Originally posted by: PingLiu

I run cplexmilp function in matlab, and the objective value is 145.4595

[x,fval,exitflag,output] = cplexmilp(f,Aineq,bineq,Aeq,beq,[],[],[],lb,ub,ctype,[],[]);

however, if I run mipopt command in cmd with the .lp file created for the same problem
then the objective value is 152.748656023068
why is there such difference?
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: John Cui

Are you sure 21-Dec-2011.lp is same as the model generated by .mat file? And could I know how did you generate 21-Dec-2011.lp?
I found 21-Dec-2011.lp has 3026 decision variables, but the other only have 3024 decision variables.

John Cui
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: PingLiu

I write a .m file by myself,and got the 21-Dec-2011.lp file, which is a little different from that generated by using

cplex.writeModel(FileName)

. but both of them are got with the same parameters*(f,Aineq,bineq,Aeq,beq,lb,ub,ctype,-1)*
actually, there are 178 fixed variables in 21-Dec-2011.lp.
so after running

cplex.readModel(FileName);
cplex.solve();
x = cplex.Solution.x;

we got only 3026 decision variables in x matrix.

but if we read the other .lp file generated by cplex.writeModel() even though the same 178 variables are fixed, when we run the same code

cplex.readModel(FileName);
cplex.solve();
x = cplex.Solution.x;

we can get all of the value of x with 3204 size.
I don't know the real reason for this difference, could you explain to me?
BTW, the index of x is not the same as that of variables, so how can we get the index of variables in x matrix?
thanks very much.
#CPLEXOptimizers
#DecisionOptimization