Re: [SIESTA-L] Regarding SCF convergence issue during optimization
Hi, you provide too few information. Of course just increasing the number of iterations won't bring you further if there is no convergence. General advises might be as follows: - Check that you are using diagonalisation and NOT order-N scheme before you fix the things and "know what you are doing". With 235 atoms the chances are, you are in order-N regime by default, and things go astray. - Check your system visually (using visualisation tools from your working XV file) whether its shape is as expected; a huge number of problems result from an error in defining the structure. - Check whether your electronic structure makes sense (the main bands are more or less where they are expected) at very first iterations, before the calculations start to markedly diverge. Do not just look at the convergence, but at other things in the output file. Make sure that the attribution of valence states in the pseudopotential and in the basis matches. Some inconsistencies may cry aloud in the output file, but you need to pay attention... - Reduce the mixing parameter. Reduce it drastically if needed (0.001 is not uncommon in some cases). Start converging with large broadening (electronic temperature), say 600 K, and then reduce it gradually as you enter into a convergence regime. The actual behaviour may depend on whether your system is isolated or periodic, insulator or metal. If your case is periodic and metallic, you may need many k-points as well (in the directions of periodicity, of course). Good luck Andrei Postnikov - Le 19 Avr 24, à 4:16, Jyotirmoy Deb a écrit : > Dear Sir/Madam, > I want to optimize a heterostructure consisting of 235 atoms. But I am facing > an > error every time "SCF did not converge in the maximum number of steps". I have > gradually increased the "MaxSCFIterations" however, I have not been able to > solve this issue. Please help me in this regard. > Thanking you > with regards > Jyotirmoy > Dr. Jyotirmoy Deb > C/O Dr. G. Narahari Sastry & Dr. Lakshi Saikia > DST-SERB National Post-Doctoral Fellow > Advanced Computation and Data Sciences Division (ACDSD) > CSIR-North East Institute of Science & Technology, Jorhat-785006, Assam, > India. > Ph. No: +917002140643/ +919435589869 > Email: [ mailto:deb.jyotirmo...@gmail.com | deb.jyotirmo...@gmail.com ] > Webpage: [ > https://urldefense.com/v3/__https://sites.google.com/view/jyotirmoydftphy__;!!D9dNQwwGXtA!QXMXS7L8cCba3hBh8KgcbwyC9C-E5X98KJTAXUguv0CMpKRil5P5cMWK5ziJhwU4-c14ht8eejcbtj_czqW_iUBo$ > | > https://urldefense.com/v3/__https://sites.google.com/view/jyotirmoydftphy__;!!D9dNQwwGXtA!TlNGLjwuIHLvaLUEXMAzZneDEOzZtFJFMRfYAnBQADgcqVNh1tJuG4DMBR8sWhbOVnrRtR-96vIssOhkbvkqvQKGf5LpgmFIQA$ > ] > -- > 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!TlNGLjwuIHLvaLUEXMAzZneDEOzZtFJFMRfYAnBQADgcqVNh1tJuG4DMBR8sWhbOVnrRtR-96vIssOhkbvkqvQKGf5IASVdD0A$ > ) -- 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] Problem using Vibra utility
Dear Diego, as your case is q=0 only, you can try your luck with my shortcut version of vibra, to be compiled from the attached zip. It does not need an .fdf file, just .XV and .FC (.XV serves just to identify the atoms; exact coordinates are not important). Tell me if you'd encounter any difficulties. Best regards Andrei Postnikov - Le 12 Sep 23, à 12:16, Diego Lopez Alcala diego.lo...@uv.es a écrit : > Dear Siesta users, > > I have been trying to compute the modes of vibrations of a molecule adsorbed > on > a semiconducting monolayer, but I am having some problems. First I run this > input: > > # General System descriptors > > SystemName vibra-2 # Descriptive name of the system > SystemLabel vibra-2# Short name for naming files > > NumberOfAtoms 164 # Number of atoms > NumberOfSpecies 5 # Number of species > > md.typeofrun fc > > MD.FCFirst 151 > MD.FCLast 164 > > AtomicCoordinatesFormat NotScaledCartesianAng > > MM.UnitsDistance Ang # what this program prints out DO NOT CHANGE > MM.UnitsEnergyeV # what this program prints out DO NOT CHANGE > MM.Grimme.S6 0.75 # Grimme-paper for PBE (correct for your functional) > MM.Grimme.D 20. # Grimme-paper (correct for your functional) > %block MM.Potentials > 1 1 Grimme111.94 3.124 # Cr, 10.1002/jcc.20495 > 1 2 Grimme 80.39 3.245 # Cr / S > 1 3 Grimme120.28 3.311 # Cr / Br > 1 4 Grimme 45.06 3.014 # Cr / C > 1 5 Grimme 12.74 2.563 # Cr / H > 2 2 Grimme 57.73 3.366 # S, 10.1002/jcc.20495 > 2 3 Grimme 86.38 3.432 # S / Br > 2 4 Grimme 32.36 3.135 # S / C > 2 5 Grimme 9.15 2.684 # S / H > 3 3 Grimme129.24 3.498 # Br, 10.1002/jcc.20495 > 3 4 Grimme 48.42 3.201 # Br / C > 3 5 Grimme 13.69 2.750 # Br / H > 4 4 Grimme 18.14 2.904 # C, 10.1002/jcc.20495 > 4 5 Grimme 5.13 2.453 # C / H > 5 5 Grimme 1.45 2.002 # H, 10.1002/jcc.20495 > %endblock MM.Potentials > > %block Chemical_Species_Label > 1 24 Cr > 2 16 S > 3 35 Br > 46 C > 51 H > %endblock Chemical_Species_Label > > PAO.BasisSize SZ > > DFTU.ProjectorGenerationMethod 1 > DFTU.PotentialShift .true. > > %block DFTU.Proj # Define DFTU projectors > Cr 1 # Label, l_shells > n=3 2 # n (opt if not using semicore levels),l,Softconf(opt) > 4.00 0.0 # U(eV), J(eV) for this shell > 0.0 # rc (Bohr), \omega(Bohr) (Fermi cutoff function) > %endblock DFTU.Proj > > # Lattice, coordinates, k-sampling > > AtomicCoorFormatOut Ang > > %block AtomicCoordinatesAndAtomicSpecies > 4.44717269 3.68308271 0.75902034 3 79.904 > . > . (Rest of atomic coordinates) > . > 5.10327585 14.06972694 8.19442056 5 1.007 > %endblock AtomicCoordinatesAndAtomicSpecies > > LatticeConstant 1.0 Ang > > %block LatticeVectors > 17.8287990.000.00 >0.00 24.3147200.00 >0.000.00 25.236237 > %endblock LatticeVectors > > %block kgrid_Monkhorst_Pack > 1 0 0 0.0 > 0 1 0 0.0 > 0 0 1 0.0 > %endblock kgrid_Monkhorst_Pack > > # DFT, Grid, SCF > > XC.functional GGA # Exchange-correlation functional type > XC.authors PBE # Particular parametrization of xc func > SpinPolarized .true. # Spin unpolarized calculation > MeshCutoff 400. Ry # Equivalent planewave cutoff for the grid > MaxSCFIterations300 # Maximum number of SCF iterations per > step > SCF.DM.Converge true > SCF.H.Converge true > > # Eigenvalue problem: order-N or diagonalization > > SolutionMethod diagon # OrderN or Diagon > ElectronicTemperature 300 K# Temp. for Fermi smearing > > # Output options > > WriteCoorInitial > WriteCoorStep > WriteForces > WriteMullikenPop1 # Write Mulliken Population Analysis > WriteCoorXmol .false. > WriteMDCoorXmol .false. > WriteMDhistory .false. > WriteCoorXmol .false. > > Once it is done I add the following lines to the input (as suggested here: > https://urldefense.com/v3/__https://docs.siesta-project.org/projects/siesta/en/latest/tutorials/basic/vibrational-properties/Benzene/index.html__;!!D9dNQwwGXtA!UewEzyC8ZoPpI3i_PmoF3Vbbcs86V_CEf3F-hofbAGex6SS1dlsBvcgSPg7HdC2VlFmiZZGfJEXGqNlpcyY$ > ) and I run the vibra utility, obtaining this error message. The > SystemLabel.bands is
Re: [SIESTA-L] Radial part of basis orbital
Isn't this information explicitly included in .ion files - after a general header, one basis function after the other? You just need to cut whichever you want and feed it to a plotting program... Best regards, A.P. ... # PAOs:__ 0 2 1 0 2.00 #orbital l, n, z, is_polarized, population 500 0.132468670984E-01 6.61018668210 # npts, delta, cutoff 0.000 0.287172515330 0.132468670984E-01 0.287187021634 0.264937341968E-01 0.287232752566 0.397406012952E-01 0.287308980728 0.529874683936E-01 0.287415720365 ... - Le 15 Fév 23, à 15:34, Emilio Artacho a écrit : > Check the WriteIonPlotFiles option > E >> On 14 Feb 2023, at 06:17, Francisco Garcia < [ >> mailto:garcia.ff@gmail.com | >> garcia.ff@gmail.com ] > wrote: >> Dear Users, >> I would like to know how to obtain the data for the radial part of a given >> basis >> orbital for plotting. E.g. the radial part of Al 3s and Al 3p for SZ basis. >> Thank you. >> -- >> 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!S0apBc_vxFM3XsCbZ2x_Ek6V_jNWS_G7uxbf6fayBqjsXL1bU6GTY35UzarXF3pSaMhFMFRLKKd94oJ0dFV9$ >> | >> https://urldefense.com/v3/__http://www.max-centre.eu/__;!!D9dNQwwGXtA!W61yKVCs4FflUUO_VbPHAuO0S3HZ8Xzc17Dr0DE6UJW862iIwupaYKBiGgBCBcoIWd0sdJ20S4blr36ZkIE3e-aI_Z3qKyHZtQ$ >> ] ) > -- > Emilio Artacho > Theory Group, Nanogune, 20018 San Sebastian, Spain, and > Theory of Condensed Matter, Department of Physics, > Cavendish Laboratory, University of Cambridge, Cambridge CB3 0HE, UK > -- > 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!W61yKVCs4FflUUO_VbPHAuO0S3HZ8Xzc17Dr0DE6UJW862iIwupaYKBiGgBCBcoIWd0sdJ20S4blr36ZkIE3e-aI_Z3qKyHZtQ$ > ) -- 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] Restarting calculations with different spin orientations
Dear Andres Tellez, my DMtune tool was written having similar tasks in mind; please feel free to download it from [ https://urldefense.com/v3/__https://www.home.uni-osnabrueck.de/apostnik/download.html__;!!D9dNQwwGXtA!UOZi7W_dTZOE_UPIXqgD4MdGacy1Fx8Kvs3HymrL_qKWalh8UtdNSwuKc4JOVzfzoPLVLIAZ5uHSbQoT-gAurVYvIzKoJQhPng$ | https://urldefense.com/v3/__https://www.home.uni-osnabrueck.de/apostnik/download.html__;!!D9dNQwwGXtA!UOZi7W_dTZOE_UPIXqgD4MdGacy1Fx8Kvs3HymrL_qKWalh8UtdNSwuKc4JOVzfzoPLVLIAZ5uHSbQoT-gAurVYvIzKoJQhPng$ ] and test whether it still works and applies to your case. There is no special documentation, but minimal explanations are included at the beginning of the source file dmtune.f. In case of problems it should not be difficult to modify, or pass me a word (with an example), I'll try to update. Best regards Andrei Postnikov - Le 31 Jan 23, à 18:14, Andres Tellez Mora a écrit : > Dear Siesta users and developers. > I am running calculations of the same structure with different spin > orientations. Since I am performing spin-orbit calculations with DFT+U, they > can take a decent amount of time to converge. Hence, I was trying to use the > .DM file of one of the calculations to start the others; however, the > DM.InitSpin block information is overridden and the calculation starts using > the same spin orientation from the converged .DM file. This even happens when > using a .DM file from a non-polarized calculation, which gives a starting > magnetization of 0.0. Is it possible to use the information of a .DM file and > start with a given spin orientation simultaneously? If this is not possible, > what else could I do to improve the convergence? I appreciate any help you can > provide. > Best Regards, > Andres Tellez. > -- > 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!UOZi7W_dTZOE_UPIXqgD4MdGacy1Fx8Kvs3HymrL_qKWalh8UtdNSwuKc4JOVzfzoPLVLIAZ5uHSbQoT-gAurVYvIzK9M_8LNA$ > ) -- 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 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 >>
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]
Dear Amal Yassin: Your unit cell vectors are all zeros; your atom belongs to specie Nr zero (must be 1), and there seems to be a problem with its coordinate. All this is evident from your output file if you read it beyond the first line. The absence of XV file is just a warning; it's no problem if the structure is correctly defined in the input file. Good luck, Andrei Postnikov - Le 18 Aoû 22, à 22:18, Amal Yassin a écrit : > Hello > When i run the simulation of Carbone , i existe this error . > Any Idea about this error , and how can resolve it ?! > Please Help.. > Best regard > siesta: WARNING: XV file not found > siesta: Atomic coordinates (Bohr) and species > siesta: 0.0** 3.77945 0 1 > siesta: Automatic unit cell vectors (Ang): > siesta: 0.00 0.00 0.00 > siesta: 0.00 0.00 0.00 > siesta: 0.00 0.00 0.00 > rcut: Wrong species 0. Have 1 > Stopping Program from Node: 0 > #0 0x7f0bf0cdbd21 in ??? > #1 0x7f0bf0cdc7ad in ??? > #2 0x7f0bf0f3038c in ??? > #3 0x55c259359aca in ??? > #4 0x55c259098e83 in ??? > #5 0x55c25901ed25 in ??? > #6 0x55c25904bd0b in ??? > #7 0x55c25904a277 in ??? > #8 0x55c258f8f74c in ??? > #9 0x7f0bf0957082 in __libc_start_main > at ../csu/libc-start.c:308 > #10 0x55c258f8f81d in ??? > #11 0x in ??? > -- > SIESTA is supported by the Spanish Research Agency (AEI) and by the European > H2020 MaX Centre of Excellence (http://www.max-centre.eu/) -- 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] Forces on crystalline atoms are not zero
Dear Francisco:as a general statement, your expectation is justified. However, unexpected things happen.Having posted an (input) / output file would help to pinpoint a problem. Without any additional information, I'd guess that your lattice vectors are not exactly fcc, or the atoms are not exactly at symmetric positions. (E.g., you think that you scalethe atom coordinates with lattice parameter, but in fact you don't).Best regardsAndrei Postnikov - Francisco Garcia a écrit : >Dear Users I performed a series of single point energy calculations on an FCC crystal by varying the lattice constant. I'm trying to generate data points to optimize the lattice constant I was expecting the force on each atom to zero by virtue of crystalline symmetry (no internal degree of freedom). But to my surprise, the forces are not zero. Increasing the Mesh cutoff from 300 Ry to higher values didn't help either. How is it possible for the atomic forces in a crystal to be non-zero? -- 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] phdos
Right.Sorry if this was not clear enough from the documentation:my "phdos" is intended to be used on Gamma phonon calculated on a (probably large) cell,that is, for a case when the number of atoms is the .XV file is the same asthe number of atoms for which eigenvectors are calculated.The idea is, to broaden a discrete spectrum of (presumably many) linesof Gamma-vibration frequencies, not to integrate the density of modes fromthe phonon dispersions over the Brillouin zone.In your case, the .XV file contains 125 atoms (5x5c1 supercell à 5 atoms)but the eigenvectors are calculated for 5 atoms only (and this over a path of many q-points).Ghe "phdos" identifies 125 atoms in the .XV file and attempts to read the eigenvectors for all of thembut finds the data for 5 atoms only, hence the error message and termination.So such behaviour is as expected.In order to solve exactly your problem I started to work some time ago on an auxiliary code which expandsa "partial" FC file into a "full" FC file (which would result from displacing all atoms in the supercelland not only "prototype" ones in the original single cell). It was not extensively testedand does not immediately work on our example although in principle it should; I'll check what is the problem and let you know.Best regardsAndrei Postnikov - AK- HYDRA a écrit : >Hello sir/ma'amHere, I got same issue every time i.e., Error reading real part >of eigenvector 1 iat=6while using 'phdos' available on github within >'Lattice_dynamics'.I am going to attach my .vector, .FC, *.fdf, and .XV file, >need suggestion please thank you -- -- 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] strange force in phonon calculation
Dear Ziba,it seems that you are randomly trying different options.Try to be more systematic.MeshCutoff of 870 Ry seems to be an overkill -but certainly not an error; if your tests show that you need as much as that, so be it.Did you have a look at your structures with visualisation tools - is everything as it should be, no obvious structure error?If your imaginary frequency is genuine and reproducible,I'd look at the eigenvector of this mode, give a small displacementto atoms along this eigenvector, and make the relaxation again. The total energy ought to be lower than in your initial starting point;this is the idea of an imaginary frequency: you displace the atomswithin the corresponding mode, and the energy goes down.Hence you were not at the equilibrium; try better.No way around.Best regardsAndrei Postnikov - Ziba Torkashvand a écrit : > Thank you Professor Postnikov,Here I will provide two examples of my workI have a hexagonal structure containing 18 atoms1. I have substituted one atom with a gaseous one and relaxed it up to 0.000794 eV/A and 0.00458798 kBarthen I calculated the phonon and fortunately, there is no negative frequency while the force for the first step is approximated as 0.008174 eV/A 2. I have substituted two atoms with two gaseous ones and relaxed it up to 0.000549 eV/A and 0.00290065 kBarthen I calculated the phonon and unfortunately, there are frequencies as -206.30292747012965 while the force for the first step is approximated as 0.015862 eV/A For both cases, I have used same inputs asFor relaxation:MeshCutoff870 Ry DM.MixingWeight 0.02 DM.NumberPulay 3 %block kgrid_Monkhorst_Pack 6 0 0 0.0 0 6 0 0.0 0 0 1 0.0 %endblock kgrid_Monkhorst_Pack For phonon: MeshCutoff750 Ry DM.MixingWeight 0.02 DM.NumberPulay 8 %block kgrid_Monkhorst_Pack 3 0 0 0.0 0 3 0 0.0 0 0 1 0.0 %endblock kgrid_Monkhorst_Pack I repeated phonon calculation for the second case with %block kgrid_Monkhorst_Pack 6 0 0 0.0 0 6 0 0.0 0 0 1 0.0 %endblock kgrid_Monkhorst_Pack and the resulting force for the first step is 0.013928 eV/A I'm confused about what is happening in these two similar examples. -- 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] strange force in phonon calculation
Dear Ziba,probably you'll need to explain better what's your problem.> Once I'm relaxing my structure up to 0.0004 eV/Ang after the first MD step of phonon calculation > without any displacement the maximum force will be 0.002 eV/Ang "The first step without any displacement" is just calculation of your relaxed structure.If it is as you say then either it was not fully relaxed, or you changed some parametersbetween the two calculations. (A guess: either the coordinates are not those from the relaxed case,or the density matrix taken was not converged).> but still there is no negative frequency.- what you mean, "still"? And how do you know "there is no negative frequency" if you've just done "the first step without any displacement"?> This time I've relaxed my structure up to 0.0002 eV/Ang but in the first step I'm receiving 0.015 eV/Ang- see above> which is leading to the negative frequencies.- for what I know, negative frequencies are are hint that the starting point for a phonon calculationwas not at the local minimum, i.e., the relaxation was not good. How negative are your frequencies?> In the successful phonon calculations I just lowered the number of kpoint in the Monkhorst pack for phonon calculation > but this time even the kpoints are the same as relaxation input.- reducing the k-mesh is certainly not a working recipe for getting phonons right; you've just had luck.Best regardsAndrei Postnikov - Ziba Torkashvand a écrit : > Hello everyone,I'm trying to do phonon calculation for some hexagonal structures.Once I'm relaxing my structure up to 0.0004 eV/Ang after the first MD step of phonon calculation without any displacement the maximum force will be 0.002 eV/Ang but still there is no negative frequency.This time I've relaxed my structure up to 0.0002 eV/Ang but in the first step I'm receiving 0.015 eV/Ang which is leading to the negative frequencies.In the successful phonon calculations I just lowered the number of kpoint in the Monkhorst pack for phonon calculation but this time even the kpoints are the same as relaxation input. Any help would be appreciated ThanksZiba -- 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] Query regarding charge calculation using SCF charge density difference
Dear Harkishan,I think Macroave (in the Util/ of the Siesta distribution)might be able to do something similar to what you need.Otherwise, I've a simple script (attached) intended to integrate(in fact, simply to sum up) the charge density given on the grid over spherescentered on atoms. The idea was to get a fair comparison with chargesestimated in other ("muffin-tin") codes. This worked fairy well even if not very fast.You can modify it to integrate over a user-defined volume.The algorithm simply runs over all points of the grid to checkwhether they fall within the defined shape.Hopefully this may help,best regardsAndrei Postnikov - Harkishan Dua a écrit : >I have a bilayer system which is supposed to act as a nanoscale capacitor. I >have performed the calculations using siesta where I have calculated the >charge stored in the capacitor from the dipole moment. But after communicating >the paper, the reviewers suggested getting the charge stored in the capacitor >using SCF charge density difference. Please guide me on how to get the charge >stored in the capacitor using SCF charge density difference in siesta. I would >be really grateful for any help in this regard. Eagerly waiting for your reply Regards Harkishan Dua -- Harkishan DuaPhD StudentDepartment of PhysicsAssam University, Silchar.C Cgrdint, a script to integrate 3-dim grid function C(i.e. LDOS, RHO, DRHO, etc.) written in SIESTA by subr. iorho C over a region of choice (a sphere centered at a given atom). C C !!! IMPORTANT -- !!! C compile this code with the same compiler switches as Siesta, C in what regards using single/double precision, C otherwise reading the data from unformatted files C written by iorho.F might be spoiled. CDon't say you haven't been warned. C!!! C C Written by Andrei Postnikov, Mar 2007 C postni...@univ-metz.fr C program grdint implicit none integer ii1,ii2,io1 parameter (ii1=11,ii2=12,io1=14) integer mesh(3),nspin,is,ii,jj,iat,nat,nz,iix,iiy,iiz, .ind,mn,ialloc,ishift,jshift,kshift,limit,nrat parameter (limit=1) ! tried translations along each lattice vector character inpfil*60,outfil*60,syslab*30,suffix*6,owrite*1 logical filexist double precision b2ang,cc_bohr(3,3),diff,coort(3),cell(3,3), . dummy,modu,rmesh(3),small,volume,srad,srad2, . dist2,charge,spin parameter (b2ang=0.529177) ! Bohr to Angstroem parameter (small=1.0d-5) integer, allocatable :: ityp(:),iz(:) double precision, allocatable :: coor_bohr(:,:) real, allocatable :: func(:,:,:,:) ! NB! single precision, as in iorho.F real sum(8) ! NB! single precision, as in iorho.F C C string manipulation functions in Fortran used below: C len_trim(string): returns the length of string C without trailing blank characters, C trim(string): returns the string with railing blanks removed C char(integer) : returns the character in the specified position C of computer's ASCII table, i.e. char(49)=1 write (6,701,advance="no") 701 format(" Specify SystemLabel (or 'siesta' if none): ") read (5,*) syslab C -- handle XV file: inpfil = trim(syslab)//'.XV' open (ii1,file=inpfil,form='formatted',status='old',err=801) C -- check number of atoms in the XV file, for allocation: do ii=1,3 read (ii1,'(3x,3f18.9)',end=801,err=801) (dummy,jj=1,3) enddo read (ii1,*,end=802,err=802) nat allocate (ityp(nat)) allocate (iz(nat)) allocate (coor_bohr(1:3,1:nat),STAT=ialloc) if (ialloc.ne.0) then write (6,*) ' Fails to allocate space for ',nat,' atoms.' stop endif rewind (ii1) C -- read in translation vectors: do ii=1,3 read (ii1,'(3x,3f18.9)') (cc_bohr(jj,ii),jj=1,3) enddo read (ii1,*) nat do iat=1,nat read (ii1,'(i3,i6,3f18.9)',end=803,err=803) .ityp(iat),iz(iat),coor_bohr(:,iat) enddo close (ii1) C C --- Look for grid data files to include: 103 write (6,706,advance="no") read (5,*) suffix inpfil = trim(syslab)//'.'//trim(suffix) open (ii2,file=inpfil,form='unformatted',status='old',err=806) write (6,*) 'Found and opened: ',trim(inpfil) read (ii2,err=807) cell C check if cell vectors match those from .XV, just for the case... diff = 0.0 do ii=1,3 do jj=1,3 diff = diff + (cell(ii,jj)-cc_bohr(ii,jj))**2 enddo enddo if (diff.gt.small) then write (6,*) ' cell vectors from XV: ' write (6,'(3f12.6)') (cc_bohr(ii,
Re: [SIESTA-L] Band structure for BiFeO3
Dear ionutghitiu,your message is a bit confusing. You say your band structures are like in other calculations,the magnetic moments are like in the literature - what is the problem, then?The list of different parameters you tried is impressive but doesn't give a clue.You have large Fe magnetic moments but you see no spin splitting in (total?) DOS and bands -are you looking at an antiferromagnetic solution? Best regardsAndrei Postnikov - ionut ghitiu a écrit : >Dear Siesta users, I've been trying to obtain the band structure of BiFeO3 for some time now, mind you it's my first project involving Siesta or dtf so please bear with me. The thing is, the band structures that I get are really similar to the ones from materialsproject for the same structure (I am using the R3C primitive cell) but for the fact that the system seems to be spin degenerate, and so I get no splitting of the bands or difference in DOS. To give you some details about the system, I'm currently using FR/SR pseudopotentials from Pseudo-Dojo and an optimized base, but I also tried with the default one (up to TZDP i think). My MeshCutoff usually was 250 Ry, but recently changed it to 350 Ry, and I also had a FilterCutoff of 200-250 Ry. Also, I've been using a 5x5x5 gamma-centered kgrid and moved to a 7x7x7 . I'm initializing the Spin as polarised and setting the spins for the 2 Fe atoms as +/- (tried with values and with InitSpin.AF). As tolerances I use 1.d-5 for DM and the default one for HTol, but also tried with 1.d-6 for DM and 4.d-4 eV for the Hamiltonian. I tried both relaxed and unrelaxed systems. The thing is the magnetic moments written in the MullikenPop are close to what I've found in literature +-4.37 for Fe, 0.004 for Bi, and 0.029 for O. Also, I've tried experimenting a little on Fe2O3 with the same R3C structure and have the same problem, while on NiO everything went well. For Fe I'm using 3s2, 3p6, 3d6 and 4s2 electrons for the base. Could you please give me any hints on what I'm doing wrong or what to try next? I feel like I'm going in circles. Thanks a lot! -- 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] Puzzle in Spin Polarized calculation in Siesta
Dear Dmitriy, as I understand, you are talking about total spin polarisation, which should normally be reproducible throughout the methods. Otherwise, in terms of local spin moments, expressed via Mulliken charges or whatever, the differences in the definition between methods (or, between basis sets) may substantially add to confusion. This said, I'd suggest you to try to separate, as possible sources of worry, the "local" issues (pseudopotentials, basis) from "global" ones (convergence in a large system with – probably – narrow bands). Your P.S. gives an impression that you are trying everything possible in different directions at the same time. For setting the "local" issues, I'd suggest a reasonable compact case, including your cobalt or whatever, to check against an all–electron method. Here you'd check the magnetism or whatever is essential for your problem to the accuracy needed, settle the XC etc., and then do not touch these issues anymore. This settled, the instability of results may come from problematic convergence, as you have a large system and probably narrow bands which swap back and forth across the Fermi level. If you have no gap but a metal, then in a large system you need to be very careful about the dense enough k-mesh, mixing schema and weights, electronic temperature etc., to see that your system really converges and not fluctuates. Good luck, best regards Andrei -- prof. Andrei Postnikov -- University of Lorraine - Laboratoire de Chimie/Physique - A2MC ICPM, 1 Bd Arago - BP 95823, F-57078 Metz Cedex 03, France - Dmitriy Muzychenko a écrit : > Dear Siesta Users and Developers > > I’m really confused by the results of the Spin Polarized calculation in > Siesta. > > My system is Bi2Te3 slab with single Co atom substitutions of a Bi atom > (typically concentration: 1 Co atom per 490-1100 Bi+Te atoms). > > The confusion arises when I start to analyze the spin properties of this > system. Depending on the pseudopotential or/and basis set I observe > completely different value of total spin polarization (S_up-S_down). For > example, numerous calculations have resulted in the following list of total > spin polarization values: 4.33, 4.19, 4.00, 3.99, 3,77, 3.52, 3.00, 2,99, > 2,90, 2.63, 2.59, 2.00, 1.99, 1.53 and 0.00. I note, that these values are > for exactly the same system, just calculated with a different pseudo(s) or > basis set(s). > > Such a wide range of total spin (from 4.33 up to 0.00) raise a logical > question, what value can I trust and why? The question is rather not in > asking for some “perfect set” of parameters, but in what the general > strategy(rule) for analyzing the results should I choose in order to filter > out clearly artifact results and then focused on the most reliable one? > > I would be grateful for any help, advise or some “rule” that will allow to > filter out incorrect results. > > Thank you in advance for any advice. > > With kind regards, > Dmitriy Muzychenko > > > > P.S. I believe that it is useless to attach here any pseudo or basis that I > used, since I have tried a lot of possible combinations of pseudo and basis > (pseudo and basis: from paper [Comp. Mat. Sci. 98 (2015) 372]; from Siesta > Web-Database; my own generated/tested pseudo and basis). I have also allowed > Siesta prog to choose basis automatically, as well as used “gen-basis” > utility for basis or choose basis manually... Each attempt gives some new > value of total spin polarization (perhaps, it can only be noted that the most > common numbers are: 4.00, 2.00 and 0.00). Orbital population analysis shows > that in all cases polarization is comes mainly from d-shell of Co atom, with > the minor extra polarization of surrounding atoms. I have used GGA PBE > approximation with Spin Polarization as well as with SOC with the similar > results. For the Co species I have tried also semicore (3s2 3p6 3d7 4s2) > configuration with close to the same total spin variation from S=4.33 to 0.00 > depending on basis set or parameters used for pseudopotential generation. > Version of Siesta that I used is 4.1.b3/b4 or latest code from “master > brunch”. -- 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] Question about the number of electrons considered in the pseudopotential
Dear Tan Shi, the issue of importance of 3p for Fe has been discussed in Eur. Phys. J. D 25, 261–270 (2003) -DOI: 10.1140/epjd/e2003-00209-3 The pseudopotentials I used there were straightforward – generated with r = 2.00, 2.00, 2.00, 1.50, 0.0, 0.7for both cases, 4 1 0.00and 4 1 6.00 Best regards Andrei Postnikov - shi tan a écrit : >Dear SIESTA experts, For Fe, I noticed that the available pseudopotential from the siesta website only includes 4s2 and 3d6 electrons. But in some published works with other codes, 3p6 electrons are also included. I suppose that it is more accurate with more electrons taken into account. Where can I find those pseudopotentials with more electrons or should I construct one by myself? This happens to a lot of metals, is it accurate to just use those pseudopotentials from the SIESTA website to describe metal behavior? I would really appreciate your help! Thank you! Sincerely,-- Tan SHI -- 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] PDOS Up and Down
Dear Rayan,in a spin-polarized case, the script writes several columns of PDOS (2, or 4 for non-collinear case)into the output file:C --- write down accumulated DOS values: - if (isferm) write (io1,310) efermi write (io1,304) do it=1,nt write (io1,305) ene(it),(dos(it,ispin),ispin=1,nspin) enddo close (io1)...Best regardsAndrei - rayan moukhadder a écrit : >Dear siesta users,A successful spin polarized siesta run was done and a .PDOS >file has been created. Now I would like to plot PDOS up and down for specific >atoms using one of Andrei Postnikov tools which is fmpdos script. Using this script I can easily identify n,l,m quantum numbers of the atom but how can I specify the z number concerning the spin(up or down), for example if I need to plot only the spin up component of the Pz orbital how can I do such action. Thank you. -- 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] SCF not converging for Cr2O3
Dear Yuvam:It might be a good idea to fix first the small periodic Cr2O3 with less than 720 atoms... then if it works and your cluster don't,I'd say, you have problem with your structure (positions of atoms);try to look at them.By the way, how about the magnetic moments?An idea to try all different pseudopotentials until it starts to convergeis usually not very fruitful...And, it might make sense to look not only at the convergence partof the output file...Good luckAndrei Postnikov - Yuvam Bhateja a écrit : >Hello everyone, I was trying to perform CG calculation of Cr2O3 consisting of 720 atoms but it is not converging in its first cycle and reaching max scf iterations i.e., 400 and even 800.I tried lowering the electronic temp to 30 eV but it gets worse.I am currently using 50 eV but still not converging. I used both psf and psml pseudopotential but still not converging. I am attaching my input file and a screenshot.Can please anyone help? -- 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] PDOS at a vacuum on top of a slab
Dear Soumyajyoti,the easiest approximation to what you probably want (especially as you mentionempty spheres) might be to use ghost atoms, see the end of the description ofPAO.Basis block in the manual. Notably the atom number -100 is supposed to add the basis of spherical Bessel functionsto whenever you place your ghost atom, – or, you assign another negative numberand tune the basis using all the usual tricks. Then you just look at the PDOSat these ghost atoms. (I never tried this, through).Otherwise I'd suggest that you calculate LDOS over a sequence of energy intervalsand then integrate each LDOS over spacial region you are interested in (a thin layer at some height above the surface, in your case).The integrated values, placed along the energy axis, will then give you a kind of histogramBest regardsAndrei Postnikov - Soumyajyoti Haldar a écrit : > Dear Siesta Users, > I am trying to calculate PDOS in the vacuum at a Z Å > height above a surface. Can you suggest how to proceed with Siesta? In some > other codes, I know one can use empty sphere to project DOS. Is there > something similar in Siesta one can set up? If yes can you > let me know how to do it? > Thanks and regards > Soumyajyoti >Dr. Soumyajyoti > Haldar > Postdoctoral Researcher | AG Heinze > Institut für Theoretische Physik und Astrophysik > Christian-Albrechts-Universität zu Kiel> Leibnizstraße 15 | 24098 Kiel | > Germany> Email: > hal...@physik.uni-kiel.de -- 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] [SIESTA] Performing geometrical optimization
Dear Yuvam, if you are new to the code it might be more "robust" to test parts of your system (just a single-cell graphene; separately a single-cell oxide; ... ) before making and relaxing a supercell out of them. The point is, the choice of basis functions might be critical in Siesta because it represents a deal between accuracy you need and the computational effort you would like to spend, and it is easier to get some feeling about this compromise when testing small systems. Exactly for this reason the question "how fast is Siesta", put in general form, is ill-defined. The answer would depend on the accuracy needed (i.e., which properties you are after), moreover – a lot – on the geometry of your system, number of neighbours to each atom, amount of vacuum in the unit cell, etc. This said, an automatic relaxation of all crystallographic degrees of freedom is technically no problem with MD.VariableCell set to T, but, in a complex system, you might need to fix some of parameters (the lattice parameter of the substrate, or else) before engaging in a global uncontrollable relaxation which might lead to weird results. This said, I'd warn you not to expect too much tutorial-like introductions from the mailing list(tutorials being accessible in abundance via the Siesta web page), but answers to specific questions.Your inquiry seems mysterious without indicating who "they" used which scripts for which optimization, and what was confusing in it.Best regardsAndrei Postnikov - Yuvam Bhateja a écrit : >Hello everyone, My name is Yuvam and I am an undergraduate student from Kolkata, India. I am new in siesta and have some experience with softwares like Quantum ESPRESSO. I want to perform geometrical optimization of my unit cell (atomic position as well as cell vectors). I have created my custom made unit cell to accommodate the graphene as well as metal oxide for gas sensing. I was following a tutorial in which they used scripts for optimization but I find it very confusing and very unsuitable as varying all 9 values of lattice vector using loops was very cumbersome. Can someone help provide any other method in which system does it by its own like in QE? And also, how fast is siesta as compared to other codes like QE, as my system consists of 200-300 atoms and using a cluster with 8 cores and 42 GB RAM. Thank you in advance. RegardsYuvam BhatejaB.Tech. 3rd yearE Shibpur -- 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] Restarting a siesta job
- RAHUL SURESH a écrit : >Hi.. Thank you for your reply. My input file is as follows:(...) scf: iscf Eharris(eV) E_KS(eV) FreeEng(eV)dDmax Ef(eV) scf:1 -557650.096820381.783520381.6763753.10403 scf:2 -4630269.6330 2077870.1836 2077870.1745864.83778 scf:3 -2261484.3148 854868.7375 854868.6831* scf:4 -1331002.0702 338981.5615 338981.4669425.38740 scf:5 -719651.9273 106868.1468 106868.1251* scf:6 -717147.807168971.176668971.0943426.43790 Can I just add MD.UseSaveCG .true. MD.UseSaveXV .true. DM.UseSaveDM .true. input file without changing anything to the fdf file and re-run? Thank you Dear Rahul:you can do whatever you want, but please take into account thatyour calculation does not converge and produces strongly meaningless results,and this not as it crashed but from the very beginning : scf: iscf Eharris(eV) E_KS(eV) FreeEng(eV)dDmax Ef(eV) scf:1 -89709.5924 -89842.8329 -89842.9104 1.93966 -6.9260 timer: Routine,Calls,Time,% =IterSCF 1 152.571 78.20 scf:2 -89677.0881 -89355.7591 -89355.9226 0.51386 -3.5728 scf:3 -89796.1277 -89389.5132 -89389.7502 0.97916 -2.3163 scf:4 -105751.2656 -88983.4222 -88983.7173 54.29636 -8.1884 scf:5 -491486.7932 -40257.9792 -40258.0558259.16731and in fact your output tells you this in plain words: scf: 999 -779797.2714 78350.029378349.9655683.29828 scf: 1000 -603514.165749775.908049775.8345* SCF_NOT_CONV: SCF did not converge in maximum number of steps. So a good idea might be to fix this, in the first place.Some suggestions:1) Is the structure correct? (Did you look at it?)If yes, could a smaller fragment than 648 atoms be selectedfor a trial calculation to check that it works correctly?2) Did you try a serial calculation? Would it converge? 3) What was the mixing parameter?Etc.Please ask this mailing lest responsibly; this is for asking questions and discussing specific problems,not for depositing millions of lines of output files.Good luckBest regardsAndrei Postnikov -- 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] Restarting a siesta job
Dear Rahul -it depends on where the job was terminated(usually, a more precise question leads to a more specific answer). If during the electronic iterations - which so far proceeded reasonably - you can normally start from the .DM file(see that the switch is set UseSaveData T).If during the structure relaxation, you can start from the last .XV file.(See that the switch is set MD.UseSaveXV T). It might be a good idea to remove the Broyden history files.If during the frozen phonons calculation, you have to startthe displacements of the last atom anew, and erase by handthe last added blocks (out of 6, calculated for +/- X/Y/Z displacements of each atom)in the .FC file -set MD.FCfirst accordingly.Best regardsAndrei Postnikov - RAHUL SURESH a écrit : >Hi users. The terminal in which the siesta job was running got terminated accidentally. How can I restart the job and how far the results are reliable with restart data file. Thank you -- Regards,Rahul -- 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] Negative translation acoustic modes in phonon band structure
Dear Igor, your phonons at Gamma are sane (you may check the frequencies, but otherwise they seem OK). This hints that a priori there is no problem with the equilibrium geometry. Now, you run a single cell > SuperCell_1 0 > SuperCell_2 0 > SuperCell_3 0 so I don't understand how you expect to get phonon dispersions from this. > (a supercell would otherwise do band folding and band structure would > have proportionally more bands). - right, this is how it is supposed to be. In "SIESTA way" you get force constants from supercell calculation, and then VIBRA Fourier transforms them and helps you to trace the phonon branches as if in a single cell. Note that the supercell size is the user's responsibility, and nothing will automatically prevent you from using a too small one... If in doubt (and in order to test things), try to make a doubled cell and calculate "exactly" a zone-boundary phonon... Best regards Andrei -- prof. Andrei Postnikov -- tel. +33-372749149 -- University of Lorraine - Laboratoire de Chimie/Physique - A2MC ICPM, 1 Bd Arago - BP 95823, F-57078 Metz Cedex 03, France - popov a écrit : > Hi, > > My Siesta calculations of phonon band structure of a spinel crystal > gives negative-energy > translational (the lowest two) bands along the Brillouin zone (see the > fig. in attachment). > > Suspecting that geometry optimized by CG in Siesta is at a saddle point, > before FC run I > optimized the geometry with various other algorithms, including those > from ASE (preconditioned > LBFGS, mdMin, GPMin) but I could not get rid of negative frequencies for > any of the optimization algorithms. The force tolerance for any of these > optimizers was 0.005 eV/Ang (default is 0.04). > > The primitive cell is rhombohedral and I use vibra for postprocessing of > the FC calculations. > Input for Vibra is similar to fdf for FC calculations, with addition of: > > SuperCell_1 0 > SuperCell_2 0 > SuperCell_3 0 > AtomicDispl 0.06 Ang > Eigenvectors True > > and I added masses of atoms in %block AtomicCoordinatesAndAtomicSpecies > in 5th column > and specified AtomicCoordinatesFormat NotScaledCartesianAng. > (should I consider InitMesh instead?) > corresponding to the real space grid step division of 5.9114/130 = 0.045 > Ang, where 5.9114 Ang > is the module of lattice vectors. I used MD.FCDispl 0.06 Ang. > > I use a primitive cell with 14 atoms for the FC calculations, i.e. I did > not make a supercell > in order to reconstruct 14*3=42 bands in band structure that is common > in literature for spinels > (a supercell would otherwise do band folding and band structure would > have proportionally more bands). > > Now I am running some additional calculations with finer MeshCutoff and > MD.FCDispl (1400 Ry, 0.02 Ang). > > I am out of options. Do you have a clue what could be cause of imaginary > frequencies of the > translational modes throughout the BZ and how to get rid of them? > > Thanks in advance, > Igor > > -- 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] problem
Dear doaa haleem : in your “111” case, your unit cell is twice longer in the X direction than in other two(did you have a look at the structure figure??) In other two cases, the lattice vectors are more cubic, yet you have Hf in the “A” positionand Ba in the “B” position (in the oxygen octahedron) of the perovskite. Is this as supposed to be?Won’t this be the other way around? Concerning the spin different from 0, this might be due to settings in your LDAU block.You switch some U on Hf and set SpinPolarized =T; what do you expect, in fact? In order to better see what’s going on you can add some printout (Mulliken charges…).I’d say, try first to get a reasonable structure in non-magnetic case without LDA+U... Best regards Andrei Postnikov - doaa haleem a écrit : >I am writing this mail to discuss a problem I have faced when I was >calculating using siesta package, I am using siesta 4.0 , I am working on >some peroveskite structure , at the beginning I have used calculations using >GGA only ,it is expected that the final structure (XSF) to be the same as the >start structure (FCC) ,but it isn't and it consists only from 5 atoms with >wrong positions..(I will attach the fdf file, XSF, and gnuplot draw that >resulted for the previous case named after "case 1") The same have happened >when using GGA+U (case 2)Trying to solve this problem I have used the geometry >constraint to fix Hf atom and all 3-O atoms as in the perovskite BaHfO3, the >resulted structure is wright but -in the DOS and PDOS - the spin up DOS isn't >the same as the spin down DOS ( I will send the fdf file , XSF and gnuplot >draw for the previous case named after "case ") I tried also to use "variable >cell command , repeating the run and changing the k-grid and mesh cut off but >the problem is still existing .It is notable that I am using the pseudo >potentials uploaded on siesta web page So, please help me to solve this >problem .I need the the structure to be right, the spin up the same as the >spin down and the total spin is zero .Thank you
Re: [SIESTA-L] EXTRACTION OF PDOS --> modified version of fmpdos
Dear Siesta users,in view of some modifications of the PDOS format in recent versions of Siesta,my script fmpdos.f (for extraction of atom-resolved partial densities of states)required modification. The updated version (hopefully able to manage bothold and new formats) is in attachment, and also available athttp://www.home.uni-osnabrueck.de/apostnik/Software/fmpdos.fThank you very much in advance for testing and sending me bug reports. Best regardsAndrei Postnikov- James Sifuna a écrit : >Hello, I have successfully done the dos and pdos calculations, I need to extract the pdos of atom 1 with n=4, l=0, m=0 etc. but i get an error.. siesta-4.0.2/Util/Contrib/APostnikov/fmpdos Input file name (PDOS): SrTiO3.PDOS Output file name : Sr.dat Extract data for atom index (enter atom NUMBER, or 0 to select all), or for all atoms of given species (enter its chemical LABEL): 1 Extract data for n= ... (0 for all n ): 0 Unknown identifier in PDOS file, line6 -0.69252E+02 Best wishes, J. program fmPDOS ! ! extracts partial densities of states from PDOS file of SIESTA, ! taking care of flexible format in index, atom_index, etc. fields ! Andrei Postnikov, University Paul Verlaine - Metz (Jul 2010) ! University of Lorraine (Jan 2019) ! andrei.postni...@univ-lorraine.fr ! implicit none integer ii1,io1 parameter (ii1=11,io1=12) integer nt,nmax,i0,ispin,nspin,norbs,it,npts,nene,is,idos,l,lref, .m,mode,mref,zeta,zvalue,atind,n,nref,indref,index,nline, .iquoted,iparsed parameter (nmax=1) ! max. number of energy values in DOS double precision ene(nmax),dos(nmax,4),dos1(4),dparsed,efermi character inpfil*60,outfil*60,string*80,llabel*80,rlabel*80, . species*6,squoted*6,chlab*6,owrite*1,s1*1,polar*6, . dtail*7 logical filexist,addos,nptdef,ene_fo,isferm external iquoted,squoted,iparsed,dparsed do it=1,nmax do is=1,2 dos(it,is)=0.d0 enddo enddo nptdef = .false. ! We don't know whether the file contains isferm = .false. ! Show the Fermi energy if included in the PDOS ! --- open input and outpute files: -- write (6,*) ' Input file name (PDOS):' read (5,*) inpfil inquire (file=trim(inpfil), exist=filexist) if (filexist) then open (ii1,file=trim(inpfil),status='old',form='formatted') else write (6,*) ' File does not exist!' stop endif write (6,*) ' Output file name :' read (5,*) outfil inquire (file=trim(outfil), exist=filexist) if (filexist) then write (6,*) ' File ',trim(outfil),' exists. Overwrite? (Y/N)' read (5,*) owrite if (owrite.eq.'Y'.or.owrite.eq.'y') then open (io1,file=trim(outfil),form='formatted',status='REPLACE') else write (6,*) ' Then rename is first. Bye...' stop endif else open (io1,file=trim(outfil),form='formatted',status='NEW') endif write (io1,309) trim(inpfil) ! --- select projected DOS which are needed: - 100 continue write (6,*) ' Extract data for atom index', .' (enter atom NUMBER, or 0 to select all),' write (6,*) ' or for all atoms of given species', .' (enter its chemical LABEL):' read (5,*) string ! check first character of string, whether it is decimal i0 = ichar(string(1:1)) if (i0.ge.48.and.i0.le.57) then ! 1st character passed is a number; ! assume the whole string is atom number mode = 1 read (string,*,err=401) indref write (io1,301)indref elseif ((i0.ge.65.and.i0.le.90).or.(i0.ge.97.and.i0.le.122)) then ! 1st character passed is a character [A-Z,a-z]; ! assume the whole string is atom label mode = 2 read (string,*,err=402) species write (io1,302) trim(species) else ! 1st character is a special symbol; presumably an error write (6,306) i0 goto 100 endif write (6,*) ' Extract data for n= ... (0 for all n ):' read (5,*) nref if (nref.ne.0) then write (6,*) ' Extract data for l= ... (-1 for all l ):' read (5,*) lref if (lref.ne.-1) then write (6,*) ' Extract data for m= ... (9 for all m ):' read (5,*) mref else mref=9 endif else lref=-1 mref=9 endif write (io1,308) nref,lref,mref ene_fo = .false. ! are we in the energy reading mode ? addos = .false. ! add this DOS to the result ? nene=0! counter for energy values nline = 0 ! global conter for lines in PDOS file ! --- read and analyze the PDOS file line by l
Re: [SIESTA-L] DOS calculation with fermi shifted to zero
Dear Riya,the DOS / PDOS come out in the "native" energy scale of the code(in which Fermi energy - listed in the last column of electronic iteration printout, under siesta: iscf Eharris(eV) E_KS(eV) FreeEng(eV) dDmax Ef(eV)- is typically several eV in the negative).In order to plot, you shift it "by hand" within your favourite plotting program.Thank you for your attention to Javier Junquera's tutorial; there is no special "Andrei Postnikov method" to plot the density of states... Best regardsAndrei Postnikov - Riya Rogers a écrit : >Dear All, I am calculating DOS of my system using the Andrei Postnikov method (given in tutorial below) : http://personales.unican.es/junqueraj/JavierJunquera_files/Metodos/Structuralproperties/PDOS-SrTiO3/Exercise-PDOS-SrTiO3.pdf I have a doubt that whether this method gives the DOS with fermi automatically shifted to zero? or Do we have to manually shift fermi to zero? Please tell me how to get DOS with fermi shifted to zero, I will be highly obliged. Thanking you RegardsRiya
Re: [SIESTA-L] File Not Converging
Dear Harsh Shah,it may sound banal, but - if you hope for a useful answer you should outline your problem in a meaningful way. "A certain system with 110 atoms did converge but the other one with 769does not" - this is not much of information.You define 2000 SCF iterations and 5000 CG steps - who of them do not converge?What is your mixing parameter, anyway? How does your failed convergence look like? In some systems (no idea what is yours) with many nearly degenerate statesa very drastic reduction of mixing might help(to get the convergence trend right, before trying more efficient mixing schemes).Best regards Andrei Postnikov - Harsh Shah a écrit : >Dear Siesta User, I am running a file in parallel with around 769 atoms in stampede2 supercomputer and my file is not converging. It has been running without giving any error for around 10-12 days and I think it should not take that much time.I ran few other files having 110 atoms and it is giving proper output. I am using:MD.MaxForceTol0.025MeshCutoff(in Ry) 80MaxSCFIterations 2000MD.NumCGSteps 5000DM.Tolerance 1.d-4 Is there anything which I should change from above? What can be other possibilities do to which the file is not converging? I would really appreciate if you could help me out. Regards,Harsh Shah
Re: [SIESTA-L] atom resolved phonon density of states
Dear Karolina:thanks for your interest; please try the attached code(a single fortran source file, including all the subroutines called).Just compile it.It is not separately documented, but you just run it and answer the questions.Please ask me if something is not clear or goes wrong,best regardsAndrei-- prof. Andrei Postnikov -- tel. +33-372749149 -- University of Lorraine - Laboratoire de Chimie/Physique - A2MCICPM, 1 Bd Arago - BP 95823, F-57078 Metz Cedex 03, France - Karolina Milowska a écrit : >Dear All, I would like to plot a phonon DOS resolved over particular atoms in my system, in a similar fashion to the standard projected DOS. How may I extract this information in practice? I have found a presentation from 2010: http://www.home.uni-osnabrueck.de/apostnik/Lectures/APostnikov-Phonons.pdf which shows what I would like to get. On slide 21, there is an information about utility - phdos. Where can I find it? Kind regards, Karolina C C phdos, a script to calculte phonon density of states C in a supercell, C using phonon eigenvectors provided by "vibrator" C Written by Andrei Postnikov, Oct 2005 Vers_0.2 CXV read format bug fixed May 2006 C apost...@uos.de C program phdos implicit none integer ii1,ii2,io1,io2,ivmin,ivmax parameter (ii1=11,ii2=12,io1=14,io2=15) integer ialloc,iat,nat,iatmin,iatmax,mode,nodif,idif,npoints integer, allocatable :: ityp(:),iz(:),jat(:,:),jtyp(:),nna(:) double precision tau(3,3),qvec(3),qq(3),delta double precision, allocatable :: mass(:),coor(:,:), . freq(:),evr(:,:,:),evi(:,:,:),wwsum(:,:),zz(:) character inpfil*60,outfil*60,syslab*30,qlab*1 character*2, allocatable :: label(:) external test_xv,read_xv,read_ev,full_dos,qres_dos C C string manipulation functions in Fortran used below: C len_trim(string): returns the length of string C without trailing blank characters, C --- read lattice vectors and atom positions from the .XV file: write (6,701) 701 format(' Specify SystemLabel of .XV file: ',$) read (5,*) syslab inpfil = syslab(1:len_trim(syslab))//'.XV' open (ii1,file=inpfil,form='formatted',status='old',err=801) call test_xv(ii1,nat) allocate (ityp(nat)) allocate (iz(nat)) allocate (mass(nat)) allocate (label(nat)) allocate (jtyp(nat)) allocate (nna(nat)) allocate (jat(nat,nat)) allocate (coor(1:3,1:nat),STAT=ialloc) if (ialloc.ne.0) then write (6,*) ' Fails to allocate space for ',nat,' atoms.' stop endif call read_xv(ii1,nat,ityp,iz,tau,mass,label,coor) close (ii1) C --- count different types in the XV file. Note that they can be changed C arbitrarily (by hand) in the XV file as compared to the Siesta run. C E.g. to select some atoms of type=4 as different, assign them type=14. C They will then get a separate column in output files. nodif=0 do 98 iat=1,nat if (nodif.gt.0) then do idif=1,nodif if (ityp(iat).eq.jtyp(idif)) then ! add atom to existing type nna(idif)=nna(idif)+1 jat(idif,nna(idif))=iat goto 98 endif enddo endif nodif=nodif+1 ! add a new type jtyp(nodif)=ityp(iat) nna(nodif)=1 jat(nodif,1)=iat 98 continue write (6,"(i5,' different types found in the XV file:')") nodif do idif=1,nodif write (6,"(i5,' atoms of type ',i5)") nna(idif),jtyp(idif) enddo C --- read and store frequencies /eigenvectors from the .vector file, C q =(0 0 0) only : C allocate for all modes, 1 through nat*3 : ivmin=1 ivmax=nat*3 C (change ivmin, ivmax above if want to restrict to less modes) allocate (freq(ivmin:ivmax)) allocate (evr(1:3,1:nat,ivmin:ivmax)) allocate (evi(1:3,1:nat,ivmin:ivmax)) allocate (zz(1:nodif)) allocate (wwsum(1:nodif,ivmin:ivmax),STAT=ialloc) if (ialloc.ne.0) then write (6,*) ' Fails to allocate space for vibration modes' stop endif write (6,702) 702 format(' Specify SystemLabel of .vectors file: ',$) read (5,*) syslab inpfil = syslab(1:len_trim(syslab))//'.vectors' open (ii2,file=inpfil,form='formatted',status='old',err=801) call read_ev(ii2,nat,qq,ivmin,ivmax,evr,evi,freq) write (6,*) 'opened and read ',inpfil close (ii2) C --- write (6,*) 'You have two options: Total density of states ', .'(sum of squares of eigenvectors),' write (6,*) 'or Q-projected density of states ', .'(convolution wit
Re: [SIESTA-L] VACF and PHONON DOS in SIESTA
Dear Suman: you can tryhttp://www.home.uni-osnabrueck.de/apostnik/download.html -> VeloCorrF Thank you for reporting bugs, Sincerely Andrei Postnikov - Suman Chowdhury a écrit : >Dear SIESTA user : Is there any utility program is SIESTA which can calculate >the velocity autocorrelation function and the fourier transform of it, because >the fourier transform of velocity autocorrelation function gives the phonon >density of states which I want to calculate..
Re: [SIESTA-L] Error in running grid2cube
- Ankita Joshi a écrit :>Dear Siesta Users, While plotting LDOS for my system using a post-processing tool grid2cube, I encountered following error:STOP grid2d: Parameter MAXP too small Does anyone knows how to solve this problem? Please give your suggestions. Dear Ankita:This is not an error but a security message to be taken literally.The suggestion to solve this problem is,open grid2cube.fin your favourite editor, correct the line parameter (maxp = 1200) to a larger value and recompile. You could have found the solution yourselfbrowsing the grid2cube.f for the stop message: np = mesh(1) * mesh(2) * mesh(3) if (np .gt. maxp) stop 'grid2d: Parameter MAXP too small' ( Do you really need the mesh more dense than that? )Best regards Andrei Postnikov
Re: [SIESTA-L] cell vector not converged
- Najmeh Honari a écrit : Dear Andrei, In terms of converging our simulations for k-mesh values, how do we know > whether the fluctuations of the "total energy" and > > "Cell Vector Mod" > are big or small? Dear Najmeh,do you refer to the recently posted results of Sunetra Das (are they "your simulations"?),or to something else? - Just in order to define what we are talking about.I'd say, the general approach is like this:- you converge the total energy (differences, see my previous comment) vs the k-ponts to the accuracy you need for your particular problem. - once you decide that the convergence is good, you write down the lattice parameters.To look at how do the lattice parameters fluctuate on increasing the k-mesh densitydoes not make much sense, because the lattice parameter is not a variational property. Best regardsAndrei Postnikov On 14 July 2018 at 03:02, Andrei Postnikov wrote: Dear Sunetra, 1) your system is 2D; your "Cell Vector Mod" is presumably the lattice parametouler in (x,y);how about the third parameter? Is there a chance that you have not fix it so that itfluctuates under the effect of stress tensor component which is a numerical noise? 2) "Is the free energy convergence enough?" - the answer fully depends onyour objective; normally this is not the absolute value of energy that you should care about -it has no special meaning anyway - but the energy difference between cases(geometries, magnetic solutions, etc.) which you want to compare in your study. Best regards Andrei Postnikov - Sunetra Das a écrit : >Dear experts, I am trying to converge my system unitcell structure for k - mesh values. Even though the Free Energy value has converged for less denser mesh, the unit cell parameter has not converged for even 21x21x1 k mesh. The system is a 2D one with 3 atoms in the unit cell. I am attaching the file containing the k values and the corresponding free energy and cell parameter values along with the plotted graph. please have a look and help me. Is the free energy convergence enough? I have used systems before where I have checked if both the parameters have converged for a given k mesh. Thank you. Regards,Sunetra Das
Re: [SIESTA-L] cell vector not converged
Dear Sunetra, 1) your system is 2D; your "Cell Vector Mod" is presumably the lattice parametouler in (x,y);how about the third parameter? Is there a chance that you have not fix it so that itfluctuates under the effect of stress tensor component which is a numerical noise? 2) "Is the free energy convergence enough?" - the answer fully depends onyour objective; normally this is not the absolute value of energy that you should care about -it has no special meaning anyway - but the energy difference between cases(geometries, magnetic solutions, etc.) which you want to compare in your study. Best regards Andrei Postnikov - Sunetra Das a écrit : >Dear experts, I am trying to converge my system unitcell structure for k - mesh values. Even though the Free Energy value has converged for less denser mesh, the unit cell parameter has not converged for even 21x21x1 k mesh. The system is a 2D one with 3 atoms in the unit cell. I am attaching the file containing the k values and the corresponding free energy and cell parameter values along with the plotted graph. please have a look and help me. Is the free energy convergence enough? I have used systems before where I have checked if both the parameters have converged for a given k mesh. Thank you. Regards,Sunetra Das
Re: [SIESTA-L] Problem in SCF convergence in spin polarized calculations
Dear Ananya:OK it doesn't converge, but are the intermediate (not converged) results meaningful, or crazy?Look at the Mulliken charges - are they more or less stable (and reasonable),in the course of iterations, or wildly fluctuating?Where is the magnetism - where it belongs (Fe impurity? ribbon edges?) Did you initialised the magnetism correctly?Do you see in the density of states more or less what you expect(knowing the DOS of pure graphene and knowing how the Fe 3d statesare supposed to be placed)?Etc. etc.So may things to look at, beyond just the convergence...My wild guess would be, ALL your carbon atomsare initialised to maximal spin, and then the things go astray... Best regardsAndrei Postnikov - Ananya Rajpoota écrit : > > >Dear users,I am interested in simulation of Graphene ribbons doped with Fe. >When I proceed with the spin-polarized calculations, the SCF in first CG run >didn't converge. It is taking 500 SCF cycles or even more during >convergence.Should I change pulay parameters? Or Linear mixing schemes? If >pulay then should I increase weight or number of pulay??Also, what else should >I correct to get faster convergence?? or Electronic temp should be reduced? My fdf file parameters are as follows:PAO.BasisSize DZPXC.functional GGAXC.authorsPBE%block kgrid_Monkhorst_Pack 1 0 0 0.0 0 1 0 0.0 0 0 49 0.0%endblock Kgrid_Monkhorst_PackMeshCutoff 370.0 Ry MaxSCFIterations 5000 SCF.Mixer.Method PulaySCF.Mixer.Variant originalSCF.Mixer.Weight 0.1SCF.Mixer.History 2SCF.Tolerance.DM 1.d-4 MD.TypeOfRun CGMD.MaxForceTol = 0.01 eV/AngMD.NumCGsteps 5000MD.VariableCell .false. SCF.Mix.First falseSCF.Mix.AfterConvergence falseSCF.RecomputeHAfterScf falseSCF.Mix densitySCF.Converge.DM true Spin polarized SCF.Mix.Spin allFixSpin falseDM.InitSpinAF false WriteCoorXmol trueSolutionMethod diagon Thank u so much...With Regards Anaya > >
Re: [SIESTA-L] Changing the crystal direction of Si
Well, your unit cell is now 4 times larger,so you need to specify 8 atoms per cell...If this is really what you want...Whatever is the meaning of your "change the crystal direction to <100> "...(do you want to expand the supercell along [100] for tracing the phonon dispersion?A doubled supercell [ 1 0 0 ] [ 0 1/2 -1/2 ] [ 0 1/2 1/2] with 4 Si atoms should work fine)Best regardsAndrei Postnikov - Alaa Akkoush (Student)a écrit : >> I have build the Silicon which has diamond structure and calculate it's >> dispersion relation. > I set the lattice vectors those of fcc and place two silicon atoms one at 0 > and other at (0.25,0.25,0.25) > as follows: > %block LatticeVectors > 0.0 0.5 0.5 > 0.5 0.0 0.5 > 0.5 0.5 0.0 > %endblock LatticeVectors > %block AtomicCoordinatesAndAtomicSpecies > 0.000.000.00 1 28.0855 > 0.250.25 0.251 28.0855 > %endblock AtomicCoordinatesAndAtomicSpecies > I want to CHANGE THE CRYSTAL DIRECTION TO <1 0 0>, so i changed the lattice > vector to > %block LatticeVectors > 1.0 0.0 0.0 > 0.0 1.0 0.0 > 0.0 0.0 1.0 > %endblock LatticeVectors > But the system is unstable!!!> Any help?
Re: [SIESTA-L] Fwd: problem with converging mesh-cutoff and K-grid points
Dear Sunetra,please have a look intohttp://www.home.uni-osnabrueck.de/apostnik/Lectures/SIESTA-tuto.pdfin particular from page 10 on.Best regardsAndrei Postnikov - Sunetra Das <sunetra.das...@gmail.com> a écrit : >Dear Andrei Postnikov Sir, Can you please guide me in how to do the 'eggbox test' ? I have gone through the tutorials and examples but am still unclear on how to do that test for a system. Will be very much thankful if you let me know the steps on testing the eggbox effect and eliminating it. I have seen the term 'eggbox energy' in siesta output file after completion of a run, but confused what to decipher from it. Thanking you and awaiting a reply. Warm regards and wishing you a happy new year in advance,Sunetra Das.
Re: [SIESTA-L] Fwd: problem with converging mesh-cutoff and K-grid points
Dear Najmeh,you are worrying about z-components of forcesand try to help it by increasing the k(xy) mesh. It means that your bands dispersion becomes ever better in the plane,however, I don't think this would have an important effect on the z-force! Three observations:1) Do you really have a problem?Your z-forces on atoms 1 and 2 do nicely sum up to zero within 0.001 - that is fine, - and the fact that they (individually) fluctuate with k simply reveals the fact that your structure is not at equilibrium. Try to get a better equilibrium - see 2).If you ultimate goal is phonons, why don't you calculate just Gammaand check if you get the acoustic modes close enough to zero?2)In order to suppress fluctuations of forces / energy, the MeshCutoff is probably much more important that k-mesh.Apparently you tried 300 Ry and did not increase it any further.It might be not sufficient, depending on pseudos / basis.(However, it might well be OK in your case, see 1). In order to understand what's going on, the eggbox test (rigidly moving all the atoms... along Z, your "problematic" direction) for different values of MeshCutoff would help.With insufficient MeshCutoff, the increase of the k-mesh alone won't help you.(and... do you really need 20 Ang in the z-direction? You only needso much that the basis functions won't overlap).3)The k-mesh is not a variational parameter, so you cannot expect a systematic lowering of the total energy(although it usually happens, albeit with fluctuations).For the phonon calculations, the total energy as such (i.e.,its absolute value) is irrelevant anyway. Best regardsAndrei Postnikov - Najmeh Honaria écrit : >Dear > Siesta users, > >I’m > trying to simulate 2D borophene δ6, in order to calculate phonon dispersion > of the material. I did convergence > test for mesh-cut-off and k-points. I selected mesh cutoff=300Ry and started > k-grid from 12x12x1 and increased it to 40x40x1. The issue is that neither > “total > energy” nor the “forces on atoms” are converged in other words fluctuations of > energies and forces are obsereved. Here are a table of some energy and forces > for different k-grid meshes of borophene > structure: > >> >K-grid > >Total > Energy (eV) > >Atomic > forces (eV/A) > >15x15 > >-206.414196 > >1 -0.01 -0.05 > -0.131499 >2 -0.02 -0.040.131227 > >17x17 > >-206.410794 > >10.00 -0.03 > -0.423281 >20.04 -0.010.423056 > >20x20 > >-206.413284 > >10.01 -0.06 > -0.287325 >2 -0.020.020.287087 > >23x23 > >-206.414429 > >10.00 -0.00 > -0.355986 >2 -0.02 -0.050.355733 > >25x25 > >-206.423565 > >1 -0.01 -0.00 > -0.416997 >2 -0.02 -0.030.416772 > >27x27 > >-206.424037 > >1 -0.02 -0.06 > -0.308230 >2 -0.010.000.307990 > >30x30 > >-206.422290 > >1 -0.01 -0.04 > -0.313348 >2 -0.00 -0.020.313111 > >33x33 > >-206.423341 > >1 -0.02 -0.08 > -0.368439 >2 -0.01 -0.050.368198 > >35x35 > >-206.422301 > >10.02 -0.04 > -0.275519 >2 -0.020.010.275263 > >38x38 > >-206.420742 > >10.020.00 > -0.288896 >2 -0.000.010.288655 > >40x40 > >-206.420015 > >1 -0.01 -0.05 > -0.357026 >2 -0.000.000.356782 > > > > >The > input.fdf file that is used in simulations is as follow: > > > > >SystemName B >SystemLabel Borophene >NumberOfAtoms 2 >NumberOfSpecies 1 >## >#Basis# >## >PAO.BasisSize DZP >## >#ChemicalSpecies# >## >%block > ChemicalSpeciesLabel > 1 > 5 B >%endblock > ChemicalSpeciesLabel >## >LatticeConstant 1.00 Ang > >%block LatticeVectors >1.6 0. > 0. >0.0 2.8200 > 0. >0.0 0. > 20.000 >%endblock > LatticeVectors > >AtomicCoordinatesFormat > Ang > >%block > AtomicCoordinatesAndAtomicSpecies >0.0 0.0 0.0 > 1 >0.8 1.41000 > 0.91000 1 >%endblock AtomicCoordinatesAndAtomicSpecies > >BandLinesScale >ReciprocalLatticeVectors > >%block BandLines >1 0.00 0.00 0.00 Gamma >200 0.50 0.00 0.00 X >200 0.50 0.50 0.00 S >200 0.00 0.50 0.00 Y >200 0.00 0.00
Re: [SIESTA-L] Structural relaxation of 2D carbon atom chain
Dear Tomasz, your carbon chain presumably doesn't want to bendbecause you keep the lattice constant (along the chain) too large, and fixed.Otherwise, if you want help please specify your problem - what you tried and what goes wrong.<< Problem starts when after hundred of calalculations the structer remaind the same >>does not seem to give a clue.Otherwise,you may have a look into my tutorial on a similar issue(an atomic - gold - chain; tendency to bend; stability; phonons) -an explaining presentation and the work directory with guidelines,http://www.home.uni-osnabrueck.de/apostnik/Lectures/Exercis-Phonons.pdfhttp://www.home.uni-osnabrueck.de/apostnik/Lectures/Execs_Barcelona_2017.zip at a Siesta school in Barcelona, May 2017.Good luckAndrei- -- prof. Andrei Postnikov -- tel. +33-372749149 University of Lorraine - Laboratoire de Chimie/Physique - A2MC ICPM, 1 Bd Arago - BP 95823, F-57078 Metz Cedex 03, France - Tomasz Kepczynski <kepczynskisie...@gmail.com> a écrit : >Ive just started using Siesta and wanted to simulate something basic. My tutor >adviced me to do for example carbon atomic chain but asked me to try simulate >it with 2 carbon atoms in a primitive cell with one placed out of the Main >axis So the atomic chain would be bend/ VVV shaped. Problem starts when after >hundred of calalculations the structer remaind the same and creats carbon >"dimers" at best Hope you can help me start my adventure with Siesta
Re: [SIESTA-L] Doping
- masoudeh maleki <masodeh.mal...@gmail.com> a écrit : >Dear Andrei,Thank you so much for your email. I'm so confused because I'm >trying to dope tinoxide with nitrogen by siesta code and I can’t optimize >three parameters such as k point, mesh cut off and lattice constant. I mean >that plot of the total energy against these parameters isn’t as well as pure >tinoxide. Before doping I optimize these parameters in bulk tinoxide and then >relaxed and after relaxed the structure. Should I do all these process in >doped tinoxide again? And would you please kindly guide me what should I do to >have acceptable plots in doped case? Should I change the coordinates doped >nitrogen atom manually or subsituating nitrogen instead of oxigen or tin is >enough ? Yours SincerelyMasoudeh Maleki Dear Masoudeh -indeed, it looks like a big confusion. OK, slowly: > I can’t optimize three parameters such as k point, mesh cut off and lattice constant. These "parameters" are not in the same standing. The lattice constant is your result,you can indeed optimise it by varying it and looking for total energy minimum. The mesh cutoff and k-points a responsible for the accuracy of calculation.There are no "optimal" value of them that would minimise the total energy.Typically, as you increase them the total energy will consistently go down.The larger (mesh cutoff and k-points cutoff) the better; however you won't like to overkill.Therefore you make a reasonable free choice of these parameters driven by analysis of whatis the accuracy of the total energy you really need. This is never the total energyby itself, but a difference of energies between different states of phases.As the target accuracy may considerably vary from task to task, so may varythe technical parameters of your calculation. > Should I do all these process in doped tinoxide again? If the chosen mesh > cutoff value is OK for pure SnO2-and- also for some system containing the > dopant it ought to be OKfor the SnO2 + dopant as well.The "necessary" k-mesh > essentially depends on the supercell size.Keep about the same density of > k-points (i.e. the same real spacecutoff radius) as you increase your > supercell. > Should I change the coordinates doped nitrogen atom manually > or > subsituating nitrogen instead of oxigen or tin is enough ?Without knowing > what is your task nobody will tell you what is "enough".In principle, the > lattice will relax around a substitutional impurity.What's a big problem to > let it , in a Siesta calculation? Best regards Andrei Postnikov On Wed, Sep 27, 2017 at 11:31 PM Andrei Postnikov <andrei.postni...@univ-lorraine.fr> wrote: - masoudeh maleki <masodeh.mal...@gmail.com> a écrit : >Dear sir, I'm using siesta code to investigate the effect of doping on bandstructure and PDOS of periodic crystal of tin oxide with tetragonal unitcell. I made a 2*2*2 supercell of it, and replaced one O atom with my dopant. According to the siesta manual, can l use NetCharge and simulatedoping together for my bulk structure to simulate doping? Or it is impossible? And if it is incorrect how can I dope my bulk structure in siesta? Yours sincerelyMasoudeh maleki Dear Masoudeh, if you explicitly add the dopant atom to your system, you don't need any NetChargesince your system is not charged. (The dopant atom comes with its correct nuclear chargeand number of electrons). As explained in the manual, the NetCharge option is providedto treat charged finite systems (molecules). Best regards Andrei Postnikov
Re: [SIESTA-L] Doping
Dear Masoudeh -indeed, it looks like a big confusion. A pity because these issues were so many timesaddressed in tutorials and - I am sure - in the mailing list.OK, slowly: > I can’t optimize three parameters such as k point, mesh cut off and lattice constant. These "parameters" are not in the same standing. The lattice constant is your result,you can indeed optimise it by varying it and looking for total energy minimum. The mesh cutoff and k-points a responsible for the accuracy of calculation.There are no "optimal" value of them that would minimise the total energy.Typically, as you increase them the total energy will consistently go down.You just make a free choice of these parameters driven by analysis of whatis the accuracy of the total energy you really need. > Should I do all these process in doped tinoxide again? If the given mesh > cutoff value is OK for pure SnO2 -and- also for - masoudeh maleki <masodeh.mal...@gmail.com> a écrit : >Dear Andrei,Thank you so much for your email. I'm so confused because I'm >trying to dope tinoxide with nitrogen by siesta code and I can’t optimize >three parameters such as k point, mesh cut off and lattice constant. I mean >that plot of the total energy against these parameters isn’t as well as pure >tinoxide. Before doping I optimize these parameters in bulk tinoxide and then >relaxed and after relaxed the structure. Should I do all these process in >doped tinoxide again? And would you please kindly guide me what should I do to >have acceptable plots in doped case? Should I change the coordinates doped >nitrogen atom manually or subsituating nitrogen instead of oxigen or tin is >enough ? Yours SincerelyMasoudeh Maleki On Wed, Sep 27, 2017 at 11:31 PM Andrei Postnikov <andrei.postni...@univ-lorraine.fr> wrote: - masoudeh maleki <masodeh.mal...@gmail.com> a écrit : >Dear sir, I'm using siesta code to investigate the effect of doping on bandstructure and PDOS of periodic crystal of tin oxide with tetragonal unitcell. I made a 2*2*2 supercell of it, and replaced one O atom with my dopant. According to the siesta manual, can l use NetCharge and simulatedoping together for my bulk structure to simulate doping? Or it is impossible? And if it is incorrect how can I dope my bulk structure in siesta? Yours sincerelyMasoudeh maleki Dear Masoudeh, if you explicitly add the dopant atom to your system, you don't need any NetChargesince your system is not charged. (The dopant atom comes with its correct nuclear chargeand number of electrons). As explained in the manual, the NetCharge option is providedto treat charged finite systems (molecules). Best regards Andrei Postnikov -- -- prof. Andrei Postnikov -- tel. +33-372749149 -- cell +33-636565128 -- University of Lorraine - Laboratoire de Chimie/Physique - A2MC ICPM, 1 Bd Arago - BP 95823, F-57078 Metz Cedex 03, France
Re: [SIESTA-L] Doping
- masoudeh malekia écrit : >Dear sir, I'm using siesta code to investigate the effect of doping on bandstructure and PDOS of periodic crystal of tin oxide with tetragonal unitcell. I made a 2*2*2 supercell of it, and replaced one O atom with my dopant. According to the siesta manual, can l use NetCharge and simulatedoping together for my bulk structure to simulate doping? Or it is impossible? And if it is incorrect how can I dope my bulk structure in siesta? Yours sincerelyMasoudeh maleki Dear Masoudeh,if you explicitly add the dopant atom to your system, you don't need any NetChargesince your system is not charged. (The dopant atom comes with its correct nuclear chargeand number of electrons). As explained in the manual, the NetCharge option is providedto treat charged finite systems (molecules). Best regardsAndrei Postnikov
Re: [SIESTA-L] Is it Possible to make a movie in SIESTA?
Dear Rishi,there is my old tool which helps to manipulate the Siesta output informationconcerning molecular dynamics or relaxation(available in MD or MD_CAR or ANI files) to write more or less directlyinto the animation file readable by XCrySDen is available fromhttp://www.home.uni-osnabrueck.de/apostnik/download.html( -> Sies2xsf ),or in the Siesta distribution, see Util/Contrib/APostnikovSee a short description in the README file and at the beginning of the code, md2axsf.fmoreover e.g.https://departments.icmab.es/leem/siesta/Documentation/Tutorials/Barcelona-2007/Postnikov-visual.pdffrom page 14 on.The XCrySDen nicely allows to run animations and save them into animated files;if you don't like XCrySDen or have problems with ityou can easily adapt the format of the resulting .axsf file(atomic types and coordinates)to any visualising software. Even if you generate a picture "by hand" from every selected frame, you'll easily find tools to merge them into a movie.Good luckAndrei -- prof. Andrei Postnikov -- tel. +33-372749149 --- University of Lorraine - Laboratoire de Chimie/Physique - A2MC ICPM, 1 Bd Arago - BP 95823, F-57078 Metz Cedex 03, France - Rishi Sreedhar <rishisr33d...@gmail.com> a écrit : >Dear SIESTA community, I was wondering if it would be possible to get the cartesian output of the input molecular cluster whilst SIESTA is running; like say after every 10-20 CG optimization moves so that one could gather all those XYZ outputs and treat each one as a frame in a movie that visually depicts the system going to it's minimum energy configuration starting from the initial state. Does anybody know if such a feature already exists or if it is a tweak that is implementable with not much difficulty? Thank you for your time, Best Regards, Rishi Sreedhar.
Re: [SIESTA-L] asterisks and NAN in output
Dear Mogus:I think, the OrderN method is generally much more fragile than diagonalisation,so that you should not expect that it goes by itself from scratch.In your case, asterisks indicate that your calculation goes astrayearly enough (iter 9).In fact you are not obliged to use OrderN just for the sake of it,once your system has more than 100 atoms. In many cases, you'llgo ahead just fine with systems of many hundreds of atoms anddiagonalisation, and spare yourself many troubles. OrderN is justan option to make calculation faster, if possible.I'd suggest the following approach:1. Check that the structure is OK. It does not have to be pre-relaxed by VASPor whatever, but you should exclude error in the units (Bohr / Ang),wrong translation vectors, atoms sitting on top of the others, etc.It seems obvious but in fact a large fraction of problems comes from his kind of things. Any visualisation of structure would be helpful.2. Make a Diagon run and save the density matrix; later on start OrderNfrom it, and not from scratch.3. Also from Diagon calculation, check that you have a band gap(in the right place), and write down the Fermi energy for the use with OrderN.4. Carefully try OrderN to see if it would work fine and faster than Diagon.Be prudent with mixing parameters. Look at how the things changefrom one iteration to the other and tune the parameters accordingly(to have a good yet stable convergence). Good luckAndrei -- prof. Andrei Postnikov -- tel. +33-372749149 -- -- University of Lorraine - Laboratoire de Chimie/Physique - A2MC ICPM, 1 Bd Arago - BP 95823, F-57078 Metz Cedex 03, France - Mochena, Mogus D. <mogus.moch...@famu.edu> a écrit : Dear Siesta Users: > I am trying to use ORDERN solutionmethod to optimize a quantum dot of CdSe > with 90 atoms of Cd and 90 atoms of Se. Actually the system was optimized > with VASP and the optimized coordinates were being used to validate SIESTA > input parameters prior to > using SIESTA for larger systems. The diagonal solution method ran fine > (single point calculation). When I tried to run the ORDERN solution method, > some results (numbers) are getting too large and producing asterisks or a NAN > (I think). I don't know how to > fix the problem. I used the diagonal method results (Md.UseSaveXV) to run the > ordern calculation. Please see the output file below. I placed the > coordinates at the end of the file. > siesta: System type = molecule > > initatomlists: Number of atoms, orbitals, and projectors:180 2520 2880 > > siesta: Simulation parameters > > siesta: > siesta: The following are some of the parameters of the simulation. > siesta: A complete list of the parameters used, including default values, > siesta: can be found in file out.fdf > siesta: > redata: Non-Collinear-spin run = F > redata: SpinPolarized (Up/Down) run = F > redata: Number of spin components= 1 > redata: Long output = F > redata: Number of Atomic Species =2 > redata: Charge density info will appear in .RHO file > redata: Write Mulliken Pop. = NO > redata: Mesh Cutoff = 100. Ry > redata: Net charge of the system = 0. |e| > redata: Max. number of SCF Iter = 100 > redata: Performing Pulay mixing using= 8 iterations > redata: Mix DM in first SCF step ? = F > redata: Write Pulay info on disk?= F > redata: Discard 1st Pulay DM after kick = F > redata: New DM Mixing Weight = 0.0500 > redata: New DM Occupancy tolerance = 0.0001 > redata: No kicks to SCF > redata: DM Mixing Weight for Kicks = 0.5000 > redata: DM Tolerance for SCF = 0.000100 > redata: Require Energy convergence for SCF = F > redata: DM Energy tolerance for SCF = 0.000100 eV > redata: Require Harris convergence for SCF = F > redata: DM Harris energy tolerance for SCF = 0.000100 eV > redata: Using Saved Data (generic) = F > redata: Use continuation files for DM= F > redata: Neglect nonoverlap interactions = F > redata: Method of Calculation= Order-N > redata: Fix the spin of the system = F > redata: Maximum number of iterations = 1000 > redata: Relative tolerance = 0.10D-07 > redata: Eta (Fermi level parameter) = 0. Ry > redata: Radius of LWFs = 9.5000 Bohr > redata: Use continuation files for LWF = F > redata: Method to build LWFs = kim > redata: Dyn
Re: [SIESTA-L] Data in SystemLabel.vectors
- T. Liu <tl...@cam.ac.uk> a écrit : > Dear All, > > After using Vibra to get the vibrational frequency, I can get such a > file called SystemLabel.vectors, and there are many coordinates (or I > may call them vectors) in the file as following, (...) > Could anyone tell me what these coordinates are about please? Dear Tao - Yes, it would make sense to call them (eigen)vectors. They are relative displacements within the mode in question times sqrt of mass, normalised to 1.0 (sum of squares over all atoms and Cartesian coordinates). Good luck Andrei Postnikov > > k= 0.000.000.00 > Eigenvector = 1 > Frequency=-0.108924 > Eigenmode (real part) >0.6144E-02 0.2053E-01 0.3226E+00 >0.6144E-02 0.2053E-01 0.3226E+00 >0.6269E-02 0.2095E-01 0.3291E+00 >0.6269E-02 0.2095E-01 0.3291E+00 >0.4731E-02 0.1581E-01 0.2484E+00 >0.4731E-02 0.1581E-01 0.2484E+00 >0.4731E-02 0.1581E-01 0.2484E+00 >0.4731E-02 0.1581E-01 0.2484E+00 >0.4731E-02 0.1581E-01 0.2484E+00 >0.4731E-02 0.1581E-01 0.2484E+00 >0.4731E-02 0.1581E-01 0.2484E+00 >0.4731E-02 0.1581E-01 0.2484E+00 >0.4731E-02 0.1581E-01 0.2484E+00 >0.1187E-02 0.3968E-02 0.6235E-01 >0.1187E-02 0.3968E-02 0.6235E-01 >0.1187E-02 0.3968E-02 0.6235E-01 >0.1187E-02 0.3968E-02 0.6235E-01 > Eigenmode (imaginary part) >0.E+00 0.E+00 0.E+00 >0.E+00 0.E+00 0.E+00 >0.E+00 0.E+00 0.E+00 >0.E+00 0.E+00 0.E+00 >0.E+00 0.E+00 0.E+00 >0.E+00 0.E+00 0.E+00 >0.E+00 0.E+00 0.E+00 >0.E+00 0.E+00 0.E+00 >0.E+00 0.E+00 0.E+00 >0.E+00 0.E+00 0.E+00 >0.E+00 0.E+00 0.E+00 >0.E+00 0.E+00 0.E+00 >0.E+00 0.E+00 0.E+00 >0.E+00 0.E+00 0.E+00 >0.E+00 0.E+00 0.E+00 >0.E+00 0.E+00 0.E+00 >0.E+00 0.E+00 -0.E+00 > > Could anyone tell me what these coordinates are about please? Are they > the deviations (xyz) from the equilibrium position of each atom under > this normal mode? What's the Unit for these coordinates? > Thanks. > > Regards, > Tao >
Re: [SIESTA-L] calculate the vibrational frequencies for a molecule with fractional charges
Dear Zhuang, I'd guess, the issue of charge is technically decoupled from the issue of vibrations. That means, to arrange a calculation for a molecule with fractional charge is one thing that may pose problems, but once this is done and the static calculation makes sense (there is a switch NetCharge; the documentation says the fractional charges are possible, but warns about difficulties), you go ahead with phonons as you'd do in any other case, with MD.TypeOfRun = FC. Best regards Andrei -- prof. Andrei Postnikov -- tel. +33-372749149 -- -- University of Lorraine - Laboratoire de Chimie/Physique - A2MC ICPM, 1 Bd Arago - BP 95823, F-57078 Metz Cedex 03, France - Zhuang Bo <zhuangbo2...@outlook.com> a écrit : > Dear Siesta users, > > I am a beginer to Siesta and I am trying to use Siesta to calculate the > vibrational frequencies for a molecule with fractional charges (for example, > 1/2 or 1/3 of charges). > I found some paper using Siesta to do such kind of calculation, such as > 10.1073/pnas.1320210111 and 10.1039/C6RA21476B, but they didn’t provide the > procedure in detail that I can follow. > I have also read through the user manual but couldn’t find the answer. > > Dose anyone here know how to do such kind of calculation using Siesta? > I really appreciate any help you can provide. > > > Zhuang --
Re: [SIESTA-L] range of the basis function
- toufik esssakhria écrit : > Dear all,> please how to "extract" the range of the basis function in > Angstroms from *.psf file? Dear Toufik,the basis function is NOT in the .psf file (which has to do with the pseudopotential);to my knowledge, you can read the basis function extension (in Bohr)in the output file, close to the beginning, e.g.: Vna: Cut-off radius for the neutral-atom potential: 6.351100 Best regardsAndrei Postnikov
Re: [SIESTA-L] Error in Using PAO.SpliNorm
Dear Ajay, the warning says, certain atoms are too close, and this most probably happens on applying the translations. Almost certainly there is an error in structure definition - I'd suggest to get a view of the atoms in the unit cell, with and without translations. SplitNorm is not a problem (at least, not at this stage..) Best regards Andrei Postnikov - Ajay_Khanna <samthecha...@gmail.com> a écrit : > Greetings Everyone, I'm wondering if someone can help me in correcting this > warning > > siesta: WARNING: Atoms 11679 11979 too close: rij =0.402646 Ang > siesta: WARNING: Atoms 11680 11963 too close: rij =0.516207 Ang > > I'm using PAO.SplitNorm 0.16 and varied it to 0.37 > But still, I'm getting this warning. Can anybody please tell me what this > warning about is?? and how can I correct this? > > Thank you. > Ajay Khanna > Student
Re: [SIESTA-L] Is relaxation is necessary during k-point optimization
Dear Rajan,as you speak about a ribbon, it looks probable that you'll have dispersionin one dimension only, so the k-mesh (to test the convergenceof results) will be linear and hardly prohibitively large.I don't understand what you mean by "relax my system at each k-point".If you want to test the convergence of results as the number of k-pointsincreases, you don't really need to relax the structure... Anyway, the result of the convergence test would depend on which properties exactlyare you after. To plot energy vs number of k-points would hardly be usefulby itself, because it is a function which goes down and down and down...You must know where to stop. You'll not be able to "evaluate optimized k-point"from the plot.Good luckAndrei Postnikov - RAJAN SINGHa écrit : >Dear Siesta Users> I have a fairly large system of about 50 atoms of graphene >ribbon. I want to optimize the same in terms of k-points and mesh cutoff.> But >if I relax my system at each k-point, It will cost too much time.> My question >is> Can I skip the relaxation fr each k point??> And plot Energies vs kpoint n >evaluate optimized k-point??> Please help me in this.With Regarda>Rajan
Re: [SIESTA-L] How to calculate single molecule energies
- I. Camps <ica...@gmail.com> a écrit : >> Dear Andrei, >> Thank you very much for your email. >> What I mean by convergence study is: > - I choose the accuracy for my energy calculations > - Do several calculations with different values of meshcutoff energies > - Plot the system energy (converged within the accuracy selected previously) > - Select the value of meshcutoff from were the system energy is stable >> Is this ok? > []'s, > Camps Dear Camps,This is OK up to the last point. What is your definition of "stable"?Does the energy have to be stable to within 0.01 eV or 0.1 meV ? This would of coursedepend on your problem under study. Moreover, the "system energy" as such(the absolute value as it comes out of the calculation) has no big meaning;what really interests you is typically the difference of this energy from someother energy (that of different phase, different magnetic state, etc.).This physically meaningful difference may converge to physically meaningful accuracy faster (or, on the contrary, slower)than you might have guessed while looking at your [system energy vs cutoff] plot. Best regardsAndrei > On Mon, Dec 19, 2016 at 8:59 PM, Andrei Postnikov <andrei.postni...@univ-lorraine.fr> wrote: A small comment to the latest remarks,sorry as it must be obvious:it hardly makes sense to run the convergence test on a calculation of just one molecule(or, in any standalone calculation) because you wont't have a cluewhat exactly, and to which accuracy, should be converged.Typically, you are after the energy difference between two (or more) cases:atoms vs molecule, free molecule + surface vs adsorbed molecule, etc.The convergence to be checkedis that of the meaningful energy difference, to the accuracy you need. Best regards Andrei Postnikov - I. Camps <ica...@gmail.com> a écrit : >Dear Riya, Its always good that, instead selecting those parameters (cutoff energy, cell parameters, numbers of k points, etc.) manually, you run convergence test. In this way, maybe you get smaller values for them and then less computational resources involved. []'s, Camps > On Sat, Dec 17, 2016 at 8:23 PM, Riya Rogers <rogers.riy...@gmail.com> wrote: Dear Dr Camps >Thank u so much. I m taking very high cutoff greater than 340Ry. N put the >Molecule in middle of big lattice. > My bond lengths are lil away frm exam abt 1.2 to 1.4%. >Thanking you >Regards > Riya > On 18-Dec-2016 2:40 am, "I. Camps" <ica...@gmail.com> wrote: >Can anyone tell me how to calculate energy of one molecule of gas. >Should I take molecule inside lattice? > Or > Should I follow the following example1: >http://personales.unican.es/junqueraj/JavierJunquera_files/Metodos/Basic/Basic-exercises.html SIESTA is designed for periodic calculation, so, you have to put your molecule inside a fictional lattice. The lattice vectors should be big enough to not get interaction between two neighborhood molecules. Take a look in the second example. Also, you can check how SIESTA is treating your system (molecular, surface, slab, etc.) in the output file. >Does k points n meshcutoff and basis set info req?Yes. K vectors has meaning only for periodic system, so, in case of a molecule, only one is needed (the gamma point (0,0,0). For the meshcutoff, you should do the convergence study as usually. The basis set should be selected according the accuracy you want or need (as in any SIESTA calculation). Regards, -- []`s >Camps > > >
Re: [SIESTA-L] How to calculate single molecule energies
A small comment to the latest remarks,sorry as it must be obvious:it hardly makes sense to run the convergence test on a calculation of just one molecule(or, in any standalone calculation) because you wont't have a cluewhat exactly, and to which accuracy, should be converged.Typically, you are after the energy difference between two (or more) cases: atoms vs molecule, free molecule + surface vs adsorbed molecule, etc. The convergence to be checkedis that of the meaningful energy difference, to the accuracy you need. Best regardsAndrei Postnikov - I. Campsa écrit : >Dear Riya, Its always good that, instead selecting those parameters (cutoff energy, cell parameters, numbers of k points, etc.) manually, you run convergence test. In this way, maybe you get smaller values for them and then less computational resources involved. []'s, Camps > On Sat, Dec 17, 2016 at 8:23 PM, Riya Rogers wrote: Dear Dr Camps >Thank u so much. I m taking very high cutoff greater than 340Ry. N put the >Molecule in middle of big lattice. > My bond lengths are lil away frm exam abt 1.2 to 1.4%. >Thanking you >Regards > Riya > On 18-Dec-2016 2:40 am, "I. Camps" wrote: >Can anyone tell me how to calculate energy of one molecule of gas. >Should I take molecule inside lattice? > Or > Should I follow the following example1: >http://personales.unican.es/junqueraj/JavierJunquera_files/Metodos/Basic/Basic-exercises.html SIESTA is designed for periodic calculation, so, you have to put your molecule inside a fictional lattice. The lattice vectors should be big enough to not get interaction between two neighborhood molecules. Take a look in the second example. Also, you can check how SIESTA is treating your system (molecular, surface, slab, etc.) in the output file. >Does k points n meshcutoff and basis set info req?Yes. K vectors has meaning only for periodic system, so, in case of a molecule, only one is needed (the gamma point (0,0,0). For the meshcutoff, you should do the convergence study as usually. The basis set should be selected according the accuracy you want or need (as in any SIESTA calculation). Regards, -- []`s >Camps > > >
Re: [SIESTA-L] siesta
Dear 木材,you might find useful to have a look athttp://departments.icmab.es/leem/siesta/Documentation/Tutorials/Lyon-2007/Postnikov-visual.pdf(from page 9 on),best regardsAndrei Postnikov - 木材a écrit : >Dear all When complete the calculation of siesta and get a file >selected.WFSX,I know it is for LUMO and HOMO,but how to deal with it and plot? >I really hope to get your hlep,hope your comment.
Re: [SIESTA-L] WRONG LOCAL MAGNETIC MOMENT OF Fe IN FeO.
Dear Anju,1) you lattice parameter seems to be too small; you define 3.04 Angwhereas in wüstite it seems to be around 4.3 Ang. Hopefully this is the source of your problem.2) Before playing around with U (up to U=14 eV, indeed!), why don't you check your calculation against the known LDA results? (Not only the magnetic moment but the placement of bands...)3) When asking for advise in searching for an error, it might be a good idea to provide an output along with your input file... Good luckAndrei Postnikov - Anju Sarohaa écrit : >Hi all, I have tried siesta-trunk-587 (latest version) but I am getting the same error. In the FeO calculation, simply to check, I gave ferromagnetic arrangement of spins for Fe. The expected local magnetic moments of Fe should come out to be 4 Bohr magneton but I am getting 0.001 Bohr magneton. Even for U=6 eV there is no band-gap. Can anyone please help me in resolving this issue? I have tried my best but didn't understand where is it wrong. I am looking forward for some suggestions. Please help me out. Here is the attached .fdf file. On Fri, Nov 25, 2016 at 9:28 PM, Anju Saroha wrote: Sorry I forgot to attach the input file. I am attaching the FeO.fdf file herewith. Please find the attachment. On Fri, Nov 25, 2016 at 9:11 PM, Anju Saroha wrote: Hi all, I am using siesta-trunk-527. I did a test run on FeO to check the installation. But I have encountered an error. Though FeO is an antiferromagnetic material, I gave the calculation for ferromagnetic case. The net magnetic moment of the unit cell is expected to be 8 Bohr magneton but I have got it as 0.5 Bohr magneton. The local magnetic moments of Fe are coming around 0.001 Bohr magneton (expected are 4 Bohr magneton). Has anyone faced a similar problem? Is it a bug in the code? If not, then what can lead to such an erroneous result? Even FeO is coming out to be metallic till U = 14 eV. Though it should become an insulator for U to be around 6 eV. Can anyone please help me to understand the error? I tried a lot to solve it but I am now clueless. Any help in the direction will be appreciated. I am curiously looking forward for suggestions. Thanks in advance. Regards -- Anju Saroha Research scholar IIT Madras Chennai-600036 India.
Re: [SIESTA-L] How to extract the force constant after siesta'scalculation for thermal conductance calculation by using nonequilibriumGreen's method
Dear Benhu Zhou, you want a lecture on how phonon calculation in Siesta works,that will replace all previous documents and tutorials. OK, I'll try to make it short:1. You have unit cell that contains M atoms. For graphene, silicon, NaCl M=2.2. If you want phonon dispersions, you construct a supercell that is a replicatedunit cell (in one dimension, or two, or three, depending on which dispersioninterests you and how far-fetching the force constants are). The supercell containssay N atoms (an integer multiple of M). If you want only Gamma phonon and no dispersion, supercell = unit cell and M=N.3. The FC calculation is organized as several inserted loops.The outer loop is over atoms in the unit cell, 1 to M. (You can performin your calculation any part of this loop, from atom FCfirst to FClast,1 <= FCfirst <= FClast <= M, and then concatenate them by hand). Every run of this loop adds 6*N lines to the .FC file.The next inner loop is over displacements, 1 to 6.For every displacement, the forces are calculated on all atoms in the supercelland written to the .FC file as a block of N lines, 3 columns (for Cartesian componentsof forces) each.As the FC calculation runs, it explains in the output file what it is doing right now,and you can inspect how the blocks are added to the.FC file.Good luckAndrei - 周本良 <blz...@hunnu.edu.cn> a écrit : >Dear Andrei Postnikov, > Thank you very much for your answer and offering relevant materials to me! >Yes, your offer materials has bee read in detail, and the similar issue > was also introdeced by Bedamani (see the website > https://www.mail-archive.com/siesta-l@uam.es/msg04969.html), and a reply was > given by Roberto. Part explanations are follows: > "Let's say FC.fdf file built by fcbuild contains N atoms, then each of them > gets a block of 6N rows by 3 columns in the *.FC file, representing the > forces on every atom, whenever every atom of your basic, reference, unit > cell moves a small distance (delta) along +x,-x,+y,-y,+z,-z. The numbers > written are not the forces themselves but forces/delta. Thus, if your unit > cell contains M atoms, the *.FC file will comprise M*6N lines." >Blame on me for being too stupid, I still have some question. >(1) as mentioned above, "a atom have 6 rows by 3 columns in the *.FC file, >representing the forces on this atom, whenever the atom of your basic, >reference, unit cell moves a small distance (delta) along +x,-x,+y,-y,+z,-z." >Dose the 6 rows represent the force constant for a atom in the unit cell when >the atom moves a small distance along +x,-x,+y,-y,+z,-z? if it is true, what >the mean for the first column, the second column and the third column are? >what the mean for the words "basis" and "reference" are? >(2) "if unit cell contains M atoms, the *.FC file will comprise M*6N lines." I >don't undstand why the total lines are M*6N, I think the total lines are 6*N >because N is the total number of atoms in the supercell. > > >Please help me, thanks! >Best regards, >Benhu Zhou > >-- Original -- > >From: "Andrei Postnikov"<andrei.postni...@univ-lorraine.fr>; >Date: Sat, Aug 20, 2016 09:59 AM >To: "siesta-l"<siesta-l@uam.es>; > >Subject: Re: [SIESTA-L] How to extract the force constant after >siesta'scalculation for thermal conductance calculation by using >nonequilibriumGreen's method > > > >Dear Benhu Zhou, >for the answer to (1), you can consult page 15 of >www.home.uni-osnabrueck.de/apostnik/Lectures/APostnikov-Phonons.pdf >(also, I think the issue was discussed many enough times in the mailing list). > >Best regards > >Andrei Postnikov > >- 周本良 <blz...@hunnu.edu.cn> a écrit : > >Dear all, > > > I am trying to use the force constant output from siesta for a phonon > transmission calculation for a graphene nanoribbon by using non-equilibrium > Green's function. The system for a graphene nanoribbons is consists of a > graphene nanoribbon (a unit cell have 8 atoms ) connected to two graphene > leads. Now I obtain the file graphene.Fc (24 atoms, 144 rows, 3 column, and > periodic boundary conditon is applied in the heat transport direction) with > the force constat matrix. But I have some questions about the file > graphene.FC. > > >(1)what is the mean for the elements in the graphene.FC file? For example, >what is the mean for the first column or the sencond culumn or the third >column? and what is the mean for the first row or the second row and so on? >I have read the guide in detail but I still do not understand. > > >(2) How I can transform the force constant to a matrix with 24X24 (basicaly
Re: [SIESTA-L] H2S Ground state has imaginary frequencies
Dear Anja, the problem of negative frequencies cannot be helped by argumentationabour agreement with experiment or other calculations. In fact it is a technical problem which you can explore very precisely,whatever your pseudos or bases. You have a certain total energy surface(as function of all coordinates). You apply a CG algorithm and arriveat what seems to be a minimum. The negative frequency hintsthat the starting point was in fact NOT a minimum, and indicateswhere to search better. This strategy is fool-proof, unless:i) your total energy surface is pathological, for whatever reason;ii) the numerical noise (fluctuations of total energy) in the course of relaxationis comparable with the energy differences you are after.The case (ii) can be systematically cured by increasing the MeshCutoff.Now as I look at your translation vectors (1.803645 15.6477590.331091), (11.8583939.3745580.435976), (0.5925850.553982 24.937136)it sems that you let them vary in the course of relaxation? -(Hopefully NOT in the course of phonon calculations? - Sorry).I see that your simulation boxis quite distorted - not an error by itself but I'd quess that your initialchoice was more "rectangular". A considerable deformation of the boxfor me is an indication that theMeshCutoff is too small.The fact that you need to go to unreasonably high MeshCutoffmight be a consequence of unfortunate choice of pseudos etc.but in principle you can eliminate the negative frequency problemjust by increasing the mesh, whatever the other settings.Best regardsAndrei - Anja Foerster <foers...@msu.edu> a écrit : > Dear Andrei Postnikov, > Thank you for your reply. I already followed the movement of the imaginary > frequency and used the new structure as the input for a new > optimization/frequency run. Essentially, I still end up with the same > optimized structure and frequency. If I compare my structure with the > literature values for the angle and distance, it is in good agreement for a > DFT calculation and as the by hand change of the angle and distance show, it > is the energetic minimum. > As for the suggestion to increase the size of the box: While for the simple > H2S molecule I can increase the box size to test if dipole-dipole > interactions are the cause, the H2S molecule is just a test system. What I > really want to calculate is H2S reacting with a 2-dimensional surface (about > 50 atoms). Here, I am not able to increase the size of the box due to the > considerable increase of the atom number and the calculation time in such a > case, especially for the frequency run. Therefore, I would be grateful if > there was a way to get the right frequencies for H2S while keeping the box > size of the 2D surface, which should be sufficiently large for H2S > the box size of the 2D surface in Ang : > 1.803645 15.6477590.331091 > 11.8583939.3745580.435976 > 0.5925850.553982 24.937136 > Thank you, > Anja > Quoting Andrei Postnikov <andrei.postni...@univ-lorraine.fr>: > > Dear Anja, > > in my experience, large negative frequencies are either trivial > > "noise"(too low mesh cutoff - apparently not your case), or, more > > frequently, a hint that you are not really at equilibrium. A good > > idea would be to look at the eigenvectorof the "most negative" mode, > > extract the relative displacements(not to forget the sqrt of masses) > > and give your atoms a small kickin this direction before resuming the > > relaxation. The negative (imaginary)frequency means - you displace > > the atoms in this mode, and the energy lowers.This might help you to > > find a better energy minimum,calculate there the frequencies again, > > etc.Another trouble (or, the reason for the behavior described > > above)could be the dipole-dipole interactions between your > > moleculesif the box is not large enough. However, the d-d interaction > > is long-ranged, so you'll have some amount of trouble at any box > > size.I don't remember if there are efficient switches to cancel it in > > this context.For a molecule like H2S, you would normally get 3 rigid > > displacementand 3 rigid rotation modes, all at zero frequency. It is > > relatively easyto bring the displacement modes to zero (by > > brute-force increasingthe mesh cutoff), whereas bringing the rotation > > modes to zero can bea hell, apparently because the "Siesta space" is > > not isotrope enough (interaction with replicas of the molecule, the > > rigid structure of mesh, ..?)Best regardsAndrei Postnikov > > - Anja Foerster <foers...@msu.edu> a écrit : > > > >> Dear all, > >> I performed a geometry optimization of H2S and then used the Vib
Re: [SIESTA-L] H2S Ground state has imaginary frequencies
Dear Anja, in my experience, large negative frequencies are either trivial "noise"(too low mesh cutoff - apparently not your case), or, more frequently, a hint that you are not really at equilibrium. A good idea would be to look at the eigenvectorof the "most negative" mode, extract the relative displacements(not to forget the sqrt of masses) and give your atoms a small kickin this direction before resuming the relaxation. The negative (imaginary)frequency means - you displace the atoms in this mode, and the energy lowers.This might help you to find a better energy minimum,calculate there the frequencies again, etc.Another trouble (or, the reason for the behavior described above)could be the dipole-dipole interactions between your moleculesif the box is not large enough. However, the d-d interaction is long-ranged, so you'll have some amount of trouble at any box size.I don't remember if there are efficient switches to cancel it in this context.For a molecule like H2S, you would normally get 3 rigid displacementand 3 rigid rotation modes, all at zero frequency. It is relatively easyto bring the displacement modes to zero (by brute-force increasingthe mesh cutoff), whereas bringing the rotation modes to zero can bea hell, apparently because the "Siesta space" is not isotrope enough (interaction with replicas of the molecule, the rigid structure of mesh, ..?)Best regardsAndrei Postnikov - Anja Foerstera écrit : >Dear all, > I performed a geometry optimization of H2S and then used the Vibra tool and a > FC run to calculate the frequency. For the optimized structure I receive 5 > negative (imaginary) frequencies (-228.9332 -135.1068-46.3373 > -0.1270 -0.0328). Since it is a small molecule, I varied both the > H-S-distance and the H-S-H angle. As you can see in the attached pdf, 1.38 > Ang is the H-S distance that corresponds to the energetic minimum. However, > only when the H-S-distance is increased to 1.40Ang are the frequencies > getting close enough to 0 for the structure to be labeled as "ground state" > based on the frequency results. > I already tried to increase the meshcutoff value to 800, which did not solve > the problem. My other critical values: MD.MaxForceTol is 0.001 eV/Ang and > DM.Energy.Tolerance is 1.0d-5 eV. I set the MD.MaxForceTol to a stricter > 0.0001 eV/Ang in the geometry optimization calculation, however even after > 400 steps is not converging. For MD.MaxForceTol 0.001 eV/Ang I receive “Tot > 0.0026530.001420 -0.001683” in the siesta output for “ siesta: Atomic > forces (eV/Ang)”. So the forces should be okay for a frequency run. > Does anyone have an explanation why the energetically lowest state has 5 > imaginary frequencies whereas an energetically higher geometry has > effectively no imaginary frequencies? > Thank you for your help, > Anja > P.S.: I'm using siesta version 3.1 >
Re: [SIESTA-L] Cholesky factorisation in rdiag error due to periodicity
Hi Arturo, you definedAtomicCoordinatesFormat Fractionalso that your atoms Nr 1 and 2 %block AtomicCoordinatesAndAtomicSpecies 4.00 0.00 1.00 3 3.00 0.00 1.00 3 as well as many others, are at identical positions. You know, in "fractional" arithmetics 4=3 (=2=1=0)You can see this explicitly as the coordinates are printed in Bohr: siesta: Atomic coordinates (Bohr) and species siesta: 152.22128 0.0 10.12043 3 1 siesta: 114.16596 0.0 10.12043 3 2 - these coordinates are separated by exactly the 1st translation vector(20.138 Ang).Indeed, this is because of periodicity (which is implicit in Siesta),however this is not an error but exactly the structure as you decribed it...The first step in solving the problem would be to figure outwhich geometry you really want to use (how long the translation vectorsand where are the atoms in the cell: placed from one another at distance =1.0 in which units?) Good luckAndrei Postnikov- Arturo Toroa écrit : >Hi all, While attempting to run a molecular dynamics simulation of an interface between a hexagonal system and a cubic system, I received the above mentioned error message. I then looked through the .out file and saw that there were numerous atoms which, although visibly located at different atoms, are said to be 0 angstroms apart. I believe this to be the result of an innate periodicity function within SIESTA. Is there a way to avoid this error besides manually attempting to delete all the conflicting atoms? Thanks again, Arturo Toro
Re: [SIESTA-L] bulk modulus by Postnikov's tool: Murnaghan equation of state fit.
Dear Manjeet: the "concept" is to add one line of code that calculatesvolume (in Bohr^3) from your lattice constant (in Ang).There is an example (commented) for bcc lattice.When programming this transformation, take care that the volumeyou calculate corresponds to the energy you have.Good luckAndrei Postnikov - Manjeet Bhatia <manjeetbhati...@gmail.com> a écrit : >Dear Andrei Sir, Thank you for your explanation. Yes, my input is lattice constant in ang and energy in eV. This means that there is no concept of lattice constant as input in murnagan fit, only volume and energy in appropriate units will give bulk modulus. Thank you. On Thu, Aug 11, 2016 at 9:14 PM, Andrei Postnikov <andrei.postni...@univ-lorraine.fr> wrote: Dear Manjeet: The mur_fit script expects the input data to be passed in two columns: (1) volume in Bohr^3, (2) Energy in Ry. These are the units internally used, in which the results are provided. In you case: # Equilibrium volume, a.u.^3 : 31.125644 # Minimum energy, Ry : -614.470876Now, it might be in the user's convenience to pass instead the volume and/or energy in other units, and convert them into Bohr^3 / Ry once they are read in. To this end, some"templates" are provided in the source file,(before the i=i+1 statement) i.e. conversion operations which can be (un)commented. In the mur_fit.f you attached, the volume is presumed to be in Ang^3 and the energy in eV. Indeed, your data file has the volume corresponding to the lowest energy = 4.61 (presumably Ang^3), and the corresponding energy = -8362.950245 (presumably eV). You'll find nearly these values, expressed in Bohr^3 and Ry,in the output from your test. The bulk modulus is crazy (and does not fitinto the print format) because the units (at least in the 1st column)are most probably wrong: for me, the values like 4.61 do not look like the volume in Ang^3.It may well be the lattice constant, in which case you'd need to activate a line to convert it into volume in Bohr^3. Summarizing: 1. Use the units of your convenience, but check that the results after the fit are correct. 2. Make sure that the energy (as in column 2) is given exactly per volume (as in column 1).Do not forget that e.g. the fcc cube has 4 atoms whereas its primitive cell has only one etc. Good luck Andrei Postnikov - Manjeet Bhatia <manjeetbhati...@gmail.com> a écrit : >Thank you for your reply,please find attached files. On Thu, Aug 11, 2016 at 6:20 PM, Anita Rani <ranianit...@gmail.com> wrote: Please send me your murn.f file and .DAT file. > I will check and then suggest the changes. > On Aug 11, 2016 12:20, "Manjeet Bhatia" <manjeetbhati...@gmail.com> wrote: Dear Siesta users, I am getting volume, energy and pressure in corresponding units but not getting bulk modulus, please suggest if i have to change something in mur_fit.f fortran file. My values are: # Equilibrium volume, a.u.^3 : 31.125644 # Minimum energy, Ry : -614.470876 # Bulk modulus, Kbar : # BP :7.709490 > >
Re: [SIESTA-L] deal with the .RHO file of siesta
Dear zyynihao, rho2xsf is not built from a single source file rho2xsf.fbut demands some others, that's why a Makefile is providedin the directory which contains the source. Chances are,the Makefile will catch up the compiler settings from the Siestainstallation, or otherwise you define FC by hand, e.g.,FC = ifortFFLAGS = -Oand type "make". It is not necessary to install and run the scriptson the same platform where you run Siesta; you can transfer the .XV, .RHO etc. filesto your local computer and run rho2xv there (if the coding of binary files is compatible,of course).Another hint: there is a README file in the distribution.In case of doubt / problems, go get the code from http://www.home.uni-osnabrueck.de/apostnik/download.htmlGood luckAndrei Postnikovan you does not compuile - 木材a écrit : >Dear all ! >when open Utile/contribe/Apostnikov and find the rho2xsf.f ,then I carry out >the order "ifort -o rho2xsf rho2xsf.f" and it show in the picture,I hope for >geting your hlep,how should I do?
Re: [SIESTA-L] Question: Proper SolutionMethod for relaxing large systems
Dear Bahareh:> But I recieve error at the very beginnig of the calculation - what kind of error? (What was the error message)?> and the job terminates badly!- how badly?It is difficult to help you as you give no details, but let me guess. You are using some old version of Siestawhere the SolutionMethod by default was diagon for less than 100 atomsand OrderN for beyond 100 atoms. That's OK, just define SolutionMethod diagonand go ahead. Yes you can. If the calculation is too heavy that's another story;you can always try OrderN later (if your system has a gap, etc. etc.)Good luckAndrei Postnikov - Bahareh Fakhraei <fakhraei.baha...@gmail.com> a écrit : > Thak you so much dear Andria, So you suggest that I can use diagon for a system consisted of about 730 atoms?But I recieve error at the very beginnig of the calculation and the job terminates badly!I thought it might be because of that. But if that's ok with diagon, I must go over it again and give it another shot. Sincerely yours, B. Fakhraei On Tuesday, August 2, 2016, Andrei Postnikov <andrei.postni...@univ-lorraine.fr> wrote: Hi, > I think there must be some misunderstanding: > the Siesta rule of thumb about ~100 atoms implies only that > for large enough system, O(N) might ultimately provide > better scaling than diagon, and ~100 is a rough estimate > of when the crossing of ~N^3 and ~N tendencies happens in practice. > However, O(N) only works with large enough band gap, and is otherwise fragile. > There is nothing wrong about using diagon to systems of any size, > and 100 atoms are not that many to make it markedly unpractical. > I'd say, go ahead with diagon as long as the calculation time > is not a problem. If it starts to become an issue > and you think you can gain by switching to O(N), > do some test on whether it works correctly and is indeed faster. > > Best regards > > Andrei > > > - RCP <pasia...@cnea.gov.ar> a écrit : > > Hi, > > > > I'm not an expert here, but the naive answer is "use order N > > method". However, that depends on the physics of your system, > > for instance, O(N) cannot be applied to metals. > > > > Regards, > > > > Roberto > > > > > > On 08/01/2016 05:42 PM, Bahareh Fakhraei wrote: > > > Hello every one, > > > > > > For relaxing the large systems (more than 100 atoms) in SIESTA, seemingly > > > *diagon* is not a good option for solution method. Why and What should be > > > used instead? > > > > > > yours, > > > > > > B. Fakhraei > > > > > > A beginner to SIESTA program > > -- >
Re: [SIESTA-L] Question: Proper SolutionMethod for relaxing large systems
Hi, I think there must be some misunderstanding: the Siesta rule of thumb about ~100 atoms implies only that for large enough system, O(N) might ultimately provide better scaling than diagon, and ~100 is a rough estimate of when the crossing of ~N^3 and ~N tendencies happens in practice. However, O(N) only works with large enough band gap, and is otherwise fragile. There is nothing wrong about using diagon to systems of any size, and 100 atoms are not that many to make it markedly unpractical. I'd say, go ahead with diagon as long as the calculation time is not a problem. If it starts to become an issue and you think you can gain by switching to O(N), do some test on whether it works correctly and is indeed faster. Best regards Andrei - RCPa écrit : > Hi, > > I'm not an expert here, but the naive answer is "use order N > method". However, that depends on the physics of your system, > for instance, O(N) cannot be applied to metals. > > Regards, > > Roberto > > > On 08/01/2016 05:42 PM, Bahareh Fakhraei wrote: > > Hello every one, > > > > For relaxing the large systems (more than 100 atoms) in SIESTA, seemingly > > *diagon* is not a good option for solution method. Why and What should be > > used instead? > > > > yours, > > > > B. Fakhraei > > > > A beginner to SIESTA program --
Re: [SIESTA-L] graphene
In graphene, the bands cross the Fermi energy in the K point. If your k-point sampling is sparsebut it so happens that it includes exactly the K point,you'll have a peak at the Fermi energy (as in your case).As either the density of k-point would increase, orthe tetrahedron integration done instead of sampling(not implemented in Siesta, to my knowledge),the importance of the K point will diminish,and the "correct" structure of DOS (smoothly converging to zeroat the zero gap) would gradually manifest itself. - Zahra Talebi <talebi_z2...@yahoo.com> a écrit : >thanks for your answer, but you know graphene is zero band gap material, for graphene with 200 atom DOS diagram shows the zero band gap but for 288 atom my DOs diagram show a pick on the point of fermi energy. do you think that it will correct if I increase the kpoints. this is my question On Monday, July 25, 2016 8:29 PM, Andrei Postnikov <andrei.postni...@univ-lorraine.fr> wrote: Hi Zahrayour both DOS distributions are essentially quite close,only that one of them (that with 288 atoms) uses much less k-points(the 6*6 mesh). The resulting energy values are too sparse,that's why you see strongly fluctuating DOS curve.The other calculation (with 200 atoms) uses much more k-points (10*10 mesh)and the resulting DOS is much more smooth.Solution:- increase the density of k-points (at least just for PDOS calculation, see the switchPDOS.kgrid Monkhorst Pack )and/or- increase the broadening of DOS by e.g.setting Electronic Temperature, or set the appropriate broadening parameter inProjectedDensityOfStatesblock.The exact "fix" depends on what exactly do you want:an accurate smooth curve or merely a reasonably looking one,without strong unphysical fluctuations.After all, this is just about the graphic representation...Best regardsAndrei- Zahra Talebi <talebi_z2...@yahoo.com> a écrit :>hi every bodyi am working on graphene surface, and I got a problem. I am trying the different super cell. for supercell with 200 atom I got the right DOS and band file. but for supercell with 288 atom I got wrong DOS but right band. I am attaching the fdf file to my e-mail. does any body knows the problem. I also attached the DOS file I really appreciate your guidezahra
Re: [SIESTA-L] graphene
Hi Zahra your both DOS distributions are essentially quite close, only that one of them (that with 288 atoms) uses much less k-points (the 6*6 mesh). The resulting energy values are too sparse, that's why you see strongly fluctuating DOS curve. The other calculation (with 200 atoms) uses much more k-points (10*10 mesh) and the resulting DOS is much more smooth. Solution: - increase the density of k-points (at least just for PDOS calculation, see the switch PDOS.kgrid Monkhorst Pack ) and/or - increase the broadening of DOS by e.g. setting Electronic Temperature, or set the appropriate broadening parameter in ProjectedDensityOfStates block. The exact "fix" depends on what exactly do you want: an accurate smooth curve or merely a reasonably looking one, without strong unphysical fluctuations. After all, this is just about the graphic representation... Best regardsAndrei - Zahra Talebia écrit : >hi every body i am working on graphene surface, and I got a problem. I am trying the different super cell. for supercell with 200 atom I got the right DOS and band file. but for supercell with 288 atom I got wrong DOS but right band. I am attaching the fdf file to my e-mail. does any body knows the problem. I also attached the DOS file I really appreciate your guide zahra
[SIESTA-L] fdf parameters
Dear Amrish: Let's sort things out:1. The energy shift is about constructing your basis functions. There is nothing you need to teston a huge supercell. Make the usual tests on small (minimal; single cell) systemsto be sure that your basis set construction permits you to obtain the reasonable description(of whatever your aims are: lattice parameters, bulk modulus, ...)2. The mesh cutoff, as it seems to me, a priori is not supposed to depend on the supercell size.Again, after you test it on prototype small systems, you can use the same valuefor large supercell.3 my workstation is not so powerful to deal with system like 500 atoms... - then:why use such workstation? Or: why use 500 atoms? Or: why use Siesta?Best regardsAndrei i am facing problem in my fdf file to take value of some parameters like mesh cutt off and energy shift parameters. my workstation is not so powerful to deal with system like 500 atoms so i don't want to waste time while test calculations on these parameters.can any one explain an alternate way to get approximate but the accurate value of mesh cutt off and energy shift for system consisting 500 atoms. -- warm regards amrish sharmaresearch scholarDr. Isha Mudahar grouppunjabi university patialahii all
Re: [SIESTA-L] script for energy vs lattice constant convergence.
Dear Javier, Mayuri: There was a comment on the related issue (how to run a sequence of similar calculations, NOT how to plot a curve) some time back, please see http://www.mail-archive.com/siesta-l%40listserv.uam.es/msg02458.html with the batch script integrated into the text of that message (from #!/bin/sh to echo $0" "$1" finished on "$dte >> $wrk0/LOG ) The script is still operational... Best regards Andrei On Wednesday, July 20, 2016 19:04 CEST, Javier Junquerawrote: Dear Mayuri: If you want to run various repetitive calculations changing the lattice constant in a compact form, you can implement it yourself starting from some of the examples provided in the Util/Scripting directory in the Siesta package. You can also take a look to the instructions and examples provided in Util/JobList. It contains utilities to organize, to run or queue, and to collect results of multiple siesta jobs. It is particularly suited for preliminary convergence tests, and it uses only fortran codes and minimal shell scripts. Once you have the energy versus lattice constant curve, you can use the script to fit it to the Murnaghan equation of state. You can download it from: http://personales.unican.es/junqueraj/ and follow the links: A self-explained Siesta tutorial Set of self-explained Siesta exercises Computing structural and electronic properties of materials Lattice constant, bulk modulus and equilibrium energy of solids. Hope this helps, Javier > hi all siesta users, > i want energy vs lattice constant graph. so i need script which is compatible > with siesta, i search about how to make script to get minima in ene vs Latt. > Cons. but proper result is not obtained. > please help me..