about cplex

1) not a good idea to add constraint in order to introduce upper cutoff or lower cutoff, in lp, it is possible to define the lower or upper objective, but in MIP, it is impossible which is very strange (should be possible in callable libraries). One way is to restrict solution limit to 1, and add a constraint to the model, this has been proved very inefficient, since it cause much more trouble to B&C.

2) solution limit denote the better solution, not any solution that has been found. if the same objective value is found , it is not counted.  If presolve gives a better solution based on the input ini solution, it is counted in the solution limit.  One strange thing is presolve gives the same objective value, but it is counted in the solution limit. A possible explanation  is that the improved solution is just slightly better, it is still counted in the solution limit, but no difference is shown on the screen, still under investigation.

3) cplex 10 withdraw the function to allow user to store the B&C tree to a file, and restart later on. But one can still use the initial solution from AMPL to enforce cplex to start from a possible solution.

4) segmentation fault happens when running ampl from outside for some scenarios, like matlab or in command line, still wondering why.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: