Numerical solutions of saddle point problems