dearl all, I've using the ''alternative fermion Hubbard'' model, with open boundary conditions, and analyzing its single-particle correlations, <cdag_up:c_up>. I'm not being able however to compute these observables for larger lattices, e.g., L=~160. I'm getting a "segmentation fault" message, does anybody know why is happening this error, maybe due to memory? If so, what would be good parameters in order to avoid it? Here is the script i'm using: LATTICE="open chain lattice" MODEL="alternative fermion Hubbard" CONSERVED_QUANTUMNUMBERS="N,Sz,Nup,Ndown" N_total=160 Sz_total=0 Nup_total=80 Ndown_total=80 mu=0 t=1 NUMBER_EIGENVALUES=1 L=160 MAXSTATES=100 MEASURE_CORRELATIONS[cdag_up c_up]="cdag_up:c_up" {U = 9.000000e-01,V=5.500000e-01,TRUNCATION_ERROR=1e-7} regards, Fernando.
Hi Fernando, If you do exact diagonalization then the problem is that you run out of memory since this needs memory that is exponential in L
On Jun 11, 2014, at 8:10, Fernando Iemini <fernandoiemini@gmail.com> wrote:
dearl all,
I've using the ''alternative fermion Hubbard'' model, with open boundary conditions, and analyzing its single-particle correlations, <cdag_up:c_up>. I'm not being able however to compute these observables for larger lattices, e.g., L=~160. I'm getting a "segmentation fault" message, does anybody know why is happening this error, maybe due to memory? If so, what would be good parameters in order to avoid it?
Here is the script i'm using:
LATTICE="open chain lattice" MODEL="alternative fermion Hubbard" CONSERVED_QUANTUMNUMBERS="N,Sz,Nup,Ndown" N_total=160 Sz_total=0 Nup_total=80 Ndown_total=80 mu=0 t=1 NUMBER_EIGENVALUES=1 L=160 MAXSTATES=100 MEASURE_CORRELATIONS[cdag_up c_up]="cdag_up:c_up" {U = 9.000000e-01,V=5.500000e-01,TRUNCATION_ERROR=1e-7}
regards, Fernando.
Hi, Fernando, Assuming that you use DMRG simulation since you specify 'MAXSTATES'. (2014/06/12 0:10), Fernando Iemini wrote:
I've using the ''alternative fermion Hubbard'' model, with open boundary conditions, and analyzing its single-particle correlations, <cdag_up:c_up>. I'm not being able however to compute these observables for larger lattices, e.g., L=~160. I'm getting a "segmentation fault" message, does anybody know why is happening this error, maybe due to memory? If so, what would be good parameters in order to avoid it?
Here is the script i'm using:
LATTICE="open chain lattice" MODEL="alternative fermion Hubbard" CONSERVED_QUANTUMNUMBERS="N,Sz,Nup,Ndown" N_total=160 Sz_total=0 Nup_total=80 Ndown_total=80 mu=0 t=1 NUMBER_EIGENVALUES=1 L=160 MAXSTATES=100 MEASURE_CORRELATIONS[cdag_up c_up]="cdag_up:c_up" {U = 9.000000e-01,V=5.500000e-01,TRUNCATION_ERROR=1e-7}
This is memory issue. The problem is that 'MEASURE_CORRELATIONS' collects all pair sites. Observing ~10000 measurements costs too much memory. Not tested, but I believe that you can avoid this issue by tweaking models.xml and lattices.xml. Change the name of some specific operator and just calculating between them. -- Ryo IGARASHI, Ph.D. rigarash@issp.u-tokyo.ac.jp tel: 04-7136-5343, fax: 04-7136-3441 OpenPGP fingerprint: BAD9 71E3 28F3 8952 5640 6A53 EC79 A280 6A19 2319
Hi Fernando, If you are using a recent version of ALPS, e.g. 2.2.0b3, you can use the new dmrg application `mps_optim`. It works the same way as the `dmrg` application, but additionally allows you to measure correlation functions for i<j, which very often is good enough. MEASURE_HALF_CORRELATIONS[cdag_up c_up]="cdag_up:c_up" Best regards, Michele -- ETH Zurich Michele Dolfi Institute for Theoretical Physics HIT G 32.4 Wolfgang-Pauli-Str. 27 8093 Zurich Switzerland dolfim@phys.ethz.ch www.itp.phys.ethz.ch +41 44 633 78 56 phone +41 44 633 11 15 fax On Jun 12, 2014, at 2:08 AM, Ryo IGARASHI <rigarash@issp.u-tokyo.ac.jp> wrote:
Hi, Fernando,
Assuming that you use DMRG simulation since you specify 'MAXSTATES'.
(2014/06/12 0:10), Fernando Iemini wrote:
I've using the ''alternative fermion Hubbard'' model, with open boundary conditions, and analyzing its single-particle correlations, <cdag_up:c_up>. I'm not being able however to compute these observables for larger lattices, e.g., L=~160. I'm getting a "segmentation fault" message, does anybody know why is happening this error, maybe due to memory? If so, what would be good parameters in order to avoid it?
Here is the script i'm using:
LATTICE="open chain lattice" MODEL="alternative fermion Hubbard" CONSERVED_QUANTUMNUMBERS="N,Sz,Nup,Ndown" N_total=160 Sz_total=0 Nup_total=80 Ndown_total=80 mu=0 t=1 NUMBER_EIGENVALUES=1 L=160 MAXSTATES=100 MEASURE_CORRELATIONS[cdag_up c_up]="cdag_up:c_up" {U = 9.000000e-01,V=5.500000e-01,TRUNCATION_ERROR=1e-7}
This is memory issue. The problem is that 'MEASURE_CORRELATIONS' collects all pair sites. Observing ~10000 measurements costs too much memory.
Not tested, but I believe that you can avoid this issue by tweaking models.xml and lattices.xml. Change the name of some specific operator and just calculating between them.
-- Ryo IGARASHI, Ph.D. rigarash@issp.u-tokyo.ac.jp tel: 04-7136-5343, fax: 04-7136-3441 OpenPGP fingerprint: BAD9 71E3 28F3 8952 5640 6A53 EC79 A280 6A19 2319
participants (4)
-
Fernando Iemini
-
Matthias Troyer
-
Michele Dolfi
-
Ryo IGARASHI