Hi. I'm a student doing research on Heisenberg spin chains, and I'm attempting to get ground state energies for longer chains than my group has done before. I've been running the DMRG program included with ALPS (not the 'simple' one) with a parameter file that looks something like:
MODEL = "spin" LATTICE = "chain lattice" LATTICE_LIBRARY = "../lattices.xml" MODEL_LIBRARY = "../models.xml" CONSERVED_QUANTUMNUMBERS = "Sz" local_S = 1/2 J = 1 V = 0 SWEEPS = 30 MAXSTATES = 60 OUTPUT_LEVEL = 1 {L = 30}
I've been able to successfully run the program, though I'm not entirely sure that all the parameters I've specified are the right ones (and that it's not just reverting to the defaults). I've been unable to find any documentation on how this particular implementation of the program works, so can anyone tell me if I'm off to the right start?
In addition, I'd like to be able to run this program with site-dependent interaction coefficients (such as J = 3*k) or so, but after much tinkering, I've been unable to find the variable 'k' that represents position on the lattice. My question is is this even possible? If so, how? Should I instead be trying to modify the source code for the program and recompile it?
Thanks for your time.
This looks fine to me but I'll ask Adrian Feiguin who is the author of the program. We are still working on the documentation.
One question: do you want to fix the simulation to a specific value of Sz?
For the site-dependent interaction you have to use an inhomogeneous lattice, created by adding an <INHOMOGENEOUS><VERTEX/></ INHOMOGENEOUS> tag to the lattice definition. Once that is done one can use the variables x, y and z in the definition of the coupling constants. This is, at the moment, supported only in the QMC codes though, but we can add it to the DMRG code.
Best regards
Matthias
On 15 Feb 2007, at 09:07, Christopher Brust wrote:
Hi. I'm a student doing research on Heisenberg spin chains, and I'm attempting to get ground state energies for longer chains than my group has done before. I've been running the DMRG program included with ALPS (not the 'simple' one) with a parameter file that looks something like:
MODEL = "spin" LATTICE = "chain lattice" LATTICE_LIBRARY = "../lattices.xml" MODEL_LIBRARY = "../models.xml" CONSERVED_QUANTUMNUMBERS = "Sz" local_S = 1/2 J = 1 V = 0 SWEEPS = 30 MAXSTATES = 60 OUTPUT_LEVEL = 1 {L = 30}
I've been able to successfully run the program, though I'm not entirely sure that all the parameters I've specified are the right ones (and that it's not just reverting to the defaults). I've been unable to find any documentation on how this particular implementation of the program works, so can anyone tell me if I'm off to the right start?
In addition, I'd like to be able to run this program with site- dependent interaction coefficients (such as J = 3*k) or so, but after much tinkering, I've been unable to find the variable 'k' that represents position on the lattice. My question is is this even possible? If so, how? Should I instead be trying to modify the source code for the program and recompile it?
Thanks for your time.
I've checked the issue with the quantum numbers, and this is the thing: if you don't specify Sz_total, the program will run in the "grand canonical", this is, using a Hilbert space with dim=2^N states. Running in the "canonical" (by setting Sz_total=0) will improve performance considerably by working in a subspace with a reduced dimension. For an example of how to do this, take a look at the parms file included with the dmrg code.
Just a coment, you declared Sz as a conserved quantum number, this will still be used in the code in order to reduce matrices in block form, already a big optimization.
Hope this helps,
Best regards, <ADRIAN>
Matthias Troyer wrote:
This looks fine to me but I'll ask Adrian Feiguin who is the author of the program. We are still working on the documentation.
One question: do you want to fix the simulation to a specific value of Sz?
For the site-dependent interaction you have to use an inhomogeneous lattice, created by adding an <INHOMOGENEOUS><VERTEX/></ INHOMOGENEOUS> tag to the lattice definition. Once that is done one can use the variables x, y and z in the definition of the coupling constants. This is, at the moment, supported only in the QMC codes though, but we can add it to the DMRG code.
Best regards
Matthias
On 15 Feb 2007, at 09:07, Christopher Brust wrote:
Hi. I'm a student doing research on Heisenberg spin chains, and I'm attempting to get ground state energies for longer chains than my group has done before. I've been running the DMRG program included with ALPS (not the 'simple' one) with a parameter file that looks something like:
MODEL = "spin" LATTICE = "chain lattice" LATTICE_LIBRARY = "../lattices.xml" MODEL_LIBRARY = "../models.xml" CONSERVED_QUANTUMNUMBERS = "Sz" local_S = 1/2 J = 1 V = 0 SWEEPS = 30 MAXSTATES = 60 OUTPUT_LEVEL = 1 {L = 30}
I've been able to successfully run the program, though I'm not entirely sure that all the parameters I've specified are the right ones (and that it's not just reverting to the defaults). I've been unable to find any documentation on how this particular implementation of the program works, so can anyone tell me if I'm off to the right start?
In addition, I'd like to be able to run this program with site- dependent interaction coefficients (such as J = 3*k) or so, but after much tinkering, I've been unable to find the variable 'k' that represents position on the lattice. My question is is this even possible? If so, how? Should I instead be trying to modify the source code for the program and recompile it?
Thanks for your time.
comp-phys-alps-users@lists.phys.ethz.ch