Good Morning,
My query is why when I run 2 equal models with the same data (one is from excel and the other is from mssql) they result in very different objective solutions. Is there a technical reason for that or is it necessarily because of the data?

Solution from excel (note : Solution relaxing....why?)


------------------------------
MIGUEL FLORES QUISPE
------------------------------

#DecisionOptimization

Are you certain that both model and data are the same? Do both models have the same dimensions (number of variables, number of constraints) before presolving?

------------------------------
Paul Rubin
Professor Emeritus
Michigan State University
------------------------------

Thanks for reply

They only differ in the number of variables because excel allows the same index variable to be used (correlative) for all arrays, while in the mssql database I must use different index variables for each array. Then the model is the same.
I understand that this should not affect the result.

What does it mean that sometimes "Relaxed Solution" and sometimes "Solution with Objective xxxxxxx" return? I am new at this. Thank you

------------------------------
MIGUEL FLORES QUISPE
------------------------------

Original Message:
Sent: Thu June 25, 2020 04:17 PM
From: Paul Rubin
Subject: Objective solution

Are you certain that both model and data are the same? Do both models have the same dimensions (number of variables, number of constraints) before presolving?

------------------------------
Paul Rubin
Professor Emeritus
Michigan State University
------------------------------

This almost certainly means that your models are not as similar as you think. If you get a "relaxed solution" then your model is actually infeasible and what you get is a minimal relaxation that makes your model feasible. Either this is due to very bad numerics or because your models are not identical. I'm afraid you have to track down where your models differ. One way to do that would be to take the solution from the feasible model and check which constraints it violates in the infeasible model. Also check the objective value of that solution in the infeasible model. Is it even the same?

------------------------------
Daniel Junglas
------------------------------

Original Message:
Sent: Thu June 25, 2020 05:39 PM
From: MIGUEL FLORES QUISPE
Subject: Objective solution

Thanks for reply

They only differ in the number of variables because excel allows the same index variable to be used (correlative) for all arrays, while in the mssql database I must use different index variables for each array. Then the model is the same.
I understand that this should not affect the result.

What does it mean that sometimes "Relaxed Solution" and sometimes "Solution with Objective xxxxxxx" return? I am new at this. Thank you

------------------------------
MIGUEL FLORES QUISPE
------------------------------


Original Message:
Sent: Thu June 25, 2020 04:17 PM
From: Paul Rubin
Subject: Objective solution

Are you certain that both model and data are the same? Do both models have the same dimensions (number of variables, number of constraints) before presolving?

------------------------------
Paul Rubin
Professor Emeritus
Michigan State University

Hi,

on top of what Daniel suggested, you can also export from within the studio the internal data and external data in flat files that you could compare.
Then you could see the differences that you have in the data wen you use excel and mysql

regards




------------------------------
ALEX FLEISCHER
------------------------------

Original Message:
Sent: Fri June 26, 2020 12:18 AM
From: Daniel Junglas
Subject: Objective solution

This almost certainly means that your models are not as similar as you think. If you get a "relaxed solution" then your model is actually infeasible and what you get is a minimal relaxation that makes your model feasible. Either this is due to very bad numerics or because your models are not identical. I'm afraid you have to track down where your models differ. One way to do that would be to take the solution from the feasible model and check which constraints it violates in the infeasible model. Also check the objective value of that solution in the infeasible model. Is it even the same?

------------------------------
Daniel Junglas
------------------------------


Original Message:
Sent: Thu June 25, 2020 05:39 PM
From: MIGUEL FLORES QUISPE
Subject: Objective solution

Thanks for reply

They only differ in the number of variables because excel allows the same index variable to be used (correlative) for all arrays, while in the mssql database I must use different index variables for each array. Then the model is the same.
I understand that this should not affect the result.

What does it mean that sometimes "Relaxed Solution" and sometimes "Solution with Objective xxxxxxx" return? I am new at this. Thank you

------------------------------
MIGUEL FLORES QUISPE


Original Message:
Sent: Thu June 25, 2020 04:17 PM
From: Paul Rubin
Subject: Objective solution

Are you certain that both model and data are the same? Do both models have the same dimensions (number of variables, number of constraints) before presolving?

------------------------------
Paul Rubin
Professor Emeritus
Michigan State University