Originally posted by: Mahmoud Fayek

Originally posted by: PhilippeLaborie

Hello, Could you say what you think is wrong with the solution:

Do you see some constraints that are formulated in the model and not satisfied in the solution?

Does the engine claim that the provided solution is optimal whereas you think you can find a better solution?

Note that if you change a model (even with some minor changes and even if the previous solution is still valid for the changed model), there is no guarantee that the engine finds the same solution.

Originally posted by: Mahmoud Fayek

Thank you Mr. Philip for your reply.

I edited the code in a way similar to that machines are personal tools to each worker so I can get the same scheduling result to check the edited code.

The original scheduling solution is more compact. that's why I think it's more optimum than the edited model's solution as the objective is to minimize tardiness.

I am really grateful for your support.

Originally posted by: Mahmoud Fayek

I have shared the complete .dat and .mod files of my model seeking help, after a long time, a lot of questions, and search trying to figure it out by myself.

I would be more than grateful if you can take a look.
https://www.ibm.com/developerworks/community/forums/html/topic?id=83707767-1e4b-4168-a431-4a3bf400b234

Thanks again for your consideration and generous help.

Respectfully,
Fayek

Originally posted by: Mahmoud Fayek

I have found my mistake. 
The objective function I was using has more than one optimum solution. Once I changed to another one with only one optimum solution like:
              " minimize max(ta in tasks) endOf(titasks[ta]) "
I got the same solution.