1: \begin{abstract}

2: We consider a lattice gas model with an infinite pairwise nonconvex

3: total interaction of the form

4: \[ 

5: V(r)=\frac{J}{r^2} + A\frac{\cos (2k_Far+\phi )}{r}\;.

6: \]

7: This one-dimensional interaction might account, for example, for

8: adsorption of alkaline elements on W(112) and Mo(112). The first

9: term describes the effective dipole-dipole interaction while the other

10: one the indirect (oscillatory) interaction; $J$, $A$, and $\phi$

11: are the model parameters, whereas $k_F$ stands for the wavevector of

12: electrons at the Fermi surface and $a$ is a lattice constant.

13: We search for the (periodic) ground states. To solve this difficult

14: problem we have applied a novel numerical method to accelerate 

15: the convergence of Fourier series.

16: A competition between the dipole-dipole and indirect interactions 

17: turns out to be very important. We have found that the reduced

18: chemical potential $\mu/J$ versus $A/J$ phase diagrams contain a

19: region $0.1 \leq A/J \leq 1.5$  dominated by several phases only with

20: periods up to nine lattice constants. Of course, the resulting

21: sequence of phases (for fixed $A/J$)   depends on the wavevector

22: $k_F$ and the phase shift $\phi$.

23: The remaining phase diagram reveals a complex structure of

24: usually long periodic phases.

25: We conjecture, based on the above results, that 

26: qua\-si\--one-\-di\-men\-sio\-nal surface states might be responsible 

27: for experimentally observed ordered phases at the (112) surface of 

28: tungsten and  molybdenum.

29: \end{abstract}

30: