Dear all,
I have a few questions about parameters in the spinmc application. In the Tutorias: Parameters web page, I found
THERMALIZATION: Number of Monte Carlo sweeps for thermalization SWEEPS: number of Monte Carlo steps (after thermalization)
-I don't understand what is the number of "copys" used to compute the averages.
-Between copy and copy to compute the averages, how many MC steps are there?
- Suppose I run with different magnetic fields. Is it possible to use the previous state as initial state for the next run?
Best regards,
Dear all,
I have a few questions about the observables that can be measured by the spinmc application.
- It is possible to compute the static magnetic structure factor?
S(q)=Sum_{ij} <S_i.S_j> Exp(i*q(ri-rj)) or
- It is possible to compute the average of correlations <S_i.S_j> ? if
not, there is possibility to include now such terms?
Best regards,
Hector
-- Dr. Hector Diego Rosales IFLP - Dto. de Física UNLP, Calle 49 y 115 s/n, 1900 La Plata, Argentina Tel: +54 221 424 6062; Fax: +54 221 423 0122 e-Mail: rosales@fisica.unlp.edu.ar
Dear all,
I have a few questions too about simulations with spinmc application.
I'll make a specific example so I hope it'll be more clear... I have to run a MC classic simulation of the XY model defined on a square lattice with periodic boundary conditions. I wrote the following input file:
LATTICE_LIBRARY="lattices.xml" LATTICE="square lattice" T=0.892936869 J=1 THERMALIZATION=50000 SWEEPS=1000000 UPDATE="cluster" MODEL="XY" BOUNDARY type="periodic" S=1 {L=64;} {L=128;}
so that the simulation is performed for two different lattice sizes. I did it following the instruction on the tutorials and it seemed to work. The questions are:
1) Which are the initial conditions at the beginning of the simulation? Are the spins spread randomly on the lattice?
2) Is it possible to save the configurations which are used for the calculations of the observables? I suppose the program generates a certain number of configurations then on one of them it takes the values of the observables and then it makes other pure dynamic MC steps and then it take again the measure of the observable on another configuration and so on...Is it correct? Can I save those configurations used for the accumulation of the observables?
3) I also would like to know if is it already possible to compute the density of vortix in this model...But I suppose this observable has to be added editing some of the spin mc files....
4) If now I want to run the same simulation but for a three dimensional simple cubic lattice is it correct to write LATTICE="simple cubic lattice" instead of LATTICE="square lattice" as above? Because I had some problems in this kind of simulation and I don't know exactly if this is the problem or if I've made a mistake somewhere else....
Thank you for the help,
best regards,
Rachele Nerattini
2011/5/16 rosales@fisica.unlp.edu.ar
Dear all,
I have a few questions about parameters in the spinmc application. In the Tutorias: Parameters web page, I found
THERMALIZATION: Number of Monte Carlo sweeps for thermalization SWEEPS: number of Monte Carlo steps (after thermalization)
-I don't understand what is the number of "copys" used to compute the averages.
-Between copy and copy to compute the averages, how many MC steps are there?
- Suppose I run with different magnetic fields. Is it possible to use the
previous state as initial state for the next run?
Best regards,
Dear all,
I have a few questions about the observables that can be measured by the spinmc application.
- It is possible to compute the static magnetic structure factor?
S(q)=Sum_{ij} <S_i.S_j> Exp(i*q(ri-rj)) or
- It is possible to compute the average of correlations <S_i.S_j> ? if
not, there is possibility to include now such terms?
Best regards,
Hector
-- Dr. Hector Diego Rosales IFLP - Dto. de Física UNLP, Calle 49 y 115 s/n, 1900 La Plata, Argentina Tel: +54 221 424 6062; Fax: +54 221 423 0122 e-Mail: rosales@fisica.unlp.edu.ar
-- Dr. Hector Diego Rosales IFLP - Dto. de Física UNLP, Calle 49 y 115 s/n, 1900 La Plata, Argentina Tel: +54 221 424 6062; Fax: +54 221 423 0122 e-Mail: rosales@fisica.unlp.edu.ar
Dear Matthias Troyer,
Thank you very much for the answers. About the last answer,
- It is possible to compute the static magnetic structure factor?
S(q)=Sum_{ij} <S_i.S_j> Exp(i*q(ri-rj)) or
- It is possible to compute the average of correlations <S_i.S_j> ? if
not, there is possibility to include now such terms?
This can easily be added.
I know which files I have to modify in the quantum case (models.xml), but in the classical MC I don't know. - Could you tell me which file I have to modify? After do that I think I have to run again ./configure, it is correct? - Is there an example in the ALPS-web page about that? because I didn't find any.
Best regards
Hector D. Rosales
Dear all,
I have a few questions too about simulations with spinmc application.
I'll make a specific example so I hope it'll be more clear... I have to run a MC classic simulation of the XY model defined on a square lattice with periodic boundary conditions. I wrote the following input file:
LATTICE_LIBRARY="lattices.xml" LATTICE="square lattice" T=0.892936869 J=1 THERMALIZATION=50000 SWEEPS=1000000 UPDATE="cluster" MODEL="XY" BOUNDARY type="periodic" S=1 {L=64;} {L=128;}
so that the simulation is performed for two different lattice sizes. I did it following the instruction on the tutorials and it seemed to work. The questions are:
- Which are the initial conditions at the beginning of the simulation?
Are the spins spread randomly on the lattice?
- Is it possible to save the configurations which are used for the
calculations of the observables? I suppose the program generates a certain number of configurations then on one of them it takes the values of the observables and then it makes other pure dynamic MC steps and then it take again the measure of the observable on another configuration and so on...Is it correct? Can I save those configurations used for the accumulation of the observables?
- I also would like to know if is it already possible to compute the
density of vortix in this model...But I suppose this observable has to be added editing some of the spin mc files....
- If now I want to run the same simulation but for a three dimensional
simple cubic lattice is it correct to write LATTICE="simple cubic lattice" instead of LATTICE="square lattice" as above? Because I had some problems in this kind of simulation and I don't know exactly if this is the problem or if I've made a mistake somewhere else....
Thank you for the help,
best regards,
Rachele Nerattini
2011/5/16 rosales@fisica.unlp.edu.ar
Dear all,
I have a few questions about parameters in the spinmc application. In the Tutorias: Parameters web page, I found
THERMALIZATION: Number of Monte Carlo sweeps for thermalization SWEEPS: number of Monte Carlo steps (after thermalization)
-I don't understand what is the number of "copys" used to compute the averages.
-Between copy and copy to compute the averages, how many MC steps are there?
- Suppose I run with different magnetic fields. Is it possible to use
the previous state as initial state for the next run?
Best regards,
Dear all,
I have a few questions about the observables that can be measured by
the
spinmc application.
- It is possible to compute the static magnetic structure factor?
S(q)=Sum_{ij} <S_i.S_j> Exp(i*q(ri-rj)) or
- It is possible to compute the average of correlations <S_i.S_j> ? if
not, there is possibility to include now such terms?
Best regards,
Hector
-- Dr. Hector Diego Rosales IFLP - Dto. de Física UNLP, Calle 49 y 115 s/n, 1900 La Plata, Argentina Tel: +54 221 424 6062; Fax: +54 221 423 0122 e-Mail: rosales@fisica.unlp.edu.ar
-- Dr. Hector Diego Rosales IFLP - Dto. de Física UNLP, Calle 49 y 115 s/n, 1900 La Plata, Argentina Tel: +54 221 424 6062; Fax: +54 221 423 0122 e-Mail: rosales@fisica.unlp.edu.ar
On May 20, 2011, at 2:03 AM, rosales@fisica.unlp.edu.ar wrote:
Dear Matthias Troyer,
Thank you very much for the answers. About the last answer,
- It is possible to compute the static magnetic structure factor?
S(q)=Sum_{ij} <S_i.S_j> Exp(i*q(ri-rj)) or
- It is possible to compute the average of correlations <S_i.S_j> ? if
not, there is possibility to include now such terms?
This can easily be added.
I know which files I have to modify in the quantum case (models.xml), but in the classical MC I don't know.
- Could you tell me which file I have to modify? After do that I think I
have to run again ./configure, it is correct?
- Is there an example in the ALPS-web page about that? because I didn't
find any.
The source code will have to be added, which I can do when I find a few hours - maybe I can do it next week.
Matthias
On May 16, 2011, at 1:44 AM, Rachele Nerattini wrote:
Dear all,
I have a few questions too about simulations with spinmc application.
I'll make a specific example so I hope it'll be more clear... I have to run a MC classic simulation of the XY model defined on a square lattice with periodic boundary conditions. I wrote the following input file:
LATTICE_LIBRARY="lattices.xml" LATTICE="square lattice" T=0.892936869 J=1 THERMALIZATION=50000 SWEEPS=1000000 UPDATE="cluster" MODEL="XY" BOUNDARY type="periodic" S=1 {L=64;} {L=128;}
so that the simulation is performed for two different lattice sizes. I did it following the instruction on the tutorials and it seemed to work. The questions are:
- Which are the initial conditions at the beginning of the simulation? Are the spins spread randomly on the lattice?
No, they are all aligned - but it is easy to change that in the code.
- Is it possible to save the configurations which are used for the calculations of the observables? I suppose the program generates a certain number of configurations then on one of them it takes the values of the observables and then it makes other pure dynamic MC steps and then it take again the measure of the observable on another configuration and so on...Is it correct? Can I save those configurations used for the accumulation of the observables?
Yes, it is possible, however this is not efficient. Using cluster updates generating a new configuration takes about as much time as reading it from disk.
- I also would like to know if is it already possible to compute the density of vortix in this model...But I suppose this observable has to be added editing some of the spin mc files....
Yes, you need to add that to the code.
- If now I want to run the same simulation but for a three dimensional simple cubic lattice is it correct to write
LATTICE="simple cubic lattice" instead of LATTICE="square lattice" as above?
Yes, that should work
Matthias
Dear Mr Troyer, thank you very much for the help. About question 2:
2) Is it possible to save the configurations which are used for the calculations of the observables? I suppose the program generates a certain number of configurations then on one of them it takes the values of the observables and then it makes other pure dynamic MC steps and then it take again the measure of the observable on another configuration and so on...Is it correct? Can I save those configurations used for the accumulation of the observables?
Yes, it is possible, however this is not efficient. Using cluster updates generating a new configuration takes about as much time as reading it from disk.
How can I read the configuration from the disk?
Another question: I already asked you if it were possible to perform a simulation of an O(4) model. I edited the file spinmc_factory.C adding these lines
else if (parms["MODEL"]=="O(4)") if (maxElemCount == 1) return new SpinSim< ONMoment<4>, MIdMatrix<double,4> >(where, parms, node); else { produceError(parms); return 0;
you said to type make install on the terminal but it doesn't work.... what can I do?
Thank you again and sorry the big amount of questions....
Best regards Rachele Nerattini
2011/5/21 Matthias Troyer troyer@phys.ethz.ch
On May 16, 2011, at 1:44 AM, Rachele Nerattini wrote:
Dear all,
I have a few questions too about simulations with spinmc application.
I'll make a specific example so I hope it'll be more clear... I have to run a MC classic simulation of the XY model defined on a square
lattice with periodic boundary conditions.
I wrote the following input file:
LATTICE_LIBRARY="lattices.xml" LATTICE="square lattice" T=0.892936869 J=1 THERMALIZATION=50000 SWEEPS=1000000 UPDATE="cluster" MODEL="XY" BOUNDARY type="periodic" S=1 {L=64;} {L=128;}
so that the simulation is performed for two different lattice sizes. I
did it following the instruction on the tutorials and it seemed to work. The questions are:
- Which are the initial conditions at the beginning of the simulation?
Are the spins spread randomly on the lattice?
No, they are all aligned - but it is easy to change that in the code.
- Is it possible to save the configurations which are used for the
calculations of the observables? I suppose the program generates a certain number of configurations then on one of them it takes the values of the observables and then it makes other pure dynamic MC steps and then it take again the measure of the observable on another configuration and so on...Is it correct? Can I save those configurations used for the accumulation of the observables?
Yes, it is possible, however this is not efficient. Using cluster updates generating a new configuration takes about as much time as reading it from disk.
- I also would like to know if is it already possible to compute the
density of vortix in this model...But I suppose this observable has to be added editing some of the spin mc files....
Yes, you need to add that to the code.
- If now I want to run the same simulation but for a three dimensional
simple cubic lattice is it correct to write
LATTICE="simple cubic lattice" instead of LATTICE="square lattice" as above?
Yes, that should work
Matthias
comp-phys-alps-users@lists.phys.ethz.ch