1: \begin{abstract}

2: % insert abstract here

3: In {\sl ab initio} calculations of electronic structures, total energies, and 

4: forces, it is convenient and often even necessary to employ a broadening

5: of the occupation numbers. If done carefully, 

6: this improves the accuracy of the calculated

7: electron densities and total energies and stabilizes the convergence

8: of the iterative approach towards self-consistency.

9: However, such a boardening may lead to an error in the calculation of the

10: forces.

11: %, and it is unclear, how significant this error will be and how it

12: %should be corrected.

13: %, but in standard schemes 

14: %it results in incorrect forces.

15: Accurate forces are needed for an efficient 

16: geometry optimization of polyatomic systems and for {\em ab initio} 

17: molecular dynamics (MD) calculations.

18: The relevance of this error and possible

19: ways to correct it will be discussed in this

20: paper.

21: The first approach is computationally very simple and in fact exact

22: for small MD time steps.

23: This is demonstrated  for the example of the

24: vibration of a carbon dimer and for the relaxation 

25: of the top layer of the (111)-surfaces of aluminium and platinum.

26: The second, more general, scheme employs 

27: linear-response theory and is applied to the calculation 

28: of the surface relaxation of Al\,(111).

29: We will show that the quadratic dependence

30: of the forces on the broadening width enables an efficient

31: extrapolation to the correct result.

32: Finally the results of these correction methods will be 

33: compared to the forces obtained by using the

34: smearing scheme, which has been proposed by Methfessel and Paxton.

35: \end{abstract}

36: