Seeking Guidance on Gurobi Optimization for Supply Chain Modeling

Awaiting user input

Quin Ackerman

• July 11, 2024 11:09

Hello everyone, :)

I am working on a project involving supply chain optimization using Gurobi, and I could use some guidance on a few points:

1. I am struggling with incorporating complex constraints related to production capacities and demand variability. Any tips on structuring these in Gurobi?
2.  What are some effective strategies to improve solver performance, especially when dealing with large datasets and complex models?
3.  How can I efficiently integrate real-time data feeds into my Gurobi model to improve decision-making accuracy?

 Can anyone help me tackle these challenges effectively. Thank you in advance for your assistance!

0

• 

  Ronald van der Velden

  Gurobi Staff

  • July 15, 2024 07:13

  Hi Quin!

  These are interesting, but also broad questions :-) I'll just share the first that comes to mind, but others may have additional ideas and you may want to add details to guide us in the right direction.

  1. There are two important steps here - first translating your rules into mathematical equations, second implementing those equations using the Gurobi API. In which of those two steps do you need help? if it's the translation into equations, can you share more details about the business rules? If it's the implementation, can you share the equations?
  2. There are many, but ideally you would first have a working model and then let the logfiles guide you towards the "pain points". Addressing performance issues too early may be wasted time, since future changes to the model may completely invalidate your performance work. You may find our webinars on performance tuning and model strengthening interesting too.
  3. I guess the question is around efficiently re-optimizing an existing model after retrieving updated data right? The most important things you can do are (a) keep the model object in-memory and update it, instead of completely building it from scratch everytime - you may not only save time building the model, but Gurobi may reuse previous results too (b) if that is not possible, you may provide variable hints / start values based on the previous iteration.

  Kind regards,
  Ronald

  0

Please sign in to leave a comment.