Collision is realized by MFThread's function MF::Database::Thread::DoCollisionOnly, while combined pre-collision and collision is realized by MFThread's function MF::Database::Thread::DoPreANDCollision. Both functions are paralelized with OpenMP. Some boundary nodes e.g. Dirichlet velocity type, require additional calculations before collision. It is realized by the pre-collision step or combined pre-collision and collision step. MFThread's flag MF::Database::Thread::m_DoPreCollision controls the execution of pre-collision step. During these steps for each thread's nodeForAutoThreading the functions to which the pointers were previously stored in the variables MF::Database::Thread::m_pBoundaryFunction (pre-collision) and MF::Database::Thread::m_pColideFunc (collision) of the MFThread object are called.