Hi,
I have an integer programming model to solve with CPLEX 10.1 Callable Library. I know that Cplex wastes lots of time to prove optimality of a solution. When i put a time limit to Cplex and solve the model, it gives optimal solution within time limit or hits time and gives integer feasible solution. In the second case, we can think that Cplex may find a good integer feasible solution using less time and it struggles to prove optimality by turning around the same integer feasible solution or make little deviation from the feasible solution for additional time. I want to learn how behave objective function with increasing time for the second case!
I want to obtain first integer feasible solution (of course, not trivial solution) or other feasible integer solutions within the time limit. Because additional time may not be meaningful to impose for finding better solution within high computational time limit.
Is there any suggestion, solution pooling or stg by using Cplex Callable Library?
thanks
melike
Topic

Re: finding first integer feasible solution to integer programming model
20110105T20:16:08ZThis is the accepted answer. This is the accepted answer.Not sure what you are trying to do. Do you just want to stop at the first feasible solution? Then set parameter CPX_PARAM_INTSOLLIM to 1.
Or do you want to do something else? 
Re: finding first integer feasible solution to integer programming model
20110106T07:01:47ZThis is the accepted answer. This is the accepted answer. SystemAdmin
 20110105T20:16:08Z
Not sure what you are trying to do. Do you just want to stop at the first feasible solution? Then set parameter CPX_PARAM_INTSOLLIM to 1.
Or do you want to do something else?
Also, whether i can understand the behaviour of objective function versus time. I mean in the case that the model can't reach optimal solution,i want to keep feasible integer solutions within the time limit. If it is possible!
Cplex may find a good solution within less time and make little improvement on objective function by adding high computational time. Wasting lots of time may be meaningless in this situation.
Thanks
melike 
Re: finding first integer feasible solution to integer programming model
20110106T09:30:21ZThis is the accepted answer. This is the accepted answer. melike
 20110106T07:01:47Z
Yes, I just want to stop at the first feasible solution!
Also, whether i can understand the behaviour of objective function versus time. I mean in the case that the model can't reach optimal solution,i want to keep feasible integer solutions within the time limit. If it is possible!
Cplex may find a good solution within less time and make little improvement on objective function by adding high computational time. Wasting lots of time may be meaningless in this situation.
Thanks
melike
In order to store or query all the solutions CPLEX has found so far use the solution pool (see manual or reference documentation). 
Re: finding first integer feasible solution to integer programming model
20110705T18:07:24ZThis is the accepted answer. This is the accepted answer. SystemAdmin
 20110106T09:30:21Z
In order to stop at the first feasible solution that CPLEX finds just set CPX_PARAM_INTSOLLIM to 1. Note that once stopped you can continue the solve to stop at the next solution if there is time left. I.e. you can invoke CPXmipopt() in a loop, always stopping at the next integral solution found until you reach some kind of a time limit.
In order to store or query all the solutions CPLEX has found so far use the solution pool (see manual or reference documentation).
I have a question,
does IntSolLim() gives the number of times Cplex finds an integer solution or the number of times it updates the incumbent solution ?
thanx
nitish 
Re: finding first integer feasible solution to integer programming model
20110706T06:14:43ZThis is the accepted answer. This is the accepted answer. NitishGargIITM
 20110705T18:07:24Z
Hi,
I have a question,
does IntSolLim() gives the number of times Cplex finds an integer solution or the number of times it updates the incumbent solution ?
thanx
nitish