Hi Fernando,

The application is using the same inputs as the former dmrg application, therefore you can reuse the tutorials you find online. In the alps tutorials you find already “adapted” Python script on how to use.

We are preparing a paper explaining the code. It will (very) soon be available.

Yes, you can measure four-points correlation functions (just enter four operators in the measurement definition). However what you want to compute is not really four-point and doing a O(N^4) calculation is overkilling. The best is using the (currently undocumented function): MEASURE_HALF_NN_CORRELATIONS[obsname] = “cdag_up:c_up:cdag_down:c_down” which will apply the four operators at potions i,i+1,j,j+1 with i<j. This is actually one of the examples we show in the release paper.

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 16, 2014, at 2:06 PM, Fernando Iemini fernandoiemini@gmail.com wrote:

dear all,

thanks for the responses, and sorry for the delay to reply. Yes, i was using the dmrg, and in fact using this new dmrg application `mps_optim`, as suggested by Michele Dolfi, solved my problem! I didn't know about this application, is there somewhere i can find a tutorial about it, or is it exactly the same input's as dmrg? And just a last point, in this application can i measure some "four-point" correlations, like "cdag_m:c_{m+1}:cdag_n:c_{n+1}" ?

best regards, Fernando.

---

Date: Thu, 12 Jun 2014 08:52:52 +0200 From: Michele Dolfi dolfim@phys.ethz.ch To: comp-phys-alps-users@lists.phys.ethz.ch Subject: Re: [ALPS-users] alternative fermion Hubbard Message-ID: 52E3FD2A-60E3-4D48-8315-1ACDF7348498@phys.ethz.ch Content-Type: text/plain; charset="us-ascii"

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