Originally posted by: fanpeng2

I would like to plot an "incumbent objective value v.s. time" curve in real time to observe the convergence. One way is to read the log file frequently and translate all lines started with "*". However, it seems to be an awkward detour. Is there a better way to do it?
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

You could use an incumbent callback to update the plot.

/Paul

Mathematicians are like Frenchmen: whenever you say something to them, they translate it into their own language, and at once it is something entirely different. (Goethe)
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: fanpeng2

The problem is that I cannot use incumbent callback, as mentioned in this link:

https://www.ibm.com/developerworks/forums/thread.jspa?threadID=326975&tstart=0
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

You can query the incumbent solution from an Info Callback as well, and Info Callbacks are compatible with dynamic search. The Incumbent Callback additionally allows you to reject a new incumbent solution, which is not possible from an Info Callback. If the only decision you wish to make based on new incumbents is either to continue the optimization or terminate it, an Info Callback will suffice.

Philip Starhill
CPLEX Research Engineer
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: fanpeng2

Can I use info callback in polish? I tried that but it is not called frequently. Sometimes it is not called more than one minute after a new incumbent is found.
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

Unfortunately, the info callback is (and can) only be called at the synchronization points in deterministic parallel mode. Otherwise, we cannot guarantee that the info callback is called in a deterministic fashion. In the case of polishing, we would typically synchronize after every node (i.e. each thread has processed one node), but a single polishing node can take quite a while and can produce a lot of new incumbents.
#CPLEXOptimizers
#DecisionOptimization