Hi, I would like to use the Quantum Wang-Landau app to obtain some benchmark results for the free energy of a spin chain. To make sure I understood the parameters I decided to do the two-qubit Heisenberg model first (with open BC's). Unfortunately, no matter how I adjust the parameters the resulting energies seem quite poor compared to my exact calcuation in Mathematica.
To be more specific, I've been adjusting the CUTOFF parameter to between 100-1000 and the number of WL steps from 16-25. The results I've been seeing are energies around -0.3 at the lowest T's (where E~-0.75 in fact) and a wide variance in energies at the highest T's, often more than 30% off the actual value.
Thanks for any help!
Dear Miles,
Could you post your parameter file?
Synge
On 2010/07/28, at 4:46, Miles Stoudenmire wrote:
Hi, I would like to use the Quantum Wang-Landau app to obtain some benchmark results for the free energy of a spin chain. To make sure I understood the parameters I decided to do the two-qubit Heisenberg model first (with open BC's). Unfortunately, no matter how I adjust the parameters the resulting energies seem quite poor compared to my exact calcuation in Mathematica.
To be more specific, I've been adjusting the CUTOFF parameter to between 100-1000 and the number of WL steps from 16-25. The results I've been seeing are energies around -0.3 at the lowest T's (where E~-0.75 in fact) and a wide variance in energies at the highest T's, often more than 30% off the actual value.
Thanks for any help!
-- -=Miles Stoudenmire=- miles.stoudenmire@gmail.com
Dear Synge, Thanks for your help. Here is a sample parameter file:
LATTICE="open chain lattice" MODEL="spin" local_S = 1/2 L = 2 CUTOFF = 100 NUMBER_OF_WANG_LANDAU_STEPS = 20 {J = 1}
This is with a very recent svn copy of ALPS.
On 27 July 2010 19:56, Synge Todo wistaria@comp-phys.org wrote:
Dear Miles,
Could you post your parameter file?
Synge
On 2010/07/28, at 4:46, Miles Stoudenmire wrote:
Hi, I would like to use the Quantum Wang-Landau app to obtain some benchmark results for the free energy of a spin chain. To make sure I understood the parameters I decided to do the two-qubit Heisenberg model first (with open BC's). Unfortunately, no matter how I adjust the parameters the resulting energies seem quite poor compared to my exact calcuation in Mathematica.
To be more specific, I've been adjusting the CUTOFF parameter to between 100-1000 and the number of WL steps from 16-25. The results I've been seeing are energies around -0.3 at the lowest T's (where E~-0.75 in fact) and a wide variance in energies at the highest T's, often more than 30% off the actual value.
Thanks for any help!
-- -=Miles Stoudenmire=- miles.stoudenmire@gmail.com
comp-phys-alps-users@lists.phys.ethz.ch