Originally posted by: gyhuang

Hi

I am using CPLEX+JAVA.

In my problem, I have to run LP multiple times, while the solution for each iteration is very close to another one. Therefore, it seems natural to use the solution of previous LP as the initial starting point of current LP.

I find the following instruction:
http://publib.boulder.ibm.com/infocenter/cosinfoc/v12r2/index.jsp
It says: In the Java API, the new examples are SchedGoalProg.java and PlantLocation.java.
However, I dont know where to find these two java files for examples.

My problem is also similar as below, however, that post seems havent got a solution yet.
http://www.ibm.com/developerworks/forums/thread.jspa?messageID=14426909&tstart=0

Thank you.
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

The examples you mentioned are not among the ones that ship with CPLEX, and I don't know where you can find them.

What I think you want, however, if different from the thread you cite, where the user was specifically trying to hot start the barrier solver with a heuristically-generated solution (if I read the thread correctly). What you want (I think) is a standard hot start -- solve an LP, make some changes, and try to use the final basis of the previous problem as a starting point for a basis for the new problem. CPLEX does this automatically as long as the advanced start switch (parameter AdvInd) is set to a value of 1 or 2, and 1 is the default. So unless you deliberately turn it off, CPLEX will try to hot start each LP from the previous solution.

/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: SystemAdmin

What are the modifications you apply to the LP between two solves? Unless you disable the AdvInd parameter explicitly CPLEX already attempts to reuse the old solution if possible.
In order to explicitly specify a starting point IloCplex.setVectors() or a combination of getBasisStatus()/setBasisStatus() may help.
#CPLEXOptimizers
#DecisionOptimization