英文:
Why CPLEX optimizer does npt stop after the limits?
问题
I have a model that I am sure the body of the model won't take like to get build up and once I call JuMp.optimize it takes very long to stop. Although I set 1000 limits.
model = Model(optimizer_with_attributes(
CPLEX.Optimizer,
"CPX_PARAM_TILIM" => 1000,
"CPX_PARAM_EPGAP" => 1e-1)) # relative gap
set_optimizer_attribute(model, CPLEX.PassNames(), true)
It's now more than three hours and still no stopping. What is the possible issue and how to fix? I repeated and checked all things three times to make sure that there is no bug somewhere, and have no other idea to try. Any help would be appreciated!
英文:
I have a model that I am sure the body of the model won't take like to get build up and once I call JuMp.optimize it takes very long to stop. Although I set 1000 limits.
model = Model(optimizer_with_attributes(
CPLEX.Optimizer,
"CPX_PARAM_TILIM" => 1000,
"CPX_PARAM_EPGAP" => 1e-1)) # relative gap
set_optimizer_attribute(model, CPLEX.PassNames(), true)
It's now more than three hours and still no stopping. What is the possible issue and how to fix? I repeated and checked all things three time to make sure that there is no bug somewhere, and have no other idea to try. Any help would be appreciated!
答案1
得分: 2
Try setting CPXPARAM_TimeLimit instead:
https://www.ibm.com/docs/en/icos/22.1.1?topic=parameters-optimizer-time-limit-in-seconds
CPX_PARAM_TILIM was deprecated a few versions ago.
It's now more than three hours and still no stopping.
Do you have the log? What is CPLEX doing for all that time?
If that doesn't work, then no suggestions. You could contact IBM for support, but this is likely an issue in CPLEX, not in JuMP or CPLEX.jl.
英文:
Try setting CPXPARAM_TimeLimit instead:
https://www.ibm.com/docs/en/icos/22.1.1?topic=parameters-optimizer-time-limit-in-seconds
CPX_PARAM_TILIM was deprecated a few versions ago.
> It's now more than three hours and still no stopping.
Do you have the log? What is CPLEX doing for all that time?
If that doesn't work, then no suggestions. You could contact IBM for support, but this is likely an issue in CPLEX, not in JuMP or CPLEX.jl.
通过集体智慧和协作来改善编程学习和解决问题的方式。致力于成为全球开发者共同参与的知识库，让每个人都能够通过互相帮助和分享经验来进步。
评论