Originally posted by: BabakArh

Hi everyone 

I have MIP model (lets call it ABC),

ABC: MIN cx

          ax<=b

         x>=0 and integer 

I have a optimal solution for ABC which took long time to be obtained.  I am about to use LP solution to heuristically find near optimal solution. therefore i need solution for LP model (LPABC).

LPABC :MIn cx

          ax<=b

         x >=0

I read some relative topics and based on them I wrote this syntax in .NET :

mipModel.SetParam(Cplex.LongParam.NodeLim, 0); // stop after first node
            mipModel.Solve();
            LPrelaxationSolution = mipModel.ObjValue;

this is not giving me LP relaxation solution and in the console the machine writes something about number of cuts (the picture is attached). also the solution it gives I think that is best MIP solution in root (node=0) not lp solution. 

if you see the picture my objective in first node is -0.0124 which i think that is MIP solution because it is under column with name of "Best Integer Solution" and in the root the CPLEX give this value as objective. also i think there is a column with name of "Objective" which have a value -0.0146 which I think that is LP solution. how can get this value? 

Originally posted by: Anjikum

Hello,

You can just Solve LP model of the same to get the LP Optimal or else in the current model you can change Variable type using "Iconversion" to convert from int to float , Use Cplex.Solve() to get LP Optimal and then convert variable type to int and you can work on MIP model.

Originally posted by: BabakArh

I dont think that is a proper way. as you see in the picture the model already has the LP objective and i think there is a way to get that value. changing the var type is last option :D I am so lazy 

The best bound is the objective value of the LP relaxation plus cuts that CPLEX adds, some of which are based on the integrality requirements of the MIP model. You can see a figure of 54 cuts in the cuts/best bound column, after which the lower bound matched the best integer value (0.0124) and you had a provably optimal solution.

If you really want the solution to LPABC (with no added cuts), then Anjikum is correct.

The dual bound after the root (which is the LP relaxation plus cuts, as Paul mentioned) is obtained from the GetBestObjValue() method, not the ObjValue property.