclass is a verb)

Joao-Dionisio · 2025-11-17T11:50:58Z

docs/tutorials/presolver.rst

+presolver that shifts variable bounds from ``[a, b]`` to ``[0, b - a]``
+and optionally flips signs to reduce constant offsets.
+
+For educational purposes, we keep our example as close as possible to SCIP's implementation, which can be found `here <https://scipopt.org/doc-5.0.1/html/presol__boundshift_8c_source.php>`__. However, one may implement Boundshift differently as SCIP's logic does not translate perfectly to Python. To avoid any confusion with the already implemented version of Boundshift, we will call our custom presolver *Shiftbound*.


Suggested change

For educational purposes, we keep our example as close as possible to SCIP's implementation, which can be found `here <https://scipopt.org/doc-5.0.1/html/presol__boundshift_8c_source.php>`__. However, one may implement Boundshift differently as SCIP's logic does not translate perfectly to Python. To avoid any confusion with the already implemented version of Boundshift, we will call our custom presolver *Shiftbound*.

For educational purposes, we keep our example as close as possible to SCIP's implementation, which can be found `here <https://scipopt.org/doc-5.0.1/html/presol__boundshift_8c_source.php>`__. However, one may implement Boundshift differently, as SCIP's logic does not translate perfectly to Python. To avoid any confusion with the already implemented version of Boundshift, we will call our custom presolver *Shiftbound*.

Joao-Dionisio · 2025-11-17T11:51:17Z

docs/tutorials/presolver.rst

+Presolvers
+###########
+
+For the following let us assume that a Model object is available, which is created as follows:


Suggested change

For the following let us assume that a Model object is available, which is created as follows:

For the following, let us assume that a Model object is available, which is created as follows:

Joao-Dionisio · 2025-11-17T11:52:34Z

docs/tutorials/presolver.rst

+
+A complete working example can be found in the directory:
+
+- ``examples/finished/shiftbound.py``


Maybe shiftbound_presolver.py or presol_shiftbound.py or some variation, to make it easier for users later on?

Joao-Dionisio · 2025-11-17T11:55:29Z

docs/tutorials/presolver.rst

+       def presolexec(self, nrounds, presoltiming):
+           scip = self.model
+
+           # Utility replacements for a few SCIP helpers which are not exposed to PySCIPOpt


Can you please interface these methods instead?
I can do it later as well, no problem

fvz185 added 14 commits October 3, 2025 20:38

Add SCIPvarIsActive function and corresponding test for variable acti…

70d39e2

…vity

Add SCIPaggregateVars function and aggregateVars method for variable …

9b69d6f

…aggregation

Add knapsack function for modeling the knapsack problem

5100b76

Add parameter settings to disable automatic presolvers and propagator…

fd41439

…s in the knapsack model

Add ShiftboundPresolver for variable domain transformation in SCIP

1c9ab3c

Add tests for Shiftbound presolver with parametrised knapsack instances

aa037f7

Update docstring in shiftbound.py to clarify presolver example and it…

94983e6

…s functionality

Add tests for Model.aggregateVars to verify aggregation functionality

908b47b

Add test for aggregation infeasibility in binary variables

e7d34cb

Remove Shiftbound presolver tests from test_shiftbound.py

7560e5f

Refactor TODO comment in test_isActive to clarify missing test cases …

0565bcb

…for fixed and aggregated variables

Update CHANGELOG to include new features: isActive(), aggregateVars()…

2f02bfa

…, and example shiftbound.py

Add tutorial for writing a custom presolver using PySCIPOpt

b3a8d0a

Add tutorial for presolver plugin to CHANGELOG

5ce2777

mmghannam self-assigned this Oct 4, 2025

Joao-Dionisio requested a review from Copilot October 4, 2025 23:57

Copilot AI reviewed Oct 4, 2025

View reviewed changes

tests/test_vars.py Show resolved Hide resolved

examples/finished/shiftbound.py Outdated Show resolved Hide resolved

examples/finished/shiftbound.py Outdated Show resolved Hide resolved

Apply suggestions from code review

3b2ac0c

Co-authored-by: Copilot <[email protected]>

mmghannam requested changes Oct 7, 2025

View reviewed changes

mmghannam reviewed Oct 7, 2025

View reviewed changes

Merge branch 'master' into feature/extend-presolver-plugin

62e9b03

Joao-Dionisio reviewed Nov 17, 2025

View reviewed changes


		Here is a high-level flow:

		1. Subclass ``MyPresolver`` and capture any parameters in ``__init__``.

	1. Subclass ``MyPresolver`` and capture any parameters in ``__init__``.
	1. Create subclass ``MyPresolver`` and capture any parameters in ``__init__``.

	For the following let us assume that a Model object is available, which is created as follows:
	For the following, let us assume that a Model object is available, which is created as follows:


		A complete working example can be found in the directory:

		- ``examples/finished/shiftbound.py``

Extend Functionality to Use the Presolver Plugin and Add a Tutorial #1076

Are you sure you want to change the base?

Extend Functionality to Use the Presolver Plugin and Add a Tutorial #1076

Uh oh!

Conversation

fvz185 commented Oct 4, 2025

Uh oh!

codecov bot commented Oct 4, 2025

Codecov Report

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull Request Overview

Reviewed Changes

Uh oh!

Uh oh!

Uh oh!

Uh oh!

mmghannam 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!

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!

mmghannam commented Oct 7, 2025

Uh oh!

fvz185 commented Oct 7, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Joao-Dionisio commented Nov 17, 2025

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!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

fvz185 commented Oct 7, 2025 •

edited

Loading