Abstract
In this thesis we want to present a way to incorporate multi-point constraints into a given finite elements system by using a projection approach. The solving method we want to incorporate this in will be a conjugate gradient (CG) algorithm. For this we first want to expalin what MPC actually are and what the general idea of a projection approach is. Next we present efficient algorithms to obtain the projection matrices and how we use those in the conjugate gradient solver. Our main framework is already given and since it is written in C++ and CUDA the final implementation also will be in C++/CUDA. Finally we will take a look at the results and the performance hit we get when using this approach. We will also take a look at how the convergence speed is affected when using a modified precoditioner.