1: \begin{abstract}

2: \pacs{71.10.Pm,71.10.Fd, 78.20 Bh}

3: The density-matrix renormalization group (DMRG) algorithm can be adapted to the

4: calculation of dynamical correlation functions in various ways which all 

5: represent compromises between computational efficiency and physical accuracy.

6: In this paper we reconsider the oldest approach based on a suitable 

7: Lanczos-generated approximate basis and implement it using matrix product 

8: states (MPS) for the representation of the basis states. The direct use 

9: of matrix product states combined with an ex-post reorthogonalization method 

10: allows to avoid several shortcomings of the original

11: approach, namely the multi-targeting and the approximate representation of the 

12: Hamiltonian inherent in earlier Lanczos-method implementations in the DMRG 

13: framework, and to deal with the ghost problem of Lanczos methods, 

14: leading to a much better convergence of the spectral weights and 

15: poles. We present results for the dynamic spin structure factor of the spin-1/2

16: antiferromagnetic Heisenberg chain. A comparison to Bethe ansatz results in 

17: the thermodynamic limit reveals that the MPS-based Lanczos approach is much more

18: accurate than earlier approaches at minor additional numerical cost.

19: \end{abstract}

20: