Hi Olivier,
Sorry, the model I described contains commercially sensitive data that we are not allowed to share. Without those data, it's not possible to run and recreate the issue. I'm wondering if you know, in general, what might be the cause of this kind of issue or if you are aware of any other case like this? Thank you.
Best,
------------------------------
Boning Yang
------------------------------
Original Message:
Sent: Thu July 06, 2023 04:02 AM
From: Olivier Lhomme
Subject: Reproducibility Issue with CPLEX 22.1.0 and Newer
Hello,
can you send a model showing the problem?
------------------------------
Olivier Lhomme
Original Message:
Sent: Sun July 02, 2023 11:00 AM
From: Boning Yang
Subject: Reproducibility Issue with CPLEX 22.1.0 and Newer
Hi All,
My research group has been using CPLEX as the optimizer for our simulation model for complex systems. Our previous experience with the older version of CPLEX, such as 12.9 and 12.10, shows that the model using CPLEX as the optimizer can produce the same result on the same machine each time as long as the parameters and input data are the same. However, as we try to set up the model using a newer version of CPLEX, namely 22.1.0, we noticed that the model can no longer produce the same result with the same input with a bigger (having more terms in the objective functions etc.) model, although with a smaller model, the results are still reproducible. The difference is rather small at the beginning, it accumulated to a more significant difference when the model became more complex and ultimately leads to completely different results. This issue challenges the reproducibility of our research and we are wondering if this is a known issue with the newer version of CPLEX or if there is a new functionality that we don't know of that may be causing this issue. Thank you.
Best,
------------------------------
Boning Yang
------------------------------