It seems that the barrier algorithm of Gurobi is not fully exploiting the sparsity and structure of the sparse QP formulation. Does someone have a reason for that?
In Model Predictive Control, the computational complexity should scale linerarly with the prediction horizon N, i.e. the algorithm needs O(N(#states+#inputs)^3) operations. However, results show that the complexity scales quadratically with the prediction horizon N.
The computation time is measured as the time needed to solve one QP problem.
Please sign in to leave a comment.