Re: [SIESTA-L] Need help with Phonon Dispersion Band Lines in Vibra
Dear Francisco, it is difficult to give a useful advice on the basis of very limited information you provide, but my impression is that your problems are not obviously related with Vibra. Some questions: 1. What (magnetic) structure are you modelling? How comes you have four atoms per AFM unit cell? Can there be two? 2. Is electronic structure (and band dispersions) correct, prior to any phonons? 3. What means "incorrect phonon dispersion"? Do you have problems with crystallography / choosing the q-path, or is your calculation basically wrong? 4. With 4 atoms as you use it so far, the Gamma phonon calculation would yield 9 modes, which would map genuine zone-center and zone-boundary modes. Do they come out reasonably? To your problem: "B asically I want to alter the band lines in input 2 so that they are equivalent to the band lines in input 1" - you have BandLinesScale pi/a in both inputs, the same lattice parameter, and the same definition of path. So if everything is correctly read, you must get the same Cartesian q-path in both cases. Either this is not so and there is something wrong with the input, or the paths are identical but your problem is elsewhere. Best regards Andrei - Le 29 Déc 22, à 0:40, garcia ff 000 a écrit : > Dear Users, > I have appended 2 Vibra inputs below for computing the phonon dispersion for > FCC > Mn. > Input 1 works fine as it gives the expected band shapes for the dispersion > (but > the frequencies are off). The main issue with input 1 is that it is not > suitable for antiferromagnetic calculations since there is only one Mn atom in > the primitive cell. > This led me to consider input 2, which has 4 atoms in the unit cell and can be > used for antiferromagnetic calculations. The issue with input 2 is that the > bandlines yield an incorrect phonon dispersion. This is what I need your help > on. Basically I want to alter the band lines in input 2 so that they are > equivalent to the band lines in input 1. > Any assistance with this, especially from the Vibra authors, would be greatly > appreciated. > Thank you very much for your kind assistance and God Bless! > Francisco > #INPUT 1 (1 atom in the FCC primitive cell; 125 atoms in Supercell) > SystemName fccMn_1 > SystemLabel fccMn_1 > NumberOfAtoms 1 > LatticeConstant 3.47 Ang > %block LatticeVectors > 0.50 0.50 0.00 > 0.50 0.00 0.50 > 0.00 0.50 0.50 > %endblock LatticeVectors > AtomicCoordinatesFormat Fractional > %block AtomicCoordinatesAndAtomicSpecies > 0.00 0.00 0.00 1 54.938 > %endblock AtomicCoordinatesAndAtomicSpecies > SuperCell_1 2 > SuperCell_2 2 > SuperCell_3 2 > AtomicDispl 0.04 Bohr > BandLinesScale pi/a > %block BandLines > 1 0.000 0.000 0.000 \Gamma > 30 2.000 0.000 0.000 X > 30 2.000 2.000 2.000 \Gamma > 30 1.000 1.000 1.000 L > %endblock BandLines > Eigenvectors True > #INPUT 2 (4 atoms in the FCC conventional cell; 108 atoms in Supercell) > SystemName fccMn_4 > SystemLabel fccMn_4 > NumberOfAtoms 4 > LatticeConstant 3.47 Ang > %block LatticeVectors > 1.00 0.00 0.00 > 0.00 1.00 0.00 > 0.00 0.00 1.00 > %endblock LatticeVectors > AtomicCoordinatesFormat Fractional > %block AtomicCoordinatesAndAtomicSpecies > 0.00 0.00 0.00 1 54.938 > 0.50 0.50 0.00 1 54.938 > 0.50 0.00 0.50 1 54.938 > 0.00 0.50 0.50 1 54.938 > %endblock AtomicCoordinatesAndAtomicSpecies > SuperCell_1 1 > SuperCell_2 1 > SuperCell_3 1 > AtomicDispl 0.04 Bohr > BandLinesScale pi/a > # The band lines below are incorrect. > %block BandLines > 1 0.000 0.000 0.000 \Gamma > 30 2.000 0.000 0.000 X > 30 2.000 2.000 2.000 \Gamma > 30 1.000 1.000 1.000 L > %endblock BandLines > Eigenvectors True > -- > SIESTA is supported by the Spanish Research Agency (AEI) and by the European > H2020 MaX Centre of Excellence > (https://urldefense.com/v3/__http://www.max-centre.eu/__;!!D9dNQwwGXtA!T-wl-ZvX-LX5xZC7QdfhJRIJ8Pmxo5HofWGt13XzKiGWhx9VgP3MXmjkoKcw2oYTy4STEEQIyWW5lU0aV4mzNGNhq7rtk7d9KA$ > ) -- SIESTA is supported by the Spanish Research Agency (AEI) and by the European H2020 MaX Centre of Excellence (http://www.max-centre.eu/)
Re: [SIESTA-L] Need help with Phonon Dispersion Band Lines in Vibra
Dear Prof. Postnikov, Special thanks again for your kind and prompt response. (i) The q-paths in input 2 must be rotated by 45 degrees to maintain consistency with the paths in input 1. I would like to kindly know the axis about which the rotation should be performed. (ii) Regarding your suggestion of using the lattice vectors [1/2 -1/2 0], [1/2 1/2 0], [0 0 1], I can't quite figure out why the q-paths will be the same as the q-paths for input 1. The primitive FCC lattice vectors in input 1 are [1/2 1/2 0], [1/2 0 1/2], and [0 1/2 1/2] (with an angle of 60 degrees between each pair of lattice vectors). The angle between the lattice vectors you proposed is 90 degrees for each pair. I'm having a difficulty grasping how this new set of lattice vectors yields the same q-paths as input 1. Any further explanation would be highly appreciated. Thank you very much Professor. On Fri, Dec 30, 2022 at 5:12 PM Andrei Postnikov < andrei.postni...@univ-lorraine.fr> wrote: > Dear Francisco, > so your AFM structure is of CuAu type. This is fine but of course > this is not the only AFM structure possible (and I don't know whether it is > realistic at all, but this of course depends on your objectives). > Now, if you want the q-path to be the same in your two settings, > you should consider that the second one is rotated by 45 degrees. > That is, if you choose the Gamma->X direction || [010] in the first > setting > it must be || [110] in the second setting, with the lattice vectors you > use. > Otherwise define the lattice vectors as [1/2 -1/2 0], [1/2 1/2 0], [0 0 1] > with the same lattice parameter as in the first setting > and enjoy the same coordinates of q points (cartesian, in terms of pi/a) > in both settings. > > Best regards > > Andrei > > > - Le 31 Déc 22, à 0:24, garcia ff 000 a > écrit : > > Dear Prof. Postnikov, > > Many thanks and appreciation for your response. I believe I found a > solution to my problem but I want to run it by you. > > First, an FCC cell with 2 unique atoms is equivalent to a tetragonal cell > (this is the smallest unit cell to model antiferromagnetism). > > Using the website > https://urldefense.com/v3/__https://www.materialscloud.org/work/tools/seekpath__;!!D9dNQwwGXtA!RtHnhiKf-1Tr0ZJZ17JGCt-WslBldkrwcANZ1KZuejXMskHtY6-_llZSytEsPLSiAktiz7DhWg1EBpjE-NmrWg$ > , the > high symmetry points in the Brillouin zone are as follows (each set of > points is scaled by the corresponding pi/a): > > Standard FCC primitive cell: Gamma (0,0,0), X(0,2,0), K(1.5,1.5,0), > W(1,2,0), L(1,1,1) > > 2-atom tetragonal cell: Gamma(0,0,0), X(0,1,0), M(1,1,0), R(0,1,0.707107), > A(1,1,0.707107), Z(0,0,0.707107). > > With this information, I believe the two Vibra inputs below, one for the > primitive FCC cell and the other for 2-atom tetragonal cell, are formally > equivalent (the last two k-points in each case, i.e. L and M, is what I'm a > bit unsure about). > > > Thank you very much for your kindness & happy holidays. > > > > (A) Primitive FCC cell: > > NumberOfAtoms 1 > > #Lattice parameters > LatticeConstant 3.47 Ang > %block LatticeVectors > 0.50 0.50 0.00 > 0.50 0.00 0.50 > 0.00 0.50 0.50 > %endblock LatticeVectors > > #Atomic positions > AtomicCoordinatesFormat Fractional > %block AtomicCoordinatesAndAtomicSpecies > 0.00 0.00 0.00 1 54.938 > %endblock AtomicCoordinatesAndAtomicSpecies > > #High symmetry Brillouin zones points scaled by pi/a: Gamma (0,0,0), > X(0,2,0), K(1.5,1.5,0), W(1,2,0), L(1,1,1) > > BandLinesScale pi/a > %block BandLines > 1 0.000 0.000 0.000 \Gamma > 30 0.000 2.000 0.000 X > 30 2.000 2.000 2.000 \Gamma > 30 1.000 1.000 1.000 L > %endblock BandLines > > > > (B) 2-atom tetragonal cell to model antiferromagnetism (this is double the > volume of the FCC primitive cell) > > NumberOfAtoms 2 > > #Lattice parameters > LatticeConstant 2.453660531 Ang #[this is the FCC lattice constant > divided by sqrt(2)] > %block LatticeVectors > 1.00 0.00 0.00 > 0.00 1.00 0.00 > 0.00 0.00 1.414213562 > %endblock LatticeVectors > > #Atomic positions > AtomicCoordinatesFormat Fractional > %block AtomicCoordinatesAndAtomicSpecies > 0.00 0.00 0.00 1 54.938 > 0.50 0.50 0.50 1 54.938 > %endblock AtomicCoordinatesAndAtomicSpecies > > #High symmetry Brillouin zones points scaled by pi/a: Gamma(0,0,0), > X(0,1,0), M(1,1,0), R(0,1,0.707107), A(1,1,0.707107), Z(0,0,0.707107) > > BandLinesScale pi/a > %block BandLines > 1 0.000 0.000 0.000 \Gamma > 30 0.000 1.000 0.000 X > 30 1.000 1.000 1.000 \Gamma > 30 2.000 2.000 2.000 M > %endblock BandLines > > > > On Thu, Dec 29, 2022 at 3:34 PM Andrei Postnikov < > andrei.postni...@univ-lorraine.fr> wrote: > >> Dear Francisco, >> it is difficult to give a useful advice on the basis of very limited >> information you provide, >> but my impression is that your problem
Re: [SIESTA-L] Need help with Phonon Dispersion Band Lines in Vibra
Dear Prof. Postnikov, Many thanks and appreciation for your response. I believe I found a solution to my problem but I want to run it by you. First, an FCC cell with 2 unique atoms is equivalent to a tetragonal cell (this is the smallest unit cell to model antiferromagnetism). Using the website https://urldefense.com/v3/__https://www.materialscloud.org/work/tools/seekpath__;!!D9dNQwwGXtA!WbTOEoZkczNXgxFdjZz1Ho66DZKcaHdWandSB0IytM7jNKQONZgfMHtxCird0d0mH_PsbycR0mostpvReHj7Iw$ , the high symmetry points in the Brillouin zone are as follows (each set of points is scaled by the corresponding pi/a): Standard FCC primitive cell: Gamma (0,0,0), X(0,2,0), K(1.5,1.5,0), W(1,2,0), L(1,1,1) 2-atom tetragonal cell: Gamma(0,0,0), X(0,1,0), M(1,1,0), R(0,1,0.707107), A(1,1,0.707107), Z(0,0,0.707107). With this information, I believe the two Vibra inputs below, one for the primitive FCC cell and the other for 2-atom tetragonal cell, are formally equivalent (the last two k-points in each case, i.e. L and M, is what I'm a bit unsure about). Thank you very much for your kindness & happy holidays. (A) Primitive FCC cell: NumberOfAtoms 1 #Lattice parameters LatticeConstant 3.47 Ang %block LatticeVectors 0.50 0.50 0.00 0.50 0.00 0.50 0.00 0.50 0.50 %endblock LatticeVectors #Atomic positions AtomicCoordinatesFormat Fractional %block AtomicCoordinatesAndAtomicSpecies 0.00 0.00 0.00 1 54.938 %endblock AtomicCoordinatesAndAtomicSpecies #High symmetry Brillouin zones points scaled by pi/a: Gamma (0,0,0), X(0,2,0), K(1.5,1.5,0), W(1,2,0), L(1,1,1) BandLinesScale pi/a %block BandLines 1 0.000 0.000 0.000 \Gamma 30 0.000 2.000 0.000 X 30 2.000 2.000 2.000 \Gamma 30 1.000 1.000 1.000 L %endblock BandLines (B) 2-atom tetragonal cell to model antiferromagnetism (this is double the volume of the FCC primitive cell) NumberOfAtoms 2 #Lattice parameters LatticeConstant 2.453660531 Ang #[this is the FCC lattice constant divided by sqrt(2)] %block LatticeVectors 1.00 0.00 0.00 0.00 1.00 0.00 0.00 0.00 1.414213562 %endblock LatticeVectors #Atomic positions AtomicCoordinatesFormat Fractional %block AtomicCoordinatesAndAtomicSpecies 0.00 0.00 0.00 1 54.938 0.50 0.50 0.50 1 54.938 %endblock AtomicCoordinatesAndAtomicSpecies #High symmetry Brillouin zones points scaled by pi/a: Gamma(0,0,0), X(0,1,0), M(1,1,0), R(0,1,0.707107), A(1,1,0.707107), Z(0,0,0.707107) BandLinesScale pi/a %block BandLines 1 0.000 0.000 0.000 \Gamma 30 0.000 1.000 0.000 X 30 1.000 1.000 1.000 \Gamma 30 2.000 2.000 2.000 M %endblock BandLines On Thu, Dec 29, 2022 at 3:34 PM Andrei Postnikov < andrei.postni...@univ-lorraine.fr> wrote: > Dear Francisco, > it is difficult to give a useful advice on the basis of very limited > information you provide, > but my impression is that your problems are not obviously related with > Vibra. > Some questions: > 1. What (magnetic) structure are you modelling? How comes you have four > atoms per AFM unit cell? > Can there be two? > 2. Is electronic structure (and band dispersions) correct, prior to any > phonons? > 3. What means "incorrect phonon dispersion"? Do you have problems with > crystallography / > choosing the q-path, or is your calculation basically wrong? > 4. With 4 atoms as you use it so far, the Gamma phonon calculation would > yield > 9 modes, which would map genuine zone-center and zone-boundary modes. > Do they come out reasonably? > > To your problem: > "Basically I want to alter the band lines in input 2 so that they are > equivalent to the band lines in input 1" - > you have > BandLinesScale pi/a > in both inputs, the same lattice parameter, and the same definition of > path. > So if everything is correctly read, you must get the same Cartesian q-path > in both cases. > Either this is not so and there is something wrong with the input, > or the paths are identical but your problem is elsewhere. > > Best regards > > Andrei > > > > > > > > > - Le 29 Déc 22, à 0:40, garcia ff 000 a > écrit : > > Dear Users, > > I have appended 2 Vibra inputs below for computing the phonon dispersion > for FCC Mn. > > Input 1 works fine as it gives the expected band shapes for the dispersion > (but the frequencies are off). The main issue with input 1 is that it is > not suitable for antiferromagnetic calculations since there is only one Mn > atom in the primitive cell. > > This led me to consider input 2, which has 4 atoms in the unit cell and > can be used for antiferromagnetic calculations. The issue with input 2 is > that the bandlines yield an incorrect phonon dispersion. This is what I > need your help on. Basically I want to alter the band lines in input 2 so > that they are equivalent to the band lines in input 1. > > Any assistance with this, especially from the Vibra authors, would be > greatly
Re: [SIESTA-L] Need help with Phonon Dispersion Band Lines in Vibra
Dear Francisco, so your AFM structure is of CuAu type. This is fine but of course this is not the only AFM structure possible (and I don't know whether it is realistic at all, but this of course depends on your objectives). Now, if you want the q-path to be the same in your two settings, you should consider that the second one is rotated by 45 degrees. That is, if you choose the Gamma->X direction || [010] in the first setting it must be || [110] in the second setting, with the lattice vectors you use. Otherwise define the lattice vectors as [1/2 -1/2 0], [1/2 1/2 0], [0 0 1] with the same lattice parameter as in the first setting and enjoy the same coordinates of q points (cartesian, in terms of pi/a) in both settings. Best regards Andrei - Le 31 Déc 22, à 0:24, garcia ff 000 a écrit : > Dear Prof. Postnikov, > Many thanks and appreciation for your response. I believe I found a solution > to > my problem but I want to run it by you. > First, an FCC cell with 2 unique atoms is equivalent to a tetragonal cell > (this > is the smallest unit cell to model antiferromagnetism). > Using the website [ > https://urldefense.com/v3/__https://www.materialscloud.org/work/tools/seekpath__;!!D9dNQwwGXtA!VCzh9W4S1t5nhfeK_65w_ZsZpJauei8vdCoYcoysbxbXQ6kbxNBuSTzR-LciHx145nkwK_JGfqplTyD_aZeQ8icIeWJtZ4jCPg$ > | > https://urldefense.com/v3/__https://www.materialscloud.org/work/tools/seekpath__;!!D9dNQwwGXtA!VCzh9W4S1t5nhfeK_65w_ZsZpJauei8vdCoYcoysbxbXQ6kbxNBuSTzR-LciHx145nkwK_JGfqplTyD_aZeQ8icIeWJtZ4jCPg$ > ] , the high symmetry points > in the Brillouin zone are as follows (each set of points is scaled by the > corresponding pi/a): > Standard FCC primitive cell: Gamma (0,0,0), X(0,2,0), K(1.5,1.5,0), W(1,2,0), > L(1,1,1) > 2-atom tetragonal cell: Gamma(0,0,0), X(0,1,0), M(1,1,0), R(0,1,0.707107), > A(1,1,0.707107), Z(0,0,0.707107). > With this information, I believe the two Vibra inputs below, one for the > primitive FCC cell and the other for 2-atom tetragonal cell, are formally > equivalent (the last two k-points in each case, i.e. L and M, is what I'm a > bit > unsure about). > Thank you very much for your kindness & happy holidays. > (A) Primitive FCC cell: > NumberOfAtoms 1 > #Lattice parameters > LatticeConstant 3.47 Ang > %block LatticeVectors > 0.50 0.50 0.00 > 0.50 0.00 0.50 > 0.00 0.50 0.50 > %endblock LatticeVectors > #Atomic positions > AtomicCoordinatesFormat Fractional > %block AtomicCoordinatesAndAtomicSpecies > 0.00 0.00 0.00 1 54.938 > %endblock AtomicCoordinatesAndAtomicSpecies > #High symmetry Brillouin zones points scaled by pi/a: Gamma (0,0,0), X(0,2,0), > K(1.5,1.5,0), W(1,2,0), L(1,1,1) > BandLinesScale pi/a > %block BandLines > 1 0.000 0.000 0.000 \Gamma > 30 0.000 2.000 0.000 X > 30 2.000 2.000 2.000 \Gamma > 30 1.000 1.000 1.000 L > %endblock BandLines > (B) 2-atom tetragonal cell to model antiferromagnetism (this is double the > volume of the FCC primitive cell) > NumberOfAtoms 2 > #Lattice parameters > LatticeConstant 2.453660531 Ang #[this is the FCC lattice constant divided by > sqrt(2)] > %block LatticeVectors > 1.00 0.00 0.00 > 0.00 1.00 0.00 > 0.00 0.00 1.414213562 > %endblock LatticeVectors > #Atomic positions > AtomicCoordinatesFormat Fractional > %block AtomicCoordinatesAndAtomicSpecies > 0.00 0.00 0.00 1 54.938 > 0.50 0.50 0.50 1 54.938 > %endblock AtomicCoordinatesAndAtomicSpecies > #High symmetry Brillouin zones points scaled by pi/a: Gamma(0,0,0), X(0,1,0), > M(1,1,0), R(0,1,0.707107), A(1,1,0.707107), Z(0,0,0.707107) > BandLinesScale pi/a > %block BandLines > 1 0.000 0.000 0.000 \Gamma > 30 0.000 1.000 0.000 X > 30 1.000 1.000 1.000 \Gamma > 30 2.000 2.000 2.000 M > %endblock BandLines > On Thu, Dec 29, 2022 at 3:34 PM Andrei Postnikov < [ > mailto:andrei.postni...@univ-lorraine.fr | andrei.postni...@univ-lorraine.fr ] > > wrote: >> Dear Francisco, >> it is difficult to give a useful advice on the basis of very limited >> information >> you provide, >> but my impression is that your problems are not obviously related with Vibra. >> Some questions: >> 1. What (magnetic) structure are you modelling? How comes you have four atoms >> per AFM unit cell? >> Can there be two? >> 2. Is electronic structure (and band dispersions) correct, prior to any >> phonons? >> 3. What means "incorrect phonon dispersion"? Do you have problems with >> crystallography / >> choosing the q-path, or is your calculation basically wrong? >> 4. With 4 atoms as you use it so far, the Gamma phonon calculation would >> yield >> 9 modes, which would map genuine zone-center and zone-boundary modes. >> Do they come out reasonably? >> To your problem: >> "B asically I want to alter the band lines in input 2 so that they are >> equivalent to the band lines in input 1" - >> you have >> BandLinesScale pi/a >> in both inputs, the same lattice parameter, and the same
