Two suggestions/questions:

1. Try an odd number of sites, since that will make the ground state unique

2. Make sure that you converge the DMRG and use a sufficient number of states and sweeps

If you then still want to add that term, then look at the tutorials to see how we make the end sites have spin-1/2 instead of spin-1 for the Haldane spin chain. You can use the same trick in your model, but I don't think that this term will solve your problem. 

On Sep 28, 2014, at 22:10, hamid mosadegh <hamid.mosadegh@gmail.com> wrote:

I want to solve extended Hubbard model
<HAMILTONIAN name="fermion Hubbard">
  <PARAMETER name="mu" default="0"/>
  <PARAMETER name="t" default="1"/>
  <PARAMETER name="V" default="0"/>
  <PARAMETER name="t'" default="0"/>
  <PARAMETER name="V'" default="0"/>
  <PARAMETER name="U" default="0"/>
  <PARAMETER name="t0" default="t"/>
  <PARAMETER name="t1" default="t'"/>
  <PARAMETER name="V0" default="V"/>
  <PARAMETER name="V1" default="V/8"/>
  <PARAMETER name="V2" default="V/27"/>
  <PARAMETER name="V3" default="V/64"/>
  <PARAMETER name="V4" default="V/125"/>
  <PARAMETER name="V5" default="V/216"/>
  <BASIS ref="fermion"/>
  <SITETERM site="i">
    <PARAMETER name="mu#" default="mu"/>
    <PARAMETER name="U#" default="U"/>
    -mu#*n(i)+U#*n_up(i)*n_down(i)
  </SITETERM>
  <BONDTERM source="i" target="j">
    <PARAMETER name="t#" default="0"/>
    <PARAMETER name="V#" default="0"/>
    -t#*fermion_hop(i,j) + V#*n(i)*n(j)

In the other code, also, the boundary potential (V/2[n_1+n_L]) is added.

I means that <n(i)> must be uniform throughout the lattice as [2,0,2,0,....,0,2] in CDW phase.

but <n(i)> evaluated using ALPS is modulated by  a wave. I want to use this potential to uniform this wave
throughout the lattice as we expect for  the ground state of infinite lattice. In this way, the effect of OBC is reduced.

On Sun, Sep 28, 2014 at 10:18 PM, Matthias Troyer <troyer@phys.ethz.ch> wrote:

It is n ot a question of finding the ground state, which the code does, but the real question is that you have to find out is what model that other code implemented. After that you can code the same model in ALPS and we can help you with that.

On 28 Sep 2014, at 20:26, hamid mosadegh <hamid.mosadegh@gmail.com> wrote:

Thanks for your attention.

I try to find the local density of fermions in the ground state of the extended Hubbard model using DMRG with OBC
for L=40 , U=0 and V=4 ( density.txt file) but the  local density in the ground state must be as localdens.dat file (calculated using the other code). I think that a potential on boundary must be added to evaluate local density in the ground state with OBC.

On Sun, Sep 28, 2014 at 6:26 PM, Matthias Troyer <troyer@phys.ethz.ch> wrote:

Why would you like to add that?

I don't understand that you need to add a term to find the ground state, since you will find the ground state of whatever model you put in.

On 27 Sep 2014, at 21:22, hamid mosadegh <hamid.mosadegh@gmail.com> wrote:

> Dear all
>
> It seems that we must put a potential on boundary to obtain the ground state using DMRG.
> How i can add a term as V/2*(n(1)+n(L)) to fermion model in models.xml
>
> Thanks
>
>
>
> --
> H. Mosadeq
> Shahr-e-Kord University (SKU),
> Shahr-e-Kord , Iran
>

--

H. Mosadeq
Shahr-e-Kord University (SKU),
Shahr-e-Kord , Iran

<n_U0V4><localdens.dat>

--

H. Mosadeq
Shahr-e-Kord University (SKU),
Shahr-e-Kord , Iran