I have a system with S_2 symmetrisation. When I perform a dimer
calculation, I found the sentence in OUTCAR:
[i]IMPORTANT INFORMATION: All symmetrisations will be switched off!
NOSYMM: (Re-)initialisation of all symmetry stuff for point group C_1.[/i]
After the dimer calculation, the structures of the two surfaces of system
become different, while in the initial system the two surfaces have the
same structure for the system has S_2 symmetrisation.
So I want to know what results would be induced when all
symmetrisations will be switched off?
Thanks a lot.
Why DIMER method switches off the symmetrisations of system?
Moderator: moderators
I don't think it ever makes sense to apply symmetry for a dimer calculation. We recommend always setting ISYM=0 for these calculations.
Imposing symmetry can make sense for a minimum structure of known symmetry. If you are searching for a saddle, however, it is unlikely that you know it is a symmetric structure. Also, symmetry can confuse the finite difference calculation in the dimer method. If the two dimer images are close enough to a symmetric structure, the symmetry will by imposed, and the finite difference algorithm will fail. If they are not near a symmetric structure, symmetry can not be applied anyways. The safe approach is to turn off symmetry.
Imposing symmetry can make sense for a minimum structure of known symmetry. If you are searching for a saddle, however, it is unlikely that you know it is a symmetric structure. Also, symmetry can confuse the finite difference calculation in the dimer method. If the two dimer images are close enough to a symmetric structure, the symmetry will by imposed, and the finite difference algorithm will fail. If they are not near a symmetric structure, symmetry can not be applied anyways. The safe approach is to turn off symmetry.
Thanks for your reply.
But I want to know which surface structure of the system I should select when the DIMER running ends? That means the structures of the two surface are different, which one is more reasonable?
In my work, I use each surface to construct a system with S_2 symmetrisation to perform a static VASP running, to see which one is favorable energetically, and then select it for the structural analyse. Whether my method is right?
But I want to know which surface structure of the system I should select when the DIMER running ends? That means the structures of the two surface are different, which one is more reasonable?
In my work, I use each surface to construct a system with S_2 symmetrisation to perform a static VASP running, to see which one is favorable energetically, and then select it for the structural analyse. Whether my method is right?
Ok, I think I see what you mean. In the old dimer code, there are two images stored in the directories 01 and 02. These images are separated by a small distance (DNdR=0.005 Ang by default). When the dimer converges to a saddle, the best estimate of the saddle is 1/2 way between these two images. You can obtain this image using our interpolate script:
interpolate.pl 01/CONTCAR 02/CONTCAR 0.5
If you need to impose symmetry on this configuration, you're on your own.
Note: in the new v2.0 of the vtst code, the dimer uses a single image, and the current location is saved in the CENTCAR file.
interpolate.pl 01/CONTCAR 02/CONTCAR 0.5
If you need to impose symmetry on this configuration, you're on your own.
Note: in the new v2.0 of the vtst code, the dimer uses a single image, and the current location is saved in the CENTCAR file.
Thanks a lot, Graeme!
But I mean that I want to study an adsorption system, and I use slab model to construct its configuration, and the adsorbed atoms are adsorbed on each side(surface) of the slab, so the system symmetry is S_2 to ensure the two sides have the same structures. But DIMER method needs to switch the symmetry off, therefore after DIMER calculations, the structures of the two side become different!
So I want to know which side should be chosed ?
But I mean that I want to study an adsorption system, and I use slab model to construct its configuration, and the adsorbed atoms are adsorbed on each side(surface) of the slab, so the system symmetry is S_2 to ensure the two sides have the same structures. But DIMER method needs to switch the symmetry off, therefore after DIMER calculations, the structures of the two side become different!
So I want to know which side should be chosed ?
Ok, I finally understand what you mean.
I don't think the dimer is going to work for this kind of symmetry. If the point of having adsorbates on each side of the slab is to zero the dipole, both adsorbates would have to be in a saddle point configuration. This structure is a second order saddle point, which is dimer is not designed to find.
You could leave one adsorbate in a minimum and find a saddle with the other, but this would defeat the purpose. Unless there is another good reason to use mirror symmetry, I would recommend using a single adsorbate, and turn on the dipole corrections.
I don't think the dimer is going to work for this kind of symmetry. If the point of having adsorbates on each side of the slab is to zero the dipole, both adsorbates would have to be in a saddle point configuration. This structure is a second order saddle point, which is dimer is not designed to find.
You could leave one adsorbate in a minimum and find a saddle with the other, but this would defeat the purpose. Unless there is another good reason to use mirror symmetry, I would recommend using a single adsorbate, and turn on the dipole corrections.