Example:
The objective value in five minutes is 5000 (with gap of 50%)
in 8 hours the objective value is 5000n(with gap of 40) (equal than in five minutes)
in 48 hours the objective value is 5000 (with gap of 35)
in seven days , the solution is reached ,the solution is 5000 (With GAP of 0%)
CPLEX has several parameters that users can adjust, in the hope of solving specific instances more quickly. They are currently listed here:
http://www-01.ibm.com/support/docview.wss?uid=swg21400023
In your situation, the first thing I would do is play with the "MIP Emphasis" parameter.
The pattern you describe is very common in solving IPs with branch-and-bound and its variants. If you need a provably optimal solution, you have to wait - sometimes indefinitely. What I often do to determine if there was a chance for finding a true optimum in reasonable time is to map the number of active (dangling) nodes and the total number of nodes created by CPLEX over time. If the two functions are still divergent even after a few days of runtime, it is highly unlikely that CPLEX would ever confirm optimality of the incumbent.
as what I did. Set the stop criteria such as time limit, or expected OBJ...
See more techniques in:
https://www.sciencedirect.com/science/article/pii/S1876735413000020
Attached, you can see that I have the same problem in all single modeling with CPLEX. I know I can limit time. In the sample, I can guess that the optimality is almost reached. But, CPLEX is still running. What I am curious about is to have control of the number of the iterations that CPLEX repeats the same gap value. For instance, to force CPLEX to stop after 100 times it repeats the same value on the GAP.
Vahid Ghomi did you solve your problem? I have been running MILP model, but for hours it is running at the same gap.
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Publication