Dear Fa,

thanks for your email. 

(1) Indeed the second vector in the triangular lattice description is incorrect and should be replaced by
   <VECTOR>a/2 a*sqrt(3)/2</VECTOR>
We have changed this in the code repository (it will become available with the next release). So please update this line in your current files for the time being.

(2) Yes, you can already implement spatially anisotropic models with different interaction terms on different types of bonds. The <BONDTERM> has a type-variable which allows this. Here is an example from the model library how-to webpage (http://alps.comp-phys.org:16080/wiki/index.php/Tutorials:ModelHOWTO) for a Bose Hubbard model:

<HAMILTONIAN name="hardcore boson">
  <PARAMETER name="mu" default="0"/>
  <PARAMETER name="t" default="1"/>
  <PARAMETER name="t'" default="1"/>
  <BASIS ref="hardcore boson"/>
  <SITETERM type="0">
    -mu*n
  </SITETERM> 
  <BONDTERM type="0" source="i" target="j">
    -t*(bdag(i)*b(j)+bdag(j)*b(i)))
  </BONDTERM>
  <BONDTERM type="1" source="i" target="j">
    -t'*(bdag(i)*b(j)+bdag(j)*b(i)))
  </BONDTERM>
</HAMILTONIAN>

If you implement a new (triangular) lattice with two different type of bonds, e.g. "0" and "1", and implement your Hamiltonian similar to the one above, will you be able to study the model you are interested in?

Please let me know, and we can further assist you
Simon

PS: I have copied this email to the ALPS user mailing list.



Dear Dr. Trebst

I'm a graduate student in physics department at berkeley.
Currently i'm trying to learn to use the ALPS library and
application and have met some problems.

I'm not sure whether you are the right person to contact.
If not, please ignore the following and take my apology.

(1). In the lattices.xml library, the triangular lattice is
<LATTICE name="triangular lattice" dimension="2">
<PARAMETER name="a" default="1"/>
<BASIS>
<VECTOR>a 0</VECTOR>
<VECTOR>a*sqrt(3)/2 a/2</VECTOR>
</BASIS>
<RECIPROCALBASIS>
<VECTOR>0 4*pi/a/sqrt(3)</VECTOR>
<VECTOR>2*pi/a -2*pi/a/sqrt(3)</VECTOR>
</RECIPROCALBASIS>
</LATTICE>

The 2nd basis vector (a*sqrt(3)/2, a/2) doesn't look correct to me.  Or
maybe I misunderstood how it is represented.

(2). In the tutorial I can't find out how to make spatially
anisotropic model. For example, anisotropic triangular Heisenberg
antiferromagnet.
It seems that the <BONDTERM> doesn't care the type of bond, or
the type of source/target sites.
Does this kind of feature have been incorporated into the applications,
or do I have to write my own code to implement this?

Thank you for your help.

Best regards.

~Fa