Decision Optimization

Hello, 

I have a model used to schedule several jobs that I declare in my .mod file as

int rJ[J] = ...;

where

range J      = 1..4;

and I then define values in my .dat file as

rJ = [1, 2, 3, 4, ]; 


I use rJ to in my model also to create sets and for constraints.

I noticed that if I change the order of the values in  rJ I get the same objective value but the solution changes and so does the computational time sometimes up to 50%.

Is this to be expected?? If so, is there any way that could help me determine which initial combination will give me the result with the minimum amount of time?

Thanks you

Sam

Hi,

you should test with several random seeds, either with the tool or through scripting https://github.com/AlexFleischerParis/howtowithopl/blob/master/randomseedvariability.mod

What you describe sounds normal.

regards

Hello, 

I have a model used to schedule several jobs that I declare in my .mod file as

int rJ[J] = ...;

where

range J      = 1..4;

and I then define values in my .dat file as

rJ = [1, 2, 3, 4, ]; 


I use rJ to in my model also to create sets and for constraints.

I noticed that if I change the order of the values in  rJ I get the same objective value but the solution changes and so does the computational time sometimes up to 50%.

Is this to be expected?? If so, is there any way that could help me determine which initial combination will give me the result with the minimum amount of time?

Thanks you

Sam