1: \begin{abstract}

2: We consider a Krylov subspace approximation method for the symmetric differential Riccati equation

3: $\dot{X} = AX + XA^T + Q - XSX$, $X(0)=X_0$. The method we consider is based on projecting

4: the large scale equation onto a Krylov subspace spanned by the matrix $A$ and the low rank factors of $X_0$ and $Q$. 

5: We prove that the method is structure preserving in the sense that it preserves 

6: two important properties of the exact flow, namely the positivity of the exact flow,

7: and also the property of monotonicity. 

8: We also provide a theoretical a priori error analysis which 

9: shows a superlinear convergence of the method. This behavior is illustrated in the numerical experiments. 

10: Moreover, we derive an efficient a posteriori error estimate

11: as well as discuss multiple time stepping combined with a cut of the rank of the numerical solution.

12: \end{abstract}

13: