November 2009 - Comp-phys-alps-users

Lattice graph.
by Michal Maik 06 Nov '09

06 Nov '09

Hi, I'm a new user to the Alps program, so this may be a trivial question. I have gone through most of the tutorials and I'm getting familiar with the program. One thing I can't figure out is how to display a picture of the lattice that I define. What I mean is how do I go from the hml code for a lattice to to a graph of it? What program can I use to display it? For example, in the tutorial you write: <GRAPH vertices="5" edges="5"> <EDGE source="1" target="2"/> <EDGE source="2" target="3"/> <EDGE source="1" target="4"/> <EDGE source="2" target="5"/> <EDGE source="4" target="5"/> </GRAPH> Then you show what this lattice looks like. How can I get that picture to display? My main concern is that since I'm new to the program I would like to see if a lattice that I think I have defined is actually the lattice that I want. Thanks.

5 6

structure function in alps
by Fu-Jiun Jiang ITP 06 Nov '09

06 Nov '09

Dear all I got two quick questions for you as follows. 1. Can one force Alps to measure only k = 0 or k = \pi/2 part of the structure function? Looking into the code, I know how this can be done in the sse_dirloop code (one just cooks up his/her measurements) , how about the looper? 2. Does alps already take into account the prefactor for each type of sublattices when measure the structure function. For example, for honeycomb lattice, since one got two type of sublattices, it is useful to use +1 for one type sublattice and -1 for another type of sublattice. Sincerely, FJ

2 1

A looper question: "Binder ratio of Generalized Magnetization"==""Binder ratio of Staggered Magnetization"" for antiferromagnetic models?
by Prsa Krunoslav 05 Nov '09

05 Nov '09

Dear ALPS maintainers and users, Thanks for your generous help and advice. I would like to calculate the Neel temperature of a 3D antiferromagnet within the Heisenberg model using the looper application as a function of ratio of the two main interactions, J2/J1. The Binder ratio B(T) for different system dimensions L seems like the easiest way to calculate the transition temperature. Both interactions are unfortunately antiferromagnetic, giving rise to a small frustration. The lattice is not bipartite anymore as soon as the J2 is introduced. This removes the possibility of using the staggered magnetization Ms, Ms^2, Ms^4 and the Binder ratio Bs=<Ms^4>/(<Ms^2>)^2. However, there is an additional observable still available called the "Generalized magnetization", its square, etc. Does its Binder ratio (observable "Binder Ratio of Generalized Magnetization") correspond to the Binder ratio for staggered magnetizations for antiferromagnetic models? Does this hold if the classical ground state of the given model is, say, spiral? I.e. can I in principle use this observable to determine the order transitions for all possible models? I am thankful for any help you may provide. All the best. Kruno --------------------------------------------- Krunoslav Prsa, PhD student, Laboratory for neutron scattering, ETH Zurich and Paul Scherrer Institute CH-5232 Villigen-PSI, Switzerland tel:+41 56 310 20 91 ---------------------------------------

4 4

Re: [ALPS-users] A looper question: "Binder ratio of Generalized Magnetization"==""Binder ratio of Staggered Magnetization"" for antiferromagnetic models?
by Prsa Krunoslav 04 Nov '09

04 Nov '09

Dear Fabien, thanks for your quick reply, the situation is much clearer to me now. However, I believe that without some modification of the looper code I will not be able to extract much information: The "Staggered" observables are only calculated if the lattice is explicitly bipartite. Do you know if there is there some other option/trick available? All the best and thanks! Kruno --------------------------------------------- Krunoslav Prsa, PhD student, Laboratory for neutron scattering, ETH Zurich and Paul Scherrer Institute CH-5232 Villigen-PSI, Switzerland tel:+41 56 310 20 91 --------------------------------------- Reply to: [ALPS-users] A looper question: "Binder ratio of Generalized Magnetization"==""Binder ratio of Staggered Magnetization"" for antiferromagnetic models? Dear Kruno, > Dear ALPS maintainers and users, > > Thanks for your generous help and advice. > > I would like to calculate the Neel temperature of a 3D antiferromagnet within the Heisenberg model using the looper application as a function of ratio of the two main interactions, J2/J1. The Binder ratio B(T) for different system dimensions L seems like the easiest way to calculate the transition temperature. > > Both interactions are unfortunately antiferromagnetic, giving rise to a small frustration. The lattice is not bipartite anymore as soon as the J2 is introduced. This removes the possibility of using the staggered magnetization Ms, Ms^2, Ms^4 and the Binder ratio Bs=<Ms^4>/(<Ms^2>)^2. > > If you are looking for antiferromagnetic order, the staggered magnetization is the correct order parameter (and its Binder cumulant an appropriate tool) - independently of the interactions being frustrated or not. > However, there is an additional observable still available called the "Generalized magnetization", its square, etc. > > Does its Binder ratio (observable "Binder Ratio of Generalized Magnetization") correspond to the Binder ratio for staggered magnetizations for antiferromagnetic models? Does this hold if the classical ground state of the given model is, say, spiral? I.e. can I in principle use this observable to determine the order transitions for all possible models? > > No - to each type of order is associated a different order parameter. The staggered magnetization is the order parameter of antiferromagnetism and therefore can locate only transitions where antiferromagnetism sets in (or out). For other orders (e.g. spiral) you must compute the corresponding order parameter. Best Fabien > I am thankful for any help you may provide. > > All the best. > Kruno > > --------------------------------------------- > Krunoslav Prsa, > PhD student, Laboratory for neutron scattering, > ETH Zurich and Paul Scherrer Institute > CH-5232 Villigen-PSI, Switzerland > tel:+41 56 310 20 91 > --------------------------------------- > > >

2 1

Quantum Wang-Landau algorithm question: applicability of QWL on frustrated lattices
by Prsa Krunoslav 02 Nov '09

02 Nov '09

Dear Prof. Troyer, Thanks for your answer. Yes, the lattice is 3D; and no, the frustration is not strong. I will follow your advice then: I'll try the QWL for the non-frustrated case, and with looper add points on the frustrated side of the phase diagram, if that will be feasible. All the best, Kruno > No, the current QWL code cannot do that. It is best to use the looper > code. Are you talking about 3D lattices? If frustration is strong the > sign problem will make simulations impossible at T_N. > Matthias > On 29 Oct 2009, at 12:20, Prsa Krunoslav wrote: > Dear ALPS maintainers, > Since nobody has yet answered to my questions from almost a month > ago, I will repeat just one question this time. > > Can the current incarnation of the QWL code handle frustrated > lattices? Specifically, I want to find the transition temperature as > a function T_N(J2/J1), where J2 and J1 are antiferromagnetic. > > > All the best, > Kruno > > --------------------------------------- > Krunoslav Prsa, Ph. D. Student > Laboratory for Neutron Scattering > Paul Scherrer Institute and ETH-Zürich > CH-5232 Villigen PSI, Switzerland > tel: +41 56 310 20 91 > mob: +41 76 386 17 99 > ---------------------------------------- > > > > -----Original Message----- > From: Prsa Krunoslav > Sent: Friday, October 02, 2009 2:56 PM > To: comp-phys-alps-users at phys.ethz.ch > Subject: Quantum Wang-Landau algorithm question: applicability of > QWL on (slightly) frustrated lattices > > > Dear ALPS maintainers, > > I have three questions with regards to your extremely powerful QWL > algorithm. > > 1. In the ALPS online documentation section for QWL algorithm there > is a note: > "Note: This first version allows the simulation of isotropic > Heisenberg spin-1/2 ferro- and antiferromagnetic models on arbitrary > non-frustrated lattices at zero magnetic field. In the future, we > plan to relax this constraint, and also provide an implementation of > the QWL perturbation expansion. " > > Has this been improved with regards to antiferro- frustrated > lattices? I am using ALPS version 1.3.3. > > 2. When I run the qwl for my slightly frustrated S=1/2 3D lattice > (consisting of triangular antiferro patterns J1-J1-J2 with J2~0.15 > J1) the program does not seem to complain (It does not seem to > complain even for the triangular lattice). Can its output for this > situation be trusted though? > > 3. I was also wondering if I can generally verify the results (say > T_N) of the frustrated model by an extrapolation from the non- > frustrated side (say ferro J2=-0.5,...,-0.1,0), thereby assuming a > mean-field like behaviour T_N~(J1-J2), for small J2. I am an > experimentalist so I may not be aware of this. Is there a general > result that would say it's not possible or at least a counterexample? > > It may be of relevance to some of the questions that in this lattice > there is a change in the classical ground state from Neel to the > triangular at J2=J1/3. > > All the best, > Kruno > > --------------------------------------- > Krunoslav Prsa, Ph. D. Student > Laboratory for Neutron Scattering > Paul Scherrer Institute and ETH-Zürich > CH-5232 Villigen PSI, Switzerland > tel: +41 56 310 20 91 > mob: +41 76 386 17 99 > ---------------------------------------- --------------------------------------- Krunoslav Prsa, Ph. D. Student Laboratory for Neutron Scattering Paul Scherrer Institute and ETH-Zürich CH-5232 Villigen PSI, Switzerland tel: +41 56 310 20 91 mob: +41 76 386 17 99 ----------------------------------------

2 1