Dear all
I test a Hubbard model with external potential (double well) using DMRG code in ALPS, but I can not get a symmetry result (density distribution) with SWEEP and MAXSTATES up to 50 and 500, respectively. It seem that I need to set more SWEEPS and STATES in calculation. But it cast to much time and is not very practical.
I found that the symmetry between spin-up and spin-down atoms is not kept in my present result, i.e. the density distributions of the two kinds of atoms are different ( I have set Sz as "CONSERVED_QUANTUMNUMBERS" and Sz_total=0 in command file).
I wonder if there are some ways for me to restrict the Sz in each site to 0 and obtain the correct result with a few SWEEPS ? or may be I should use a better series of STATES ?
Any comments are welcome. thank you.
chenahai
On 26 Apr 2012, at 19:46, Chen Ahai wrote:
Dear all
I test a Hubbard model with external potential (double well) using DMRG code in ALPS, but I can not get a symmetry result (density distribution) with SWEEP and MAXSTATES up to 50 and 500, respectively. It seem that I need to set more SWEEPS and STATES in calculation. But it cast to much time and is not very practical.
You will probably need many more STATES, and it can help if you always keep the full number of states and don't ramp them up. You can also look at our recent arXiv paper on multi-grid DMRG for a discussion of the issue that you might have observed and a possible solution.
I found that the symmetry between spin-up and spin-down atoms is not kept in my present result, i.e. the density distributions of the two kinds of atoms are different ( I have set Sz as "CONSERVED_QUANTUMNUMBERS" and Sz_total=0 in command file).
Is there a different number of up and down atoms or is just the distribution not symmetric?
I wonder if there are some ways for me to restrict the Sz in each site to 0 and obtain the correct result with a few SWEEPS ?
Restricting Sz=0 on each site means that no singly occupied sites will be allowed. This is probably not what you want.
or may be I should use a better series of STATES ?
yes, use more states
Matthias Troyer
Dear prof. Matthias
Thanks for your reply.
Is there a different number of up and down atoms or is just the
distribution not symmetric? There are both a different number of up and down atoms and a asymmetric distribution. When I increase SWEEPS and MAXSTATES from 30,300 to 50,500, there are slight in the result (< <10^-3).
Restricting Sz=0 on each site means that no singly occupied sites will
be allowed. This is probably not what you want. For a simple Hubbard model with external potential, the operators for up and down atoms in Hamiltonian are symmetry. There will not be singly occupied sites in the result.
Thanks again.
Chen Ahai
comp-phys-alps-users@lists.phys.ethz.ch