Originally posted by: 81XJ_Jakob_Kaths

Hello,

I'm using CPLEX to solve a MIQP problem in order to calculate control inputs for a model predictive control algorithm. Now I try to speed up my code and I was wondering if I could hand over a start point in each iteration as I always have a good guess for that because of my reciding control horizon. Is that possible using the Python API?

Thanks!

I am not clear what you mean by "iteration".

Using cplex.MIP_Starts.add() you can provide an initial integer feasible solution for the MIP search.

If you want to provide a starting point for the algorithm that solves the node relaxation then you will have to use a solve callback.

Originally posted by: 81XJ_Jakob_Kaths

Thanks, I will try that. By iteration I just mean that I recalculate the solution in a certain interval. But for each interval I already have a good guess that I want to provide as a start value.

Originally posted by: 81XJ_Jakob_Kaths

Sorry, one more question. For the moment I want to use this for MIQP. Is it as well working for LP, MILP, QP, (MI)(QC)QP that way? Thanks in advance.

cplex.MIP_Starts is for everything that involves integral variables (and/or SOS constraints). For pure LP, QP, QCP, use cplex.start.