So far our understanding has been that no information is stored between two runs fired up from the GAMS IDE.
If this is true, every symbol should be initialized in the same way, when comparing two runs.
But, we have been able to reproduce a weird behaviour that challenges this understanding:
Run 1: Uses Solver A, finishes at a solution (let's say S1)
Run 2: Uses Solver B, finishes at S1
Run 3: Uses Solver A, finishes at S1
Run 4: Uses Solver B --> Finishes at a different solution.
This are all runs started from the IDE using the GUI button, not different solve statements in the same file. Nothing was changed except the solver via an option statement. We are trying to pinpoint if the different solution at which run 4 finishes is caused by the internal solver process or because some information is stored between runs while GAMSIDE remains open.
We are now in the process of testing how this changes when closing GAMSIDE between run 3 and 4, but the solve process is lengthy so I thought I could ask here for confirmation / refutation of the fact that any information is stored between runs.
Best regards and thanks!
questions about GAMS' tools
1 post • Page 1 of 1