Jaromił Najman

Hi Ruibing,I apologize for not coming back to you for so long. We figured out that the issue is numerical. Due to feasibility and optimality tolerances, it may happen that some node relaxation may ...

Hi Ruibing,as of version 13, the convexity recognition of Gurobi for nonlinear terms is very weak. Thus, despite that your model is clearly convex, Gurobi reports that it is non-convex. We are work...

Hi Ruibing,In addition to what Riley already stated, we are looking at the latest model you provided and trying to understand why Gurobi reports a different global optimum depending on the seed.I w...

Hi Ruibing,The latest ru model you provided also behaves well, both with version 12 and 13. Again I would recommend sticking with v13. Again, I tested both versions over 5 seeds and all runs proved...

Hi Ruibing,The latest model you provided (the ro case) behaves well, both with version 12 and 13 (I would recommend sticking with v13 to profit from latest bug fixes and performance improvements). ...

Hi Ruibing,Thank you for sharing the reformulated model. It behaves way better. However, I am afraid that there is a bug or numerically troublesome behavior showing up on this model. I will need to...

Hi Ruibing,Thank you for sharing. This is very helpful. I see something that might be problematic for a nonlinear solver.You have terms of the formlog(y / (constant + sum a_i x_i ))where \(y\) and ...

Which solution should I trust? Which is trully global optimal?It is hard to say without the model at hand. It might be a bug in v13, a bug in v12, or even a bug in AMPL.Could you please extract the...

I still do not fully understand your answer to my second question, so let me clarify what confuses me. My objective is to maximize revenue. After I added an additional constraint to Model B, the ob...

I tried Model B with Gurobi by avoiding zero denominators and get a solution from Gurobi. Gurobi output is below. I am writing to ask if the solution is a valid global optimal solution. Thank you v...