Скачать или смотреть Why Pyomo is Assigning Floats to the Integer Domain: Debugging Your Model

Understand how to solve feasibility issues in `Pyomo` and ensure your integer constraints are respected during optimization.
---
This video is based on the question https://stackoverflow.com/q/71163026/ asked by the user 'frunk' ( https://stackoverflow.com/u/18190284/ ) and on the answer https://stackoverflow.com/a/71164718/ provided by the user 'jsiirola' ( https://stackoverflow.com/u/6749874/ ) at 'Stack Overflow' website. Thanks to these great users and Stackexchange community for their contributions.

Visit these links for original content and any more details, such as alternate solutions, latest updates/developments on topic, comments, revision history etc. For example, the original title of the Question was: Pyomo assignes floats to Integer domain

Also, Content (except music) licensed under CC BY-SA https://meta.stackexchange.com/help/l...
The original Question post is licensed under the 'CC BY-SA 4.0' ( https://creativecommons.org/licenses/... ) license, and the original Answer post is licensed under the 'CC BY-SA 4.0' ( https://creativecommons.org/licenses/... ) license.

If anything seems off to you, please feel free to write me at vlogize [AT] gmail [DOT] com.
---
Understanding the Problem

Are you facing an issue with Pyomo where the solver is assigning float values to variables that should strictly be integers? In this guide, we'll explore a specific example of dealing with this problem and help you understand what's going wrong and how to fix it.

The Scenario

You have a simple model where you want to assign unique instance IDs across an instance_map while ensuring that each ID occurs exactly once. However, when running your optimization problem using ipopt, you get a warning indicating that the problem is infeasible, along with unexpected float values in an integer domain.

Your initial code looks something like this:

[[See Video to Reveal this Text or Code Snippet]]

On executing this code, you encounter a peculiar output resulting in float values instead of integers. Let's dive into why this happens and how to address it.

The Solution

Understanding the Solver's Behavior

Infeasibility Issue:

The message "termination condition: infeasible" indicates that the solver cannot find a viable solution that satisfies all constraints. This is compounded by what ipopt is designed to do.

Continuous vs. Discrete Variables:

The ipopt solver is primarily a continuous interior point solver. According to the nature of optimization, this means it treats integer variables as continuous unless instructed otherwise.

Look out for warnings like "Treating binary and integer variables as continuous." When those messages appear, it implies that your solver is not properly handling the integer constraints.

Steps to Resolve the Issue

Step 1: Change the Solver

Consider using a solver that is designed for discrete optimization problems that natively handles integer constraints, such as:

glpk: A GNU Linear Programming Kit.

cbc: Coin-or branch and cut.

Example of how to call a different solver:

[[See Video to Reveal this Text or Code Snippet]]

Step 2: Check Your Constraint Definitions

Ensure that your constraints are formulated correctly. They should explicitly restrict the instances_map variable to meet the unique instance requirements.

Re-evaluate the logic of your unique_instances_check function, ensuring that it adequately captures the unique assignment condition.

Step 3: Add Logging Information

To debug effectively, enable the logging information of the solver:

[[See Video to Reveal this Text or Code Snippet]]

This way, you can track how the solver proceeds and pinpoint where it may be running into infeasibility.

Conclusion

By understanding the limitations of the ipopt solver with respect to integer constraints and choosing a solver that aligns better with your optimization needs, you can effectively resolve the issue of floats appearing where integers should be. Always ensure your constraints are coherent and monitor solver logs for insights.

The key takeaway here is to ensure you're using the right tools for your optimization task in Pyomo. With the correct approach, your integer assignments can be accurately reflected in the optimization results.