Originally posted by: R06J_Sohaib_AFIFI

Hello, 

I had  a strange MIP start behavior I hope you can explain it to me.

Working with C++ concert, I'm doing a for loop where for every iteration I do exactly the same thing : I create my env,model, cplex (locally) and read a mip start from a vector. solve and than end the env.

The problem is that cplex consider my mip start as valid just in the first iteration ! I even tried to export the model and the mip start and executed conflict in the interactive solver and it consider it valid all the time.

Is there any problem with my tests ?

Could you please attach log files for the first (the one that accepts the MIP start) and second iteration (the one that rejects the MIP start) of your loop? That could help to analyze what is going on.

Also, are you sure that the vector from which you read the MIP start is still the same in the second loop? Maybe print it out to be extra sure.

Originally posted by: R06J_Sohaib_AFIFI

Thank you for your reply

The log file is attached. the vector is the same. 

For the mip start I'm providing just the boolean variables with the value true. 

This looks indeed quite strange. What version of CPLEX do you use? What happens if you write out the MIP start to an MST file right before calling solve()? Are the two MST files exactly the same?