Re: [Phonopy-users] Primitive cell
Brought to you by:
atztogo
Menu
▾
▴
Re: [Phonopy-users] Primitive cell
|
From: Atz T. <atz...@gm...> - 2012-03-02 02:53:26
|
Hi, I think what you have done is correct. I have calculated TiO2 anatase. It seems the interaction range is large. I made 4x4x1 supercell of the tetragonal cell and the imaginary frequencies have disappeared. This can happen sometimes. I calculated firstly the elastic constants and see if the lattice stability condition is satisfied. If it is satisfied, the acoustic instability is considered wrongly calculated and I guess the supercell size may be not enough. Next I calculated phonon by 2x2x1 supercell and the unstable acoustic mode direction was found along the basal plane in the reciprocal space of the conventional unit cell. So I thought I don't need to expand in z-axis (and also c is enough long already). Togo On Thu, Mar 1, 2012 at 8:52 PM, Cereda, Silvia <sil...@kc...> wrote: > Thanks for your reply! Very kind. > > I think that the manual is really fine, but I had to check the source code for some clarifications. > Anyway, I think that some examples could be very very useful! > > I understand what you mean, very useful. Thanks! > But please can I ask you what you think about my current issue, since maybe I still miss some key points. > > I have an anatase cell. The primitive unit cell is > p1 = a 0.0 0.0 > p2 = 0.0 a 0.0 > p3 = a/2 a/2 c/2 > and contains 6 atoms. > > The conventional unit cell is orthorombic and contains 12 atoms > c1 = a 0.0 0.0 > c2 = 0.0 c 0.0 > c3 = 0.0 0.0 c > > When I ask phonopy to check the symmetry on the 12atoms POSCAR >> phonopy --symmetry > phonopy finds successfully the primitive matrix as p1, p2, p3 (in PPOSCAR), but the > Bravais lattice (BPOSCAR) turns out to be c1, c2, c3. > > Now, if I use the 12atoms cell (c1,c2,c3) and make the supercell (96atoms) > and on this 96 atoms supercell I want to compute the phonons which > primitive cell should I use? I would say ( [1 0 0],[0 1 0],[0.5 0.5 0.5]). > But, in this way the phonons are totally wrong! I mean when compared with the published one. > So, I think I am wrong. Is it possible that in this case the symmetry points are no more the ones > of the original symmetry group? > > If I instead use the 6atoms cell (p1,p2,p3) as initial configuration to make then the sueprcell, > I would use np.eye(3) as primitive_matrix. Am I correct? > > Now, since I keep getting a negative band in the acustic branch, can you suggest me something > that I can check. I converged all my results to hell (kpoint,cutoff,forces) and this frequency > is always negative. Since I tried with LDA/PAW in VASP, and also with Castep (using phonopy as a module) > and with some classical potentials as well, I cannot imagine there is a problem of convergency. > I more positive in thinking about a mistake I am doing in using phonopy. > > For all the systems I tried so far (all cubic or orthorombic) I have never had any problem, > so I am worried about the symmetry of this particular one. > > Thanks really a lot for any suggestion/idea. > It would be really appreciated. > Thanks > Silvia > > > ________________________________________ > From: Atz Togo [atz...@gm...] > Sent: 01 March 2012 03:10 > To: Cereda, Silvia > Cc: pho...@li... > Subject: Re: [Phonopy-users] Primitive cell > > Hi, > > I'm surprised at somebody who I don't know using the phonopy Python > module. Should I make the code cleaner and prepare a better manual? > >> When I use phonopy with VASP I do not have to define any primitive matrix for bands calculation. I assume that it is calculated by the code itself. Right? > > No. There is no option to search primitive lattice vectors in phonopy. > This is because there are infinite number of ways to choose primitive > axes. > > The definition of primitive_axis is shown here. > http://phonopy.sourceforge.net/setting-tags.html#primitive-axis > In the case using phonopy by calling phonopy command: > 1. Read POSCAR as unit cell. a_u, b_u, c_u are obtained from POSCAR. > 2. If you give PRIMITIVE_AXIS, then a_p, b_p, c_p are set according to > the definition. > 3. If you don't set PRIMITIVE_AXIS, a_p = a_u, b_p = b_u, c_p = c_u. > 4. Then all the calculations are made for a_p, b_p, c_p. > >> However when I use phonopy as a library I have to define explicitly the primitive cell. > > Yes at the post-process setting. > http://phonopy.sourceforge.net/phonopy-module.html#post-process > But if you only want to use the original unit cell, you can pass just > np.eye(3) for primitive_axes. I will make this as default in the next > version. > Currently, the code is as follows. > > def set_post_process(self, > primitive_matrix, > sets_of_forces=None, > force_constants=None, > is_nac=False): > > But this should be > > def set_post_process(self, > primitive_matrix=np.eye(3, dtype=float), > sets_of_forces=None, > force_constants=None, > is_nac=False): > >> Now, with cubic/ortorombic systems it is straightforward, but in non-cubic systems I am a little bit confused (mainly due to the rows/columns order). >> Since I am struggling with some negative frequencies (that should not be there) in the acustic band of a non-cubic system, I am worried that my primitive cell definition is wrong. >> >> So, is there a way, given the initial unit cell, to get the primitive cell by calling some phonopy libraries. >> Or, how can I check using phonopy with VASP which primitive cell is used for bands calculation. > > If the lattice of your unit cell follows a definition of, say, Bravais > lattice, then you can obtain a primitive matrix referring > International Tables of Crystallography Volume A, Part 5, > Transformations in Crystallography. If the initial unit cell is not > ordinary one, you can obtain a unit cell with a Bravais lattice > considering crystal symmetry as BPOSCAR using --symmetry option. > >> Second question, let's suppose that your system can be obtained by replicating two possible unit cells: a minimal non-cubic one, and a slightly larger cubic cell (twice atoms). >> Let's say that we prefer to use the cubic cell. Which primitive cell should we use in bands calculation? The one referring to the cubic system? > > Usually a primitive cell has to be set correctly following a > definition of Crystallography. One reason is that the special points > are defined for the proper primitive cells. You can see the shapes of > the first Brillouin zone at > http://www.cryst.ehu.es/cryst/get_kvec.html . > > For the creation of supercell, in most cases, to create supercell with > high symmetry and near isotropic. If you elongate in one direction, > often the crystal symmetry is broken in the supercell, this makes > calculations heavier. But this elongation is useful if you are > interested in a specific smaller k-point in a direction in reciprocal > space. > > Best regards, > > Togo > >> >> I hope that my questions are clear. >> >> Many thanks for any help you can provide! >> Thanks&Regards >> S >> ------------------------------------------------------------------------------ >> Virtualization & Cloud Management Using Capacity Planning >> Cloud computing makes use of virtualization - but cloud computing >> also focuses on allowing computing to be delivered as a service. >> http://www.accelacomm.com/jaw/sfnl/114/51521223/ >> _______________________________________________ >> Phonopy-users mailing list >> Pho...@li... >> https://lists.sourceforge.net/lists/listinfo/phonopy-users > > > > -- > Atsushi Togo > http://atztogo.users.sourceforge.net/ > atz...@gm... -- Atsushi Togo http://atztogo.users.sourceforge.net/ atz...@gm... |
