Home

FULL-POTENTIAL LMTO PROGRAMS ”NMT” USER's MANUAL

1. PARAMETER NBASMAX 9000 PARAMETER MBASMAX 1500 BASIS SET EXPANSION IN PLANE WAVES OPTIMAL NUMBER OF PLANE WAVES C KK CELL DATA PARAMETERS FEA AA hhh PARAMETER NSHEMAX 60 PARAMETER NMSHMAX 6000 PARAMETER NPOLMAX 12 MAX NUMBER OF POINTS IN INTERSTITIAL MESH MAX NUMBER OF SURFACE GRID POINTS MAX NUMBER OF POLYNOMS C eH FREQUENTLY USED SETTINGS ooo kkk kkk kkk kkk PARAMETER NPRECISION 2 PARAMETER NGGAMAX 3 PARAMETER NOPRMAX 48 7 PARAMETER NENVMAX 30_ PARAMETER NVCMAX 6000 PARAMETER NTERMAX 15 7 WITH DOUBLE PRECISION WITH GGA FOR EXCHANGE CORRELATION MAX NUMBER OF GROUP OPERATIONS MAX NUMBER OF POLYHEDRON PLANES MAX NUMBER OF STRUCTURE VECTORS MAX NUMBER OF BROYDEN ITERATIONS Cok KBR k k k k k k ok ok COMBINATIONS KKK K K 2K K K K 2K K FK 2K K FK K 2K K K K 2K K K K K ok K PARAMETER NLENMAX NRADMAX 1 4 NPAT NPLWMAX 1 NSPINMAX PARAMETER NWIGMAX LMAXT 1 2 LMAXT 1 2 LMAXT 3 3 PARAMETER LMMAXV LMAXV 1 2 PARAMETER LIMLMB LMAXB 1 2 PARAMETER LIMLMT LMAXT 1 2 PARAMETER LMMAXV1 LMAXV 2 2 PARAMETER LIMLMB1 LMAXB 2 2 PARAMETER LIMLMT1 LMAXT 2 2 PARAMETER PARAMETER PARAMETER LMMAXV2 LMAXV 1 2 LMAXV 1 2 LMAXV 3 3 LIMLMT2 LMAXT 1 24LMAXT 1 2 LMAXT 3 3 LIMLMB2 LMAXB 1 2 LMAXB 1 2 LMAXB 3 3 8 1 Main parameters This set of parameters is applicable
2. The input files to the program are INIFILE and STRFILE The output is the number of irreducible k points generated according to the point group and the list of points which can be stored in the output file An additional option is provided to generate the irreducible points with its minimal lenght This means that among a set of all k G points G is a reciprocal lattice vector the vector with minimal lenght will be picked out 12 7 Program SURF The directory SURF contains sample input files for the Data Explorer The latter is the UNIX software which can be used to plot 3D graphs like Fermi surfaces Sample input files and the programs are provided 50 13 ADDITIONAL INPUT HUBFILE WARNING Due to a permanent development of this part of the program some input data may differ from realization The LDA U method is described in Ref 7 It turns out to be drastically improve the results comparing to LDA when doing the calculations of the strongly correlated systems Another option avaiable here is constrained LDA calculations See Ref 8 for a complete description To include LDA U and or LDA C option a special HUBFILE must be created An example of this file for NiO system nmt dat nio nio hub is given below INPUT DATA FOR INCLUDING CORRELATED STATES GENERAL SETTINGS LDA U1 1 Scheme Ry l Units eV Ry avaiable Cubic 4 Cubic Spherical harmonics Complex Real Complex input output Yes N y n occ
3. necessary when plotting the Fermi surface Then BNDFILE which contains energy band in the IBZ must be created see the description of the input parameter ibnd 2 in chapter describing INIFILE Then use MESH program to set up energy band in the 3 dimensional cubic mesh No special configuring of the program is required Just compile all the files link them to get executable make exe The input to the program are 3 files INIFILE STRFILE and BNDFILE After that specify three vectors and their divisions which set a cubic mesh FFT mesh in the space The output from the program is the values of the band energies at all points of this cubic mesh Since no interpolation is assumed here the irreducible points generated with BNDFILE must commensurate with this cubic mesh As additional input you must select which bands are necessary to widthdraw e g only those which cross the Fermi energy The Fermi surface can then be plotted using Data Explorer see the description of program SURF or other existing software 12 4 Program MLSQ Program MLSQ makes least squares polynomial fit to the set of data specified in the input file The program is useful to find different derivatives of the total energy forces etc No special configuring of the program is required Just compile the file mlsq f link it and get executable mlsq exe Sample input data file is provided 12 5 Program PLOT A few programs are located in this directory Also a library of the russ
4. xy x2 y2 3z2 1 REAL DN yz ZX xy x2 y2 3z2 1 IMAG DN REAL UP ZX xy x2 y2 3z2 1 IMAG UP yz ZX xy x2 y2 3z2 1 REAL DN yz ZX xy x2 y2 3z2 1 IMAG DN yz ZX xy x2 y2 0 0000001 0 State 3d for Ni2 spin up dn Hubbard potentials yz 0 2507736 0 0 0000297 0 0 0000297 0 0 0002219 0 0 0001281 0 yz 0 0000000 0 0 0000000 0 0 0000000 0 0 0000002 0 0 0000001 0 yz 0 2122868 0 0 0000231 0 0 0000231 0 0 0000512 0 0 0000295 0 yz 0 0000000 0 0 0000000 0 0 0000000 0 0 0000002 0 0 0000000 0 0000001 0 0000003 ZX xy 0000297 0 0000297 2507736 0 0000297 0000297 0 2507736 0002219 0 0000000 0001281 0 0002563 ZX xy 0000000 0 0000000 0000000 0 0000000 0000000 0 0000000 0000002 0 0000000 0000001 0 0000003 ZX xy 0000231 0 0000231 2122868 0 0000231 0000231 0 2122868 0000512 0 0000000 0000295 0 0000591 ZX xy 0000000 0 0000000 0000000 0 0000000 0000000 0 0000000 0000002 0 0000000 0000000 0 0000002 13 1 General Settings No case sensitivity is assumed in the input parameters described below gt LDA U1 1 Ry 2 Cubic Complex 7 Yes d Append 2 e Scheme A number of different formulae have been programmed for the LDA U technique For a complete 0 0000000 x2 y2 0 0002219 0 0002219 0 0000000 0 2359186 0 0000000 x2 y2 0000002 0000002 0000000 0000000 0000000 x2 y2 0000512 0
5. If cout 1 the program will terminate and you have to remove old nio out manually A useful hint here is i see the OUTFILE after the execution ii decide whether the execution was OK or some mistake occurs iii if it was OK rename nio scf to nio scO and nio out nio ou0 if it was not OK remove bad output nio scf and nio out correct input data and start the job again This always allows to keep the switches iscf 2 and iout 2 The last option with the OUTFILE is iout gt 2 In this case the output will be redirected to the channell assigned to the terminal output Specify iout 6 for the terminal output under UNIX systems Specify iout 5 for the terminal output under VMS systems The output from the program is made using standard WRITE UNIT IUN where IUN 1 when iout lt 3 and IUN IOUT when out gt 2 3 6 Other Data for NMT pack Other Data for NMT Pack 1 4 nff nef 8 8 8 2 n1 n2 n3 nc 32 32 32 0 02 0 04 1 5 nfft1 nfft2 nfft3 epsR epsG kbz rbz e nff number of filled bands in the main valence panel above the semicore If nff 1 this number will be determined automatically from the knowledge of the total valence charge and the guessed number of bands crossing the Fermi level see below e nef number of bands crossing the Fermi level or larger This parameter is used for calculating the Fermi energy and DOS If this number is not exactly known or if empty bands should also 22 be taken into account for plotting DOS a
6. nmt lib cell The program reads INIFILE and generates atomic cells according to the weights specified The output information will contain the muffin tin spheres and circumscribed spheres found in this way which should be placed back into the INIFILE See the description of nmt lib cell program in the chapter USING LOCAL LIBRARY for more details Imaz t maximal angular momentum l not l 1 for the basis functions i e for the decom position of the tails coming from other atoms Normally it is 4 in the ASA calculation 6 for PLW calculation and 8 for the CEL calculation Imax b maximal l actually included in basis This may not be less than number of non zero columns in the description of any valence states see below lmazx v maximal l value for the expansion of the charge density and the potential in spherical harmonics Normally it is the same as Imaz t The following block is repeated for each from 1 to nkap d T states for E 0 6 Ry 34 main quantum numbers 00 basis set 00 choice of Eny 5 0 5 Eny 0 4 0 Dny f states for E 0 4 Ry 4 main quantum numbers 17 FOoOWrFAB HF OWF basis set choice of Eny 0 5 0 5 Eny 3 0 4 0 Dny states for E 1 4 Ry main quantum numbers basis set choice of Eny 0 5 0 5 Eny 3 0 4 0 Dny e sp df states for E 0 4 Ry This is a comment In each of the following lines only Imax b 1 first numbers are read the only exception is
7. 0000000 0000000 x2 y2 0000000 0000000 0000000 0000000 0000000 53 are 3z2 1 0000000 0000000 0000000 0000000 0000000 3z2 1 0000000 0000000 0000000 0000000 0000000 3z2 1 0000000 0000000 0000000 0000000 0000000 3z2 1 0000000 0000000 0000000 0000000 0000000 GO O O O O oo0ooo GO O OO O O G OTO C9 REAL UP ZX xy x2 y2 3z2 1 IMAG UP yz ZX Xy x2 y2 3z2 1 REAL DN yz ZX xy x2 y2 3z2 1 IMAG DN yz ZX xy x2 y2 3z2 1 State 3d for Ni2 spin up dn Hubbard potentials are yz ZX xy x2 y2 3z2 1 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 yz ZX xy x2 y2 3z2 1 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 yz ZX xy x2 y2 3z2 1 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 yz ZX xy x2 y2 3z2 1 0 0000000 0 0000000 0 0000000 0 0000
8. 0000000E 00 Calculated average square of electron velocities lt Vx 2 gt 0000000E 00 lt Vy72 gt 0000000E 00 lt Vz 2 gt 0000000E 00 Calculated bare plasma frequencies in eV om_p x 0000000E 00 om_p y 0000000E 00 om_p z 0000000E 00 of fully filled bands 16 input 14 of bands crossing Ef 0 input 4 Energy bands at the Gamma point for spin up states are 58822 48708 21930 23623 23623 40584 40584 42354 47532 47532 55798 65522 65522 65546 65826 7 7 Constructing the charge density When the charge density is calculated program RHOFUL see source file rhoful f the following output lines allow to check for the correct normalization If it is more the a few per cent more in the ASA something is going wrong If overlap matrix is not positive define or the ghos bands occur the renormalization coefficient can strongly deviate from unity Watch out then for the mistakes in the INIFILE kKKKKKK BZINT finished CPU time 354 7200 FKK Kk k KR KKK kkkKAK ROFULL started CPU time 354 7200 k K Kk AK aK CK Valence charge in whole elementary cell must be 44 00000 Valence charge found via fourier transform is 44 10053 Renormalization coefficient of the val density is 9977204 7 8 Renormalizing core levels The renormalization of the deep core levels program RENCOR see source file rencor f for each atom results in the following output table xxxkxxk ROFULL finished CPU time 37
9. 2 l d 1 6 Total of non zero components found 27 3 m m m m 6 5 2 4 gt 53 E2 el a S26 al 1 0 0 0 1 2 This message gives the accuracy in expanding pseudo Hankel functions for different 1 s in plane wave in the interstitial region for the input cutoff estimated from the input FFT grid The ratio RH S H S is the ratio between pseudoHankel function and the real Hankel function at the sphere boundary The pseudoHankel function should coincide with the Hankel function in the interstitial region but it is smooth inside the sphere regular at the origin The ratio GH S RH S is the ratio between the pseudoHankel function found using Fourier transform and calculated in the real space Diviations show the accuracy The accuracy is worsen for higher l s but is acceptable usually a few percent l o 1 ou 3 n 4 5 6 Nplw Ecut Ry RH S H 3862 3862 3862 3862 3862 3862 3862 91 91 91 91 91 91 91 00 00 WO 00 1 00000 99997 99978 99895 99642 99036 97841 S GH S RH S 1 0000 1 0000 99969 99853 1 0027 1 0162 1 0063 The information below contains the set up for using the Ewald method to sum up the structure constants Minim differences between tail energies and the poles of free electron Green function show how far is the singularity on the real axe Note that when using the positive x a small i
10. 3 or 1 4 Units Electronvolts or Ry input units for the energies in this file are avaiable Specify either eV or Ry e Harmonics The input can be given either in spherical or in cubic harmonics representation Input Output can be either real or complex e Occupancies in this file are either listed below yes key or absent no key e Occupancies in this file will be either overwritten during the execution overwrite option or appended append option to the end of the file The latter is recommended for use 13 2 Description of Correlated States 2 of correlated states 1 3d 0 588 0 601227 0 378773 atom nl f0 f2 f4 Slater integrals 2 3d 0 588 0 601227 0 378773 atom nl f0 f2 f4 Slater integrals e of states which are considered as correlated must be given e for each of the correlated state selected by atom number main quantum number orbital quantum number Slater integrals must be given For d electrons the knowledge of on site Coulomb U and exchange integral J defines set of Slater integrals as follows U FW J FO F 14 and FO F 0 625 13 3 Table of Occupation Numbers At the end of the HUBFILE matrix of the occupation numbers and or correction to the LDA potential for each of the correlated state must be given A specific input is selected by the first key 1 stands for the input occupation matrix for each correlated state 2 stands for the occupa
11. along b c a 1 0000 orthorombicity along c ibas 1 basis in Cartesian sys ibz 1 automatic BZ choice icalc 1 using built in calc istrn 1 distort cutoff sphere ndivi 4 polyhedron generation ndiv2 10 polyhedron surface grid nvec 300 vectors in Evald method alpha 1 0000 splitting factor there lt PRIMITIVE TRANSLATIONS gt L 50000 gt 50000 gt 1 0000 Rix Rly Riz 50000 gt 1 0000 gt 50000 R2x R2y R2z L 1 0000 gt 50000 gt 50000 R3x R3y R3z lt BASIS ATOMS IN CELL gt 00000E 00 00000E 00 00000E 00 for Nil L 1 0000 1 0000 gt 1 0000 for Ni2 L 50000 gt 50000 gt 50000 for 0 1 5000 gt 1 5000 gt 1 5000 for 0 lt STRAIN MATRIX gt 1 0000 00000E 00 00000E 00 Sxx Sxy Sxz 00000E 00 1 0000 00000E 00 Syx Syy Syz 00000E 00 00000E 00 1 0000 Szx Szy Szz lt INVERSE STRAIN MATRIX gt 1 0000 00000E 00 00000E 00 Rxx Rxy Rxz 00000E 00 1 0000 00000E 00 Ryx Ryy Ryz 00000E 00 00000E 00 1 0000 Rzx Rzy Rzz lt POINT GROUP DESCRIPTION gt ikov C Cubic system lt RECIPROCAL LATTICE gt 2 50000 gt 7 50000 gt 1 5000 Glx Gly Giz L 50000 gt 1 5000 gt 7 50000 G2x G2y G2z C 1 5000 gt 7 50000 gt 7 50000 G3x G3y G3z lt BRILLOUIN ZONE gt E 33 50000 50000 1 5000 K1x Kly Klz 50000 1 5000 50000 K2x K2y K2z 1 50
12. and upper energy limits and the number of divisions like for instance 0 6 0 8 400 Then in the last iteration partial densities of states are put in the same file for 400 energies in the example above As soon as there is no well justified way to define partial l decomposed densities of states in the full potential calculation what is actually calculated is the total density of states per unit cell decomposed proportionally to n contributions within the muffin tin spheres The density of states can be plotted using plotdos exe program located at nmtlib plot directory See chapter USING NMTLIB LIBRARY for details e iscf lt scffile gt This is the file for storing the output charge density The file is updated at the end of every iteration Since there are now two SCFFILEs one for the input charge density and another is for the output charge density they must have different names in order to avoid rewritting of the input file Usually the input charge density file has an extension scO nio scO in the example for NiO while the output SCFFILE has an extenison scf nio scf in the example 21 for NiO Specifying iscf 0 will not produce the output SCFFILE Actually it will be produced and erased at the end of the execution Specifying iscf 1 will produce the output SCFFILE If the file with the same name already exists the program will terminate Specifying iscf 2 will produce the output SCFFILE regardless if the file with the same na
13. crystalline structure For example HCP structure cannot be described in the coordinate system with the rotations about 60 degrees along x axe If it is necessary to use another rotational system set symmetry code to A arbitrary and give the file name of KOVFILE describing your own rotational system afterwards See chapter ADDITIONAL INPUT KOVFILE for details 4 7 Primitive Translations for BZ For ibz 1 these vectors are ignored If 2b2 0 the whole Brillouin zone is represented as a parallelepiped the three primitive translations of which are set in this lines The parallelepiped is divided along three directions number of divisions is specified in the INIFILE then the group symmetry operations are applied to sort out non equivalent k points The Brillouin zone should correspond to the translation vectors as defined in Primitive Translations prior to applying eventual orthorombic displacement and strain to the latter In this case the program will automatically introduce the correct distortion of the Brillouin zone 4 8 High symmetry direction settings are used for calculating the energy in a set of directions listed here The program will do it if the parameter ibnd is set to 1 in the list of output parameters of the INIFILE Note that if it is necessary to output the energy bands at the tetrahedron corners set parameter ibnd 2 4 9 Settings for PLOBANDS program Not used by the fp lmto code This is a list of symbols which
14. defines how to plot the energy bands using nmtlib plot plobands exe utility See chapter USING NMTLIB LIBRARY for details 29 5 INPUT CHARGE DENSITY FILE SCFFILE The charge density distribution in the unit cell is stored in the unformatted SCFFILE This is the third input file to the NMT package Since the representation of the charge density is different for the ASA CEL or PLW programs this file is unique for each of it That means that you cannot start execution say of the PLW package giving as an input file the output SCFFILE made by the ASA program To prepare the input SCFFILE at the begining there exists an auxilary program SCFM located inside nmt lib scfm directory It uses Mattheiss s prescription for constructing the input charge density as a superposition of the atomic charge densities To be able to do it the directory atomdat must exist and contain a library of the atomic charge densities for all elements of the periodic table The atomdat directory is distributed together with the NMT package To properly configure SCFM i go to nmt lib scfm directory ii edit the file makescfm f iii locate a string C HERE IS THE PATH TO THE ATOMIC DATA DIRECTORY DIR u a4 savrasov atomdat iv change this path to your own path v compile link all the files and get an executable make exe file To use the you also need INIFILE and STRFILE to be prepared because for the proper execution the crystalline st
15. for all NMT programs Al e NSPINMAX is either 1 for non spin polarized version of the program or 2 for spin polarized version including spin orbit coupling parameter nspin in INIFILE cannot exceed this value e NPAT is the maximum number of atoms allowed to consider parameter natom in INIFILE cannot exceed this value e NSORTMAX is maximum number of atoms of different kind parameter nsort in INIFILE cannot exceed this value e NSYMMAX is maximum number allowed for non zero components in the expansion of the charge density in spherical harmonics If LMAXV 1 gives the total number of components many of them can be skipped because of the symmetry of the crystall Therefore only non zero components are stored inside the program The list of terms allowed by symmetry is printed at the beginning of the OUTFILE The maximal number of terms found for any atom must not exceed NSYMMAX Usual choose NSYMMAX as one half of LMAXV 1 or so e LMAXV is the maximum value for the one center spherical harmonics expansion of the charge density and the potential parameter Imaz v in INIFILE cannot exceed this value e LMAXB is the maximum value of l for the orbitals actually included in the basis parameter Imaz b in INIFILE cannot exceed this value e LMAXT is the maximum value of 1 for the expansion of tails for the orbitals coming from different sites parameter Imax t in INIFILE cannot exceed this value Usually LMAXT LMAXV e N
16. of the self consistent calculation The names of these files are explicitly given at the beginning of the rat lt atomname gt file and they can be changed Putting 0 as the 1 st character of a file name suppresses creation of the corresponding file Note that most of the rat lt atomname gt files contain starting energies which are already self consistent eigenvalues for atoms They can be used as a useful reference during the work x FREE ATOM CALCULATION FOR FE 3D6 482 1 den fe DENSITY FILE 1 tab fe TABLE OUTPUT 1 inf fe SCF INFO FILE 10 4 J JCORE Mus De Lo ZN IONICITY 0 3 30 PHI ITERMAX 5 E 5 1 E 3 EPS LEVEL EPS POT 3 KPOT OF EXCHANGE SEE TEXT O FEX 1 E DEPENDENT SHAM KOHN POTENTIAL 0 ATOMIC NUMBER IF NON POINT NUCLEUS 1 INITIAL ENERGIES 1 listed below O unknown ORBITAL N L J ELECTRONS EIGENVALUE 181 2 1 0 0 0 0 5 2 0 5 1321178E 02 281 2 2 0 0 0 0 5 2 0 6 0166799E 01 2P1 2 2 0 1 0 0 5 2 0 5 1966747E 01 2P3 2 2 0 1 0 1 5 4 0 5 1051799E 01 351 2 3 0 0 0 0 5 2 0 6 9521535E 00 3P1 2 3 0 1 0 0 5 2 0 4 5606623E 00 3P3 2 3 0 1 0 1 5 4 0 4 4464996E 00 3D3 2 3 0 2 0 1 5 4 0 6 5538833E 01 3D5 2 3 0 2 0 2 5 2 0 6 4413460E 01 481 2 4 0 0 0 0 5 2 0 5 0025746E 01 48 12 2 Program CHIQ The program CHIQ calculates imaginary real part of the susceptibility function in the constant matrix element approximation The corresponding formula for the real part is give
17. pseudofunction pseudocharge density or pseudopotential The values of the functions must coincide with 46 the values of the pseudofunction in the interstitial region Make plots using LINE DAT GRID DAT with the standard software programs Analogious program exists in the nmtcel lib grid directory Since NMTCEL uses only one center spherical harmonic expansions no pseudocharge densities pseudopotentials calculated These one center expansions often are slowly convergent and it is not advised to make good charge density plots with the CEL package No plotting facilities for the charge density exists in the ASA package 11 3 Program SCF1 Program SCF1 is helpful to generate SCFFILEs for supercells If a supercell is made of a few original cells it first advised to make self consistent SCFFILE for the small cell and then built input supercell SCFFILE using it At the moment the program SCF 1 only exists for the ASA and CEL verisons of the NMT Location nmtasa lib scf1 and nmtcel lib scf1 directories No special configuring is required just compile all the programs link them to get make exe executable module The program asks for the SCFFILE of the original cell then a number of atoms number of sorts in the supercell is specified After that you must indicate how to copy the atoms from the old cell to the new supercell At the end name for the supercell SCFFILE must be given 11 4 Program SCFM Program SCFM is designed to construct
18. the Eny line where lmaz b 2 numbers are read see below For instance if Imax b 2 and you define something for the f states this will have no effect e 448 4 main quantum numbers e 1 1 1 0 basis set either 1 if the state is included in basis or 0 In the example for NiO 2k basis set is chosen for Ni s p and d states with the tail energies 0 4 and 1 4 Ry Usually the distance between energies is 1 Ry or so to avoid linear dependency of LMTOs of the same L character but different by 2 The same 24 basis is chosen for the broad 2p states of oxygen On the other hand narrow 2s oxygen states are treated with 1 0 6 Ry e 331 0 choice of E for each state it may be 0 E is taken from the Eny line and fixed throughout the iterations 1 D is taken from the Dny line and fixed throughout the iterations E are adjusted to these Dny 2 E is found as the energy of the bound state inside the atomic sphere Advisable for the states which are semicore like but treated as bands in the main valence panel If the bound state cannot be found the eigenvalue is lying in the continious spectrum the will be fixed according to Dy l 1 3 E is adjusted throughout the iterations to the center of gravity of the occupied band from the band structure calculation at the first iteration however the D from the Dny line is used and Eny ignored During the self consistency the Eny numbers will be stored to the SCFFILE In c
19. to be calculated e nmt lib directory containing the programs for helping to construct input data files and un derstand output information For example subdirectory nmt lib scfm contains the program which creates input charge density distrubtion Subdirrectory nmt lib cell containes the pro gram which helps to find the muffin tin sphere radii Subdirectory nmt lib grid creates two dimensional grid for plotting the self consistent charge densities Each subdirectory containes source data files f object files o and the executable file usually named make exe In contrast to the local libraries which exist inside nmtasa nmtcel nmtplw there also exists a global library directory nmtlib The latter containes general purpose application programs The programs understand output data files of all NMT codes These programs are e nmtlib plot programs to plot energy bands and densities of states e nmtlib chiq program to calculate real imaginary part of the susceptibility in the constant matrix element approximation e nmtlib mlsq least squares fit program and sample input data to fit total energies and forces in order to find equillibrium configuration bulk modulus phonon frequencies etc e nmtlib mesh program to generate whole 3D grid of bands from the knowledge of the data at the irreducible part e nmtlib surf sample input data to plot 3D graphs like Fermi surfaces by Data Explorer e nmtlib q
20. 00 50000 50000 K3x K3y K3z Cell Volume 260 9273 The primitive translations positions of atoms in the basis reciprocal lattice vectors and the Brillouin zone data are given after taking into account b a and c a ratios and after applying the strain matrix The data in direct space are given in the units of lattice parameter the data in reciprocal space are given in units 2r The cell volume Qe is printed in atomic units 7 The following output contains the information about the number of plane waves to be used for the repre sentaion of the charge density and the potential in the interstitial region This value together with the energy cutoff is found according to the input FFT grid 32 32 32 in this case Number of plane waves to be used 3862 Plane wave energy cutoff will be 91 81580 Ry Reciprocal lattice sphere radius 12 27803 2pi a Fast fourier transform divisions 32 32 32 FFT optimal divisions calculated 32 32 32 Note that the line FFT optimal divisions calculated is for your information only if optimally calculated divisions do not coincide with the input divisions place optimal divisions into the INIFILE The following output messages contain information about using the acceleration in calculating the interstitial potential matrix elements see also a description of parameters kbz rbz in the INIFILE The last message here compares plane wave series read from the input SCFFILE and generated in this set up In the e
21. 000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 4 HUBBARD CORRECTION TO THE LDA POTENTIAL State 3d for Nil spin up dn Hubbard potentials are yz ZX xy x2 y2 3z2 1 0 2122868 0 0000231 0 0000231 0 0000512 0 0000295 0 0000231 0 2122868 0 0000231 0 0000512 0 0000295 0 0000231 0 0000231 0 2122868 0 0000000 0 0000591 0 0000512 0 0000512 0 0000000 0 2786051 0 0000000 o 0 0000295 0 0000295 0 0000591 0 0000000 0 2786051 yz ZX xy x2 y2 3z2 1 0 0000000 0 0000000 0 0000000 0 0000002 0 0000000 0 0000000 0 0000000 0 0000000 0 0000002 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000002 0 0000002 0 0000002 0 0000000 0 0000000 0 0000000 0 0000000 0 0000000 0 0000002 0 0000000 0 0000000 yz ZX xy x2 y2 3z2 1 0 2507736 0 0000297 0 0000297 0 0002219 0 0001281 0 0000297 0 2507736 0 0000297 0 0002219 0 0001281 oo 0 0000297 0 0000297 0 2507736 0 0000000 0 0002563 0 0002219 0 0002219 0 0000000 0 2359186 0 0000000 0 0001281 0 0001281 0 0002563 0 0000000 0 2359186 yz ZX xy x2 y2 3z2 1 0 0000000 0 0000000 0 0000000 0 0000002 0 0000001 0 0000000 0 0000000 0 0000000 0 0000002 0 0000001 0 0000000 0 0000000 0 0000000 0 0000000 0 0000003 0 0000002 0 0000002 0 0000000 0 0000000 0 0000000 54 REAL UP yz ZX xy x2 y2 3z2 1 IMAG UP yz ZX
22. 000512 0000000 0 2786051 0000000 x2 y2 0000002 0000002 0000000 0000000 0000000 oo0oooo o ooo oo0oooo Scheme O 0000000 are oooo ooo O OG Og 3z2 1 0001281 0001281 0002563 0000000 2359186 3z2 1 0000001 0000001 0000003 0000000 0000000 3z2 1 0000295 0000295 0000591 0000000 2786051 3z2 1 0000000 0000000 0000002 0000000 0000000 Units eV Ry avaiable Cubic Spherical harmonics Real Complex input output y n occupancies avaiable 3z2 1 REAL UP yz ZX xy x2 y2 3z2 1 IMAG UP yz ZX Ey x2 y2 3z2 1 REAL DN yz ZX xy x2 y2 3z2 1 IMAG DN yz ZX xy x2 y2 3z2 1 Overwrite Append occupancies description see also file nmt run hubpot f LDA standard LDA Can be used to withdraw LDA occupation numbers matrix LDA LDA U1 2 LDA4 occupation numbers must be given see below LDAG LDA LDAG U1 3 LDA4 U1 4 LDA4 below 55 U with the average occupanicies U in spherically averaged form U1 1 standard LDA U Starting occupancies must be listed see below U with another double counting Together with the starting occupancies LDA LC constrained LDA calculations A constrained part of the potential must be specified see LDA CU1 1 combination of constrained LDA with the LDA U techniqie scheme 1 1 1 2 1
23. 05 xy 0 0000005 0 0000005 0 0000000 0 0000000 0 0000000 x2 y2 0 0000003 0 0000003 0 0000005 0 0000000 0 0000000 3z2 1 State 3d for Ni2 spin up dn occupation numbers are yz ZX xy x2 y2 3z2 1 REAL UP 51 9440717 0000402 0000402 0 0004752 0002744 yz 0000000 0000000 0000000 0000005 0 0000003 yz 0 9500596 0 0000275 0 0000275 0 0000241 0 0000139 yz 0000000 0000000 0000000 0000003 0 0000002 o ooo oooo GO O O O O ooo G O O O O 0000402 9440717 0000402 0004752 0002744 ZX 0000000 0000000 0000000 0000005 0000003 ZX 0000275 9500596 0000275 0000241 0000139 ZX 0000000 0000000 0000000 0000003 0000002 oooo o oo0oooO 0000402 0000402 9440717 0000000 0005487 xy 0000000 0000000 0000000 0000000 0000005 xy 0000275 0000275 9500596 0000000 0000278 xy 0000000 0000000 0000000 0000000 0000004 FOR 0004752 O 0000000 O 0 0000000 0 oooo 0004752 1046295 x2 y2 0000005 0000005 0000000 0000000 0000000 x2 y2 0000241 0000241 0000000 9774884 0000000 x2 y2 0000003 0000003 0000000 0000000 0000000 0 0002744 0 0002744 0 0005487 0 0000000 0 1046295 3z2 1 0 0000003 0 0000003 0 0000005 0 0000000 0 0000000 3z2 1 0 0000139 0 0000139 0 0000278 0 0000000 0 9774884 3z2 1 0 0000002 0 0000002 0 000000
24. 1 0600 aR a a kK K k K k KK kkKKAK RENCOR started CPU time 371 0600 M kkk a K K K K K Orbital n 1 j tt el Levels Ry for Nil Zcor 12 000 1s1 2 1 0 1 2 2 600 7874 2s1 2 2 0 1 2 2 70 59245 2p1 2 2 1 1 2 2 61 48175 2p3 2 2 1 3 2 4 60 19977 3s1 2 3 0 1 2 2 6 755245 7 9 Evaluating total energy The work of the program evaluating the total energy program ENERGY see source file energy f starts by printing the partial numbers and densities of states for every atom and for every energy panel Different contributions to the total energy are printed afterwards x xxkxxk RENCOR finished CPU time 371 2300 FKK kk kK KK kK K kkKKAK ENERGY started CPU time 371 2800 k K kk aK aK CK Occupation numbers for the ith panel lt E gt 3373072 gt Summed over tail energies partial states for Nil spdf TOS 2261 2236 7 846 2214E 01 spdf MAG 2004E 02 2576E 03 1 762 3051E 04 38 spdf up TOS 1140 1117 4 804 1105E 01 spdf dn TOS 1120 1119 3 042 1109E 01 spdf up DOS 0000E 00 OOOOE 00 OOOOE 00 0000E 00 spdf dn DOS OOOOE 00 OOOOE 00 OOOOE 00 0000E 00 EEE kk kkk kk BAND ENERGY 35 824906474871 CORE ENERGY 3584 3342723301 POTENTIAL ENERGY 10079 778848994 KINETIC ENERGY 6459 6196701892 COULOMB ENERGY 12559 261981630 EXCH CORR ENERGY 276 39922038385 HUBBARD ENERGY 115 65537813379 TOTAL ENERGY 6491 6969099585 EEEo ooo kk k k kkk kk kkk kk XK XA XA XX
25. 34 0 0000269 0 0000000 0 0000000 3z2 1 0 0031954 0 0031954 0 0063909 0 0000000 0 3862361 3z2 1 0 0000044 0 0000044 0 0000089 0 0000000 0 0000000 REAL UP ZX xy x2 y2 3z2 1 IMAG UP yz ZX xy x2 y2 3z2 1 REAL DN yz ZX xy x2 y2 3z2 1 IMAG DN yz ZX xy x2 y2 3z2 1 oooo O OOO ooo o yz 9740251 0005125 0005125 0055350 0031956 yz 0000000 0000000 0000000 0000077 0000044 yz 9814365 0001126 0001126 0003418 0001973 yz 0000000 0000000 0000000 0000233 0000134 ooo l o ooo C e E N e e ZX 0005125 9740251 0005125 0055350 0031956 ZX 0000000 0000000 0000000 0000077 0000044 ZX 0001126 9814365 0001126 0003418 0001973 ZX 0000000 0000000 0000000 0000233 0000134 oo0ooo O OGO OG xy 0005125 0005125 9740251 0000000 0063912 xy 0000000 0000000 0000000 0000000 0000089 xy 0001126 0001126 9814365 0000000 0003947 Xy 0000000 0000000 0000000 0000000 0000269 l o ooo oooo D O O x2 y2 0055350 0055350 0000000 3862469 0000000 x2 y2 0000077 0000077 0000000 0000000 0000000 x2 y2 0003418 0003418 0000000 9468114 0000000 x2 y2 0000233 0000233 0000000 0000000 0000000 3z2 1 0 0031956 0 0031956 0 0063912 0 0000000 0 3862469 3z2 1 0 000004
26. 4 0 0000000 0 0000000 yz ZX xy x2 y2 3z2 1 IMAG UP yz ZX xy x2 y2 3z2 1 REAL DN yz ZX xy x2 y2 3z2 1 IMAG DN 2 TABLE OF State 3d for Nil spin up dn occupation numbers yz 0 9814361 0 0001127 0 0001127 0 0003406 0001967 yz 0000000 0000000 0000000 0000233 0 0000134 yz 9740245 0005125 0005125 0055347 0 0031954 yz 0000000 0000000 0000000 0000077 0 0000044 o D OO O oooo O O OGO l o ooo ooooo ooo ooo 0 o OCCUPATION NUMBERS FOR LDA ZX 0001127 9814361 0001127 0003406 0001967 ZX 0000000 0000000 0000000 0000233 0000134 ZX 0005125 9740245 0005125 0055347 0031954 ZX 0000000 0000000 0000000 0000077 0000044 oo0ooo oo0ooo oooo xy 0001127 0001127 9814361 0000000 0003933 xy 0000000 0000000 0000000 0000000 0000269 xy 0005125 0005125 9740245 0000000 0063909 xy 0000000 0000000 0000000 0000000 O 0000089 0 oooo o oooo ooo x2 y2 0003406 0003406 0000000 9468045 0000000 x2 y2 0000233 0000233 0000000 0000000 0000000 x2 y2 0055347 0055347 0000000 3862361 0000000 x2 y2 0000077 0000077 0000000 0000000 0000000 State 3d for Ni2 spin up dn occupation numbers 52 are 3z2 1 0 0001967 0 0001967 0 0003933 0 0000000 0 9468045 3z2 1 0 0000134 0 00001
27. 4 0 0000044 0 0000089 0 0000000 0 0000000 3z2 1 0 0001973 0 0001973 0 0003947 0 0000000 0 9468114 3z2 1 0 0000134 0 0000134 0 0000269 0 0000000 0 0000000 REAL UP yz ZX xy x2 y2 3z2 1 IMAG UP yz ZX xy x2 y2 3z2 1 REAL DN yz ZX xy x2 y2 3z2 1 IMAG DN CONSTRAINED PART 3 oo0oooOo oo0ooo oo0ooo G T ET yz 0000000 0000000 0000000 0000000 0000000 yz 0000000 0000000 0000000 0000000 0000000 yz 0000000 0000000 0000000 0000000 0000000 yz 0000000 0000000 0000000 0000000 0000000 DO O O G G O O O O DOGO OOG O G O O GO O ZX 0000000 0000000 0000000 0000000 0000000 ZX 0000000 0000000 0000000 0000000 0000000 ZX 0000000 0000000 0000000 0000000 0000000 ZX 0000000 0000000 0000000 0000000 0000000 OF THE LDA POTENTIAL State 3d for Nil spin up dn Hubbard potentials ODO OGOGO Ooo0ooo O OOGO e a a e O xy 0000000 0000000 0000000 0000000 0000000 xy 0000000 0000000 0000000 0000000 0000000 xy 0000000 0000000 0000000 0000000 0000000 xy 0000000 0000000 0000000 0000000 0000000 O OD QOO Oo oo0ooo D O OOO ooooo x2 y2 0000000 0000000 0000000 0000000 0000000 x2 y2 0000000 0000000 0000000 0000000 0000000 x2 y2 0000000 0000000 0000000
28. F key connected with the adjustment of the radial wave functions This substitutes previously used key NOVR in the INIFILE Also a bug in calculating exchange correlation energy for IXC 4 Vosko et al parametrization of LSDA functional has been found It is not expected to influence much on the calculated results The bug is corrected in following versions ASA1 60 CEL3 80 PLW2 50 Generalized gradient corrections after Perdew et al added and tested Two forms are avaiable here GGA91 and GGAQ6 45 11 USING LOCAL LIBRARY Programs in the local libraries nmt lib are designed for helping to construct input data for the main code and to read output files from the main code Usually they understand input output files only from the main code located in the same global subdirectory e g nmtplw lib serves NMTPLW code etc In contrast to that there exists a global library NMTLIB which contains different application programs serving any of NMT packages See chapter USING NMTLIB LIBRARY for the detailed description 11 1 Program CELL Program CELL is helpful to find different sphere radii before running the main code Ecept ASA code where the atomic sphere radii are completely determined by setting the weights for every atom see the description of parameter weight in the INTFILE the mt sphere and circumscribed sphere must be carefully chosen before running the main program Program is located in nmtplw lib cell and nmtcel lib cell director
29. FILE must be either recalculated or you must specify icon 0 when using the switch bnd 1 The bands can be plotted using the program plobands exe located in nmtlib plot See chapter USING NMTLIB LIBRARY for details Setting ibnd 2 is another option which allows to store the energy bands in the tetrahedron mesh of the BZ Since the bands are stored at the end of the self consistent cycle keep it in mind and set of iterations to 1 manually if necessary This storage can be necessary to plot the Fermi surface or to calculate real imaginary parts of susceptibilites See chapter USING NMTLIB LIBRARY for details e ipot lt potfile gt Specify ipot 1 or 2 to store the potential into the POTFILE The potential can be plotted afterwards using the program located at nmt lib grid See chapter USING LOCAL LIBRARY for details e ifat lt fatfile gt Specify ifat 1 together with the switch ibnd 1 to store the information about the partial orbital character of the bands This is extremily usefull to understand the chemistry of the compound The fat bands can be plotted with help of the tight bindindg LMTO ASA programs developed in Stuttgart FATFILE must be rewritten into the format understandable by the TB LMTO using the program nmtlib plot fat exe See chapter USING NMTLIB LIBRARY for details e idos lt dosfile gt Specify idos 1 to calculate the density of states Here a starting DOSFILE should first be prepared containing one line with the lower
30. FULL POTENTIAL LMTO PROGRAMS NMT USER s MANUAL S Yu SAVRASOV Max Planck Institute fuer Festkoerperforschung D 70569 Stuttgart Germany Department of Physics and Astronomy Rutgers University Piscataway NJ 08854 October 13 2000 Contents 1 INTRODUCTION 2 INSTALLATION 3 MAIN CONTROL FILE INIFILE 3 1 Control Parameters 2 2 e 3 2 Exchange correlation functional 2 ee 3 3 Iterative Procedures Limits and Accuracies 0 00 000 pe eee 34 ATOMIC Data A Pee A pees 3 5 Output Control Parameters e 3 6 Other Data for NMT pack e 4 STRUCTURE CONTROL FILE STRFILE 41 Main Parameters 2 ouies aeea araa ee 42 Primitive Translations circa he ae ee ek ay as EO ee a 4 3 Basis in original Unit Cell ee es 4 4 Basis in distorted Unit Cell 2 a a E a E A e a E A E R G AH E aE ah E O 4 6 Point Group Description E E TR a R N ee 4 7 Primitive Translations for BZ 2 ee N 4 8 High symmetry direction settings oaoa ee 4 9 Settings for PLOBANDS program 0 200000 ee eee 5 INPUT CHARGE DENSITY FILE SCFFILE 6 RUNNING THE PROGRAM 10 10 11 18 20 23 24 25 25 25 26 26 27 27 27 28 29 7 OUTPUT MESSAGE FILE OUTFILE 7 1 7 2 7 3 7 4 7 5 7 6 7 7 7 8 7 9 7 10 7 11 Reading input data deneon es doetor te a ha Pe ee a A Preparing structure constants 2 0 e Computine sphere Padil ies ee as ni a
31. KAPMAX is the maximum value for the number of kappa s which can be used in a single channel parameter nkap in INIFILE cannot exceed this value e NDIMMAX is the maximum value for the dimension of the hamiltonian without taking into account the number of spins in case spin orbit coupled calculation The H matrix is dimensioned as NDIM MAX NSPINMAX inside the program An upper limit for NDIMMAX is obviously NKAPMAX NPAT LMAXB 1 2 However chosing different Imax b for different atoms and or skipping some orbitals from the basis can make NDIMMAX much smaller e NPANMAX is the maximum value for the number of panels valence plus all semicore allowed to consider e NRADMAX is the maximum value for the number of radial points at the logarithmic scale If the actual value exceeds this number the program will use NRADMAX e KMAX is the maximum number of k points allowed to generate inside the irreducible Brillouin zone 8 2 Parameters for plane waves This set of parameters is only effective within PLW package e NPLWMAX is the maximum number of plane waves allowed for the Fourier transform of the charge density and the potential in the interstitial region If the actual number exceeds NPLWMAX the program terminates e NFFTMAX is the maximum number for the FFT grid evaluated If the actual number evaluated as m1 1 m2 1 m3 1 see input FFT grid in the INIFILE exceeds NFFTMAX the program termi nates e NBASMAX is the ma
32. TER FILE PARAM DAT for the detailed description The file PARAM DAT is included into every source code during the compilation time by INCLUDE statement The second comment concerns a scratch file storage To minimize core memory some data during the run are temporary stored in the scratch files To be able to do this a scratch directory must exist on any particular node where execution of the program is performed To create executable file go to the directory nmt run 1 Edit PARAM DAT and install the necessary size of arrays Sample PARAM DAT file is provided 2 Edit the file setup f and specify the path to the scratch directory Also check that other items match your computer settings 3 Edit the file timel f and specify the call to the system subroutine to learn CPU time 4 Compile all programs link them to get exectable file main exe Under UNIX using AIX XL Fortran Compiler this looks like x1f cOw f to compile only with optimization and suppress ing all warning messages The command xlf cCg f will compile only suppress optimizationm and provide debugging information To link use the command xlf o o main exe To create a load map use the command xlf o o main exe bloadmap map At the end of the map file a total ammount of the core memory allocated by the program is printed out All NMT programs have three main input files e INIFILE main control data file which is the same for all NMT codes e SCFFILE i
33. X Y Z except Z axe e Z rotation along Z e I inversion e C combination 24 total of elements in the Hexagonal group U operation equivalent 1 Z operation rotations along Z 2 element pi 3 Z operation 3 62 Z operation Z operation Z operation A operation pi 1 0 0 A operation rotations along arbitrary axe 9 pi sqrt 3 2 1 2 0 A operation 8 pi 1 2 sqrt 3 2 0 A operation 7 pi 0 1 0 A operation 12 pi 1 2 sqrt 3 2 0 A operation 11 pi sqrt 3 2 1 2 0 I operation inversion 13 C operation combinations 14 13 2 C operation 15 13 3 C operation 16 13 4 C operation 17 13 5 C operation 18 13 6 63 19 13 20 13 21 13 22 13 23 13 24 13 10 11 12 operation operation operation operation operation operation 64 16 Acknowledgents I greatly acknowledge Dr Andrej Postnikov who has initiated writting of this manual Part of the developments has been done in collaboration with my brother Dr Dmitrij Savrasov Special thanks to Prof Ole Andersen and Dr Ove Jepsen who are my LMTO teachers 17 COPYRIGHT These programs are a free software for scientific and or educational purposes It is not allowed to redistribute them without prior written consent of the Copyright owners It is illegal to commercially distribute these programs as a whole or incorporate any part of it in
34. XX XX XX Xk XK XA XA XX XX XX XX 7 7 10 Evaluating forces The evaluation of forces program FORCES see source file forces f is the next step at the iteration Note that the Hellmann Feynmann forces are not accurate and large incomplete basis set corrections must be taken into account See the description of the input parameter npfr in the INIFILE CALCULATED FORCES AT THE CENTERS OF ATOMS gt Position gt 00000E 00 00000E 00 00000E 00 for Nil HF FORCE Fx 0000000E 00 Fy 0000000E 00 Fz 0000000E 00 CR FORCE Fx 0000000E 00 Fy 0000000E 00 Fz 0000000E 00 HF CR FORCE Fx 0000000E 00 Fy 0000000E 00 Fz 0000000E 00 XK XA XX XX X 7 11 Mixing charge densities The last step at the iteration is mixing of the input charge density and the output charge density to prepare the input density to the next interation The self consistency of the charge density can be simply watched by comparing the input output charges of the spheres at the iteration Values S inp and S out The same is done for magnetization M inp and M out Values I inp and I out stand for the interstial charges After Broyden mixing procedure the ne charge density and magnetization are constructed the corresponding charges and magnetic moments within spheres are also printed out A usefull parameter for watching whether the Broyden mixing is properly working is the iteration weight At the begining this number is set to one If
35. ap energies E k complex numbers If Re x gt 0 then Im x must be non zero and close to 0 03 Ry to avoid singularities in the Ewald summations Then sets of data for all non equivalent sorts follow Each set includes 14 for Nit 2 O 28 D0 12 D0 1 1 0 0 50 58 7 Zz zcor lr icor ispl split mass 2 100 3 500 1 1 0 5 mt sphere rou sphere weight rloc 626 lmax t 1lmax b 1lmax v e for Nil title for every atom Note that this character string maximum 10 letters will be read and widely used in the output file Therefore it is recommended to use the format for EL e z atomic number if empty sphere specify z 0 e zcor deep core charge i e those atomic states which will be treated as levels not as bands The program can treat three kinds of states i Valence states i e those which form the bands These states form main valence panel and will be found by diagonalizing LMTO hamiltonian ii Semicore states i e states which have small but negligible band width These states have small hybridization with the valence states and therefore are treated in separate energy panels They also found by diagonalizing LMTO hamiltonian corresponding to each of the semicore state iii Deep core states which are found by solving Dirac s equation for free atom with the potential taken as the spherical part of the crystalline potential until the MT sphere and zero outside it A simple rule to sort out the states over these three
36. ase of restart calculation they will be read from it to obtain a smooth continuation 0 5 0 5 0 5 0 5 Eny 1 0 2 0 3 0 4 0 Dny It should be noted that in the Eny line Imazx b 2 numbers are read For the states which are not included in basis but used in the decomposition of the tails coming from other atoms the Imax b 2 th value in the Eny line is used as the energy of radial functions with higher l s in terms of which the decomposition is performed The prescription of chosing this value is to place it into the center of 18 gravity of the whole occupied part of the band This parameter is not crutial for the calculation a good estimate for it is the middle between bottom of the band and the Fermi energy General prescription for choosing tail energies for the programs working with the MT geometry CEL PLW the interstitial region can be large therefore an additional variational freedom of the basis functions is desired For the states forming broad energy bands 2 or 3 kappa basis set must be chosen to make sure that the result of the band structure calculation is well convergent The values for these kappa s are not that important the only condition is that they should be separated from each other by the energy of the order 1 Ry to avoid linear dependency of the LMTOs A few hints can be given here i negative kappa basis set is formed with x12 0 1 Ry k22 1 0 Ry and 637 2 5 Ry The advantage of the negative ene
37. bic and hexagonal symmteres are understandable automatically if the symmetry code is set to either C or H see the description of the symmetry code in the chapter describing STRFILE However the same effect can be reached if the symmetry code is set to A and one of the KOVFILEs shown below is created You can make your own KOVFILE using these examples KOVFILE for describing the cubic rotational system is listed below All the expressions must be recognizable by the CALC facility see the description of the calculator in the chapter describing STRFILE Cubic rotations are given after Kovalev Note the difference between the present and Kovalev numbering which is given by a second column Notations are the following e U equivalent operation e A arbitrary rotation along X Y Z except Z axe Z rotation along Z e I inversion e C combination 48 total of operations in the Cubic group operation equivalent operation rotations along arbitrary axe 20 element pi 2 1 0 0 A operation 2 pi 1 0 0 A operation 19 3 pi 2 1 0 0 A operation 22 pi 2 0 1 0 A operation 3 pi 0 1 0 A operation 24 3 pi 2 0 1 0 Z operation rotations along Z axe 15 pi 2 Z operation 59 4 pi Z Operation 14 3 pi 2 A operation 5 2 pi 3 1 1 1 A operation 9 4 pi 3 1 1 1 A operation 10 2 pi 3 1 1 1 A operation 8 4 pi 3 1 1 1 A ope
38. bove the Fermi energy nff may be set smaller than the true number of filled bands One can always use combination nff 0 and nef total of bands This however may lead to unnecessary large storage of the expansion coefficients AKA which are written on the scratch file for the bands crossing the Fermi level pole this value is read but not used by the NMT programs n1 n2 n3 nc divisions of the Brillouin zone along three directions for the tetrahedron inte gration nc sets first division for semicore states other two will be determined automatically Every k point is described by the set of three integers k1 k2 k3 according to k k k YE O 3 n na n3 k where G1 G2 and Gg are the reciprocal lattice vectors For calculating the bands along high symmetry directions switch ibnd 1 described above the first value of n1 will control the division of every direction the number of k points generated in every line is n1 1 mi m2 m3 divisions of the unit cell for the fast Fourier transform Every r point of the FFT grid is described by i1 i2 i3 according to r _R R Rs 4 m1 ma m3 where R1 R2 and Rg are the primitive lattice itranslations The rule of thumb is 16 divisions between nearest neighbours guarantees the sufficient accuracy Another estimate useful for complex structures is total number of divisions m1 x m2 x m3 should be not less than 4000 x the number of atoms then distribute the division
39. ched Usually convergency of the total energy up to 0 1 mRy corresponds to the charge density self consistency of the order 102 Charge Density Self Consistency 1047040E 05 1495354E 06 Magnetization Self Consistency 9249301E 06 1891719E 09 40 8 PARAMETER FILE PARAM DAT The dimensions inside the program are defined using PARAMETER statements All parameter statments are collected in one file called PARAM DAT and are included in every program using INCLUDE statement An example of PARAM DAT is given below C eo MAIN PARAMETERS OF PARAMETER NSPINMAX 2 PARAMETER NPAT 7 PARAMETER NSORTMAX 5 7 PARAMETER NSYMMAX 50_ PARAMETER LMAXV 6 PARAMETER LMAXB 3 PARAMETER LMAXT 6 PARAMETER NKAPMAX 4 PARAMETER NDIMMAX 150 PARAMETER NPANMAX 5 7 PARAMETER NRADMAX 450 PARAMETER KMAX 400 THE ARRAYS eee RK SPIN POLARIZED VERSION WITH SPIN ORBIT COUP MAX NUMBER OF ATOMS IN UNIT CELL MAX NUMBER OF DIFFERENT TYPE ATOMS MAX NUMBER OF NON ZERO COMPONENTS IN RHO V LMAX FOR THE POTENTIAL AND CHARGE DENSITY LMAX FOR BASIS SET LMAX FOR TRIAL AND WAVE FUNCTION EXPANSION MAX NUMBER OF TAILS IN TRIAL FUNCTION MAX DIMENSION OF HAMILTONIAN MAX NUMBER OF PANELS MAX NUMBER OF POINTS IN RADIAL MESH MAX NUMBER OF K POINTS Cerebro PLANE WAVE DATA PARAMETERS KKK KKK PARAMETER NPLWMAX 40000 MAX NUMBER OF PLANE WAVES IN FOURIER TRANSF PARAMETER NFFTMAX 135000 FAST FOURIER TRANSFORM GRID
40. d to mixing parameter migs in the linear mixing scheme It was found that mixb cannot be small and it is usually of the order 0 3 0 4 e mizh this is linear minxing parameter for higher l components l gt lbr of the charge density Since it is assumed that these components do not influence much the self consistence loop they are mixed within linear mixing scheme and do not stored for all previous iterations 3 4 Atomic Data Atomic Data 4 3 2 1 natom nsort nspin norbs 8 0510 1 00 lattice parameter v v0 1 2 3 3 is iatom 3 0 6 0 0 0 4 0 03 1 4 0 03 nkap ekap in valence states e natom total number of atoms per unit cell e nsort the number of non equivalent atoms 13 e nspin 1 for non spin polarized calculations 2 for spin polarized calculations e norbs 1 without spin orbit coupling 2 incuding effects of spin orbit coupling This works only in combination with the switch nspin 2 In this case the size of the hamiltonian is doubled Note that in the PARAM DAT file PARAMETER NDIMMAX the size of the hamiltonian need not to be increased Inside the program the H matrix is dimensioned as HMAT NDIMMAX NSPINMAX NDIMMAX NSPINMAX Note also that you cannot specify nspin 1 and norbs 2 i e spin orbit coupling and NO spin polarization If spin orbit coupling is ON then spin polarization is always assumed In case non magnetic calculation is required like Pb for example do not specify an in
41. does not change the accuracy of the lattice sums but may accelerate the calculation of those Parameter a 1 was chosen from the condition of fastest calculation for system with the number of atoms of order 10 at the work station IBM RISC 6000 Primitive Translations First three Cartesian components of the first vector then the second etc as in subr READSTR 4 3 do Ivec 1 3 READ 1 RBASR I Ivec I 1 3 enddo Basis in original Unit Cell The same as above In fact this data are not currently in use but may be used for programmer s information 4 4 Basis in distorted Unit Cell These lines are really used to set the basis Note that the atomic positions may be provided either in Cartesian coordinates for ibas 1 or in the units of primitive translations for ibas 0 In the output file true Cartesian coordinates are printed out after all transformations 27 4 5 Strain matrix This matrix performs the linear transformation of the translation vectors and the basis positions according to R new S R old For reciprocal lattice the rule is G new G old S 1 This may be convenient for introducing strain or rotating the whole setup to other axes This matrix should always be provided in the default case it is simply the unit matrix The matrix is read in row by row as do I 1 3 READ 1 STRAIN I J J 1 3 enddo In restoring the true translation vectors and basis positions the program first appl
42. e gt nio bnd ibnd lt bndfile gt gt nio pot ipot lt potfile gt nio fat ifat lt fatfile gt gt nio dos idos lt dosfile gt nio scf iscf lt scffile gt nio out iout lt outfile gt This section explicitly gives the names of files which may be used by the program Generally but with a few exceptions the switch in the first position has the following meaning 0 file is not created or created as temporary 1 file will be opened as a new one and saved if file exists the execution will be terminated 2 file already exists and will be read if file does not exist the switch will be automatically set to 1 6 the contents of the file will be printed out to the terminal channel 6 icon lt confile gt file for storage the structure constants Operative with all NMT programs If icon 0 CONFILE will be created as temporary in the scratch directory the structure constants will be calculated for every k point and at every iteration This might save some disc space since no storage of the structure constants for all k points is performed If icon 1 CONFILE will be created as permanent and the structure constants will be calculated and stored in this file for all k points If CONFILE exist the program will terminate Setting icon 2 is more general option If CONFILE does not exist it will be created the program will automatically set switch icon 1 if CONFILE already exist the stored informa
43. e of the multiple kappa basis is not important single kappa basis is always OK The tail energy can be fixed to 0 or slightly smaller value 0 1 Ry to avoid structure consnant singularity as was proposed in the original paper 1 After the blocks distributing the valence states over different x s follows the description of the semicore panels semicore states are 1 of semicore states 3 1 4 0 0 0 n 1 energy for 3p state e 1 number of semicore states which will be treated as band states in separate energy windows without hybridizing with the main valence panel e 31 4 0 0 0 principal quantum n angular momentum I tail energy K here 3p state for every state This is self explanatory and it should be only noted that a possibility is provided that semicore states from different atoms belong to the same panel and are therefore allowed to hybridize for this one should indicate the same tail energy for them Semicore states with different tail energies are 19 treated as independent During the iterations E s for the semicore states are chosen to be the bound energies for the spherical part of the potential for a given iteration Here the block Atomic Data is finished The last block is 3 5 Output Control Parameters Output Control Parameters 2 NNOOOOOOO nio con icon lt confile gt nio sr f isrf lt srffile gt nio psi ipsi lt psifile gt nio scr itmp lt tmpfil
44. e potential sites producing the polyhedron planes The expirience shows that in most cases it is sufficient to use ndivi 2 If however the program prints incorrect value of defined cell volume set ndivl 1 The best way indeed is to try to find different atomic weights which define the polyhedron geometry This parameter has only meaning for the CEL version not for ASA or plane waves versions ndiv2 cell surface grid parameter for taking interstitial integrals Increasing this parameter makes the calculation of the interstitial cell integrals more accurate but also more time consiming The recommended value is 20 This parameter has only meaning for the CEL version not for ASA or plane waves versions nvec average number of lattice vectors used in the Ewald summation The program will take the number of vectors in reciprocal space approximately twice of nvec and the number of vectors in direct space half of nvec see also next paragraph The program does not check the convergency of lattice sums but prints out the accuracy of the calculation see below which is the relative cotribution to the sum going from the largest vectors a is usually equal to 1 Since the ratio between numbers of generated vectors in reciprocal and direct spaces is fixed by four see previous paragraph the possibility is provided to scale this ratio by a factor of a For example if a 1 4 the number of generated vectors in both spaces will be the same This scaling
45. e the sphere radii the calculation may be continued from the previous SCFFILE All needed interpolations to another meshes are done automatically inside the program Specifically for the CEL package If one changes the sphere radii the surface structure constants SRFFILE should be recalculated if they were stored before 31 7 OUTPUT MESSAGE FILE OUTFILE Here description of the output messages is given Also short introduction to the structure of the program is described see Figure 2 Consider one particular iteration for NiO system by the program NMTPLW The structure and the output messages of NMTASA and NMTCEL are similar 7 1 Reading input data The execution of the NMTPLW package source file main f starts from reading the input data controled by INIT subroutine see file init f Beginning of the OUTFILE contains the information read from the INIFILE lt lt lt INPUT INIFILE READ gt gt gt Band Structure Calculation of NiO lt CONTROL PARAMETERS gt lift 3 Self consistency is on lmto 1 Unscreened LMTO is on novr 2 Overlap matrix via Sdot npf 0 Atomic forces are off lt EXCHANGE CORRELATION gt ixc 3 Janak Williams Morruzi no GGA lt CONVERGENCY DATA gt niter 50 of iterations epstot 1000000E 08 total energy convergency epsrho 1000000E 09 charge density convergency epsmag 1000000E 09 magnetization convergency lbroy 2 Broyden mixing is switched
46. f substitutes previous parameter nint The latter was used to control the calculation of interstitial overlap integrals All current versions of the programs calculate overlap matrices using energy derivative of the structure constants nint 2 is set in the program setup f as a default value The new parameter lrwfis used to control adjustment of the radial wave functions to the spherical part of the potential Normally the radial Schr dinger s equation is solved with the spherical potential at the current iteration In case of spin polarized calculation the equation is solved for both spin up and spin down potential and therefore radial wave functions have a spin dependence There are however special cases when it becomes useful to solve radial Schr dinger s equation not with the spin dependent potential but with the average potential Vup Vdn 2 This eliminates explicit dependence of the radial wave functions from the spin index It is necessary for example when calculating susceptibility functions using linear response theory Another option is provided to froze radial wave functions for one particular spherical potential and do not recalculate them at every iteration of the self consistency If radial wave functions are frozen then the calcuation of forces is exact in the sence that the calcuated force is exact derivative of the LMTO expression of the total energy without any further assumptions The parameter lrwf can take one of the fol
47. for NMTCEL program Setting lift 1 is useful if the structural information is necessary for a number of jobs which will be executed spontaneoulsly 2 Calculate structural data files and make one calculation of the energy bands This is useful if energy bands are necessary to calculate and store in one of the output files No BZ integration is performed and the new charge density is not constructed 3 The whole self consistent looping is switched on This normally includes ppreparation of structural data files calculation of energy bands integration over BZ charge density construction evaluation of total energy forces and preparation of the new charge density for the next iteration e Imto 10 1 Unscreened original LMTO All NMT programs use unscreened long range LMTO representation as originally formulated in Ref 1 2 Screened tight binding LMTO is under construction It is avaiable in its simplest version in the program FTBASA not automatically distributed but can be obtained upon request Other versions FTBCEL and FTBPLW are not currently avaiable e ncou is the parameter which is only used by the NMTCEL program ncou 1 switches on the Coulomb corrections to the full potential This frequently improves the accuracy of the calcula tion in case of open lattices and supercells by solving more accurately the Poisson equation in the interstitial region The parameter has no effect on NMTASA and NMTPLW codes e lrw
48. g 1 basis vectors are given in Cartesian system 0 basis vectors are given in units of primitive translations switch for Brillouin zone construction 1 Brillouin zone translations are set up automatically as reciprocal lattice translations 0 BZ translations are read from the corresponding lines in the STRFILE The idea is it may be useful if the automatic Brillouin zone has too pathological shape for dividing into tetrahedra switch for using the calculator to translate the expressions icalc 1 the calculator is on for interpreting the expressions in the sections Primitive Trans lations Basis Strain Matrix Brillouin zone High symmetry settings Every line can contain any simple expressions brackets are allowed without restrictions special functions like COS SIN TAN EXP LOG SQRT CBRT X 1 3 are allowed in the format of FORTRAN but nesting of the special functions is not allowed Special constant PI 3 1415 can be specified Degrees like or are also allowed but the expressions in brackets cannot contain special functions listed above Expressions are separated by commas and at the end any comment only after is allowed No letter case sensitivity is assumed Look also at some comments in in the nmt run calc f file Some simple examples are Primitive Translations for the hexagonal lattice 1 0 0 0 0 0 Ax Ay Az 11 lt note for the c
49. g Ske doe EO Ea EE ee AR Finding full potentials Pa 2 ig te ee e a ee Ge Pa ee ea Calculating energy bands 2 2 a Integrating over Brillouin zone e Constructing the charge density ee Renormalizing core levels ee Evaluating total energy ee Evaluating forces so 24 2 ose en A a Pee eR ee ge aes Mixing charge densities ooa 8 PARAMETER FILE PARAM DAT 8 1 8 2 8 3 8 4 8 5 Main parameters cio Bes ew Ae Rad ate ee ee ee Re ae ee Parameters for plane waves aoaaa a Parameters tor cells 2 24 20 e 008 eror rt Re ee AR ee ee Frequently used settings 2 2 5 sn asa a A ee ee Estimation of the needed core memory 9 ERROR MESSAGES 9 1 9 2 9 3 9 4 Errors connected with PARAM DAT e Errors connected with input 2 e Errors connected with iterative procedures 2 2 ee ee Other errors 5 E ke A oe Re oe ek Sos YRS 10 VERSIONS DIFFERENCES 11 USING LOCAL LIBRARY 11 1 11 2 11 3 11 4 Program CELL gcc oe eta ag e a HR eA ke ee Ree be Eel AO Sa OE See Car es Program GRID rair 8 ee A Eek Bo hee be Ber BE ok Pee e Rad Program SOEN wy hea AAA A A eB th Be ads Program SCEM Lois ee Pe ae ee a a ee Bagh ee Bae ale 12 USING NMTLIB LIBRARY 12 1 12 2 12 3 12 4 12 5 12 6 12 7 Programm ATOM Er shies pag Sat AAA PELE AOS OU ey OS bee eS AA Program CHI Qs ic igo wos Se AAA ee BEEN AAA eS Program MBSE A 8 asa ae Rae eo hee
50. ian GRAFOR package is located here e program plobands exe is designed to plot energy bands along high symmetry directions To create an executable module compile the files plobands f calc f Also compile contents of the subdirectory GRAFOR Link these files to get executable plobands exe 49 The input to this program is BNDFILE and STRFILE The BNDFILE must be created by any of the NMT packages with the key bnd 1 The end of the STRFILE should contain a list of high symmetry directions to plot For example the following control line X g L Z R A will produce two joined pictures with the bands along X T and T L directions and sperately two joined pictures with the bands along Z R and R A directions Small letter g is understood as greek I The dot at the end of the control line is understood as the end of the control line The comma is understood as making free space between two plots Any of the directions X g g L etc must be calculated and stored in the BNDFILE as explained in sections describing INIFILE and STRFILE The output from the program has a format understandable by the SURFER GRAPHER software It can be directly plotted using view exe plot exe utilities of this software Alternatively it can be rewritten into the postscript format using the program plotps exe see below e program plotdos exe is designed to plot partial densities of states To create an executable module com
51. ibroy 1 after n ibroy iterations admixb 3000000 initial guess for Jacobian admixh 3000000 mixing for high 1 gt 1broy lt ATOMIC DATA gt l natom 4 of atoms in unit cell nsort 3 of atoms of different type par 8 051000 lattice parameter in a u nspin 2 including spin polarization norbs 1 without spin orbit coupling nkap 3 of kappas in valence panel Etaili 60000 Hankel tail energy Etail2 40000 Hankel tail energy Etail3 1 4000 Hankel tail energy lt OUTPUT DATA gt icon 1 save str const in nio con iftr 0 no storage fourier con ibnd 0 no storage of bands idos 0 no d o s calculation ipot 0 no storage of full potential iscf 1 save full density in nio scf iout gt 1 print current output lt OTHER DATA gt nff 14 of bands below EF 32 nef 4 of bands crossing EF ndiv 8 8 8 tetr mesh for valence panel ndic 2 2 2 tetr mesh for semicore panels lt ADDITIONAL INPUT FILES gt ichub gt 0 with Hubbard correction after nio hub ichop 0 with no hopping matrix icopt 0 with no optical properties This information is printed only to test the corectness of the input data The next output lines contain the information read from the STRFILE lt lt lt INPUT STRFILE READ gt gt gt xxkxxkk Structure Data for NiO lt CONTROL PARAMETERS gt natom 4 of atoms in unit cell b a 1 0000 orthorombicity
52. ies No specific configuring of the source files is necessary Just compile them link and obtain make exe Program CELL reads INIFILE and STRFILE and prints information about the calcuated sphere radii The basic input used for generating the spheres is again atomic weights as it has been explained above Example of the output lines contains input mt sphere is 2 100000 input as sphere is 2 610231 cell mt sphere is 2 108595 cell as sphere is 2 605776 touch mt sphere is 2 108595 nrad 318 znuc 28 zcor 18 circums sphere is 3 486185 npln 18 The input mt sphere is the one which has been read from the INIFILE The cell mt sphere touch mt sphere and circums sphere are those which have been generated according to the input weights Specifically both cell mt sphere and the circumscribed sphere come as a result of generating the atomic polyhedra while touching spheres are just blowed up proportionally to the weights until they do not overlap See also chapter OUTPUT MESSAGE FILE OUTFILE for more information An important comment should be said when moving the atoms is required Then to avoid numerical inaccuracy it is highly advised to use the same mt spheres circumscribed spheres for all atomic geometries To do it one first generates STRFILEs for all needed structures then fixes the weights in one INIFILE computes mt spheres circumscribed spheres at every geometry and then chooses the minimum values for the mt sphere radii maximum va
53. ies the orthorombic scaling DO 10 I 2 3 I 2 for b a I 3 for c a do IVEC 1 3 RBASR 1I IVEC RBASR I IVEC ORTH I direct lattice BBASR I IVEC BBASR I IVEC ORTH I recipr lattice enddo do IATOM 1 NATOM TAU R I IATOM TAU R I IATOM ORTH I basis positions enddo 10 CONTINUE and then the strain matrix DO IVEC 1 3 do I 1 3 RBAS I IVEC 0 DO do J 1 3 RBAS I IVEC RBAS I IVEC STRAIN I J RBASR J IVEC enddo enddo ENDDO C APPLY STRAIN FOR BASIS VECTORS T Sx T DO IATOM 1 NATOM do I 1 3 TAU I IATOM 0 DO do J 1 3 TAU I IATOM TAU I IATOM STRAIN I J TAUR J IATOM enddo enddo ENDDO and correspondingly for the reciprocal lattice 4 6 Point Group Description Symmetry code either C for cubic systems or H for hexagonal systems Cubic system contains 48 operations of a cube hexagonal system contains 24 operations of a hexagon it incudes rotations about 60 degrees along z axe Applying all symmetry operations for cubic or hexagonal group the program picks up those which are consistent with the actual crystal structure Non symmorfic operations are found as well Therefore your choice is only to decide whether a particular structure belongs to the cubic or hexagonal symmetry Note that since C or H rotational operations assume a certain coordinate system as rotation about 60 degress along z axe not 28 x or y axes the same coordinate system should be used to describe
54. iterative procedures A number of iterational procedures is programmed inside the NMT package in order to find the Fermi energy or the E values from a fixed set of D Limiting number of iterations and the accuracy is set in the file nmt run setup f If the number of iterations is exceeded here the message is printed and execution is terminated 9 4 Other errors Some warninngs and errors are connected with the lost of accuracy in solving differential equations or in integrational procedures Another type of errors can be due to not positevely defined overlap matrix which is most likely due to an error in the input The overlap matrix defined with the non overlapping MT spheres is always positevely defined When MT spheres overlap there is a warning message 44 10 VERSIONS DIFFERENCES This section traces differenes between current and previous versions of the program ASA1 10 CEL0 30 PLW2 01 or earlier not important ASA1 20 CEL0 41 PLW2 10 Broyden mixing added ASA1 30 CEL3 50 PLW2 20 LDA U added ASA1 40 CEL3 62 PLW2 30 Spin orbit coupling added ASA1 42 CEL3 62 PLW2 32 Relativistic solution of the semicore problem is rewritten no references to the Liberman atomic program exist anymore ASA1 50 CEL3 70 PLW2 40 Includes the possibilty to calculate hopping integrals for tight binding cal culations Does not actually work well ASA1 51 CEL38 71 PLW2 41 Marginal internal changes ASA1 52 CEL3 72 PLW2 42 Contains LRW
55. itial splitting of the potential see parameter split below If after self consistency is reached for non magnetic spin orbit coupled calculation the spin polarization is necessary specify some splitting and set ispl 2 see also below Do not forget to set ispl 0 after one run since ispl 2 will always split the potential at the begining of every run Notes to orbital magnetizm spin polarized spin orbit coupled calculation makes non zero average orbital moment The program calculates orbital contribution to the magnetic mo ment and prints it out However no contribution to the potential arizes from the orbital moment in LSDA Therefore the spin densities remain unchanged inside the program In all places where the magnetic moment is calculated and printed out it is SPIN magnetic mo ment WITHOUT orbital contribution The orbital contribution is printed out separately and must be added to the spin moment in order to obtain the total magnetic moment Notes to the group symmetry since spin orbit coupling operator lowers the symmetry of crystal group do not wonder if after switching SO coupling the crystal group will contain only 8 operations instead of 48 in the cubic case e lattice parameter in atomic units e v v uniform compression e is iatom for each atom from 1 to natom gives the sort of this atom The sequence of atoms as in the STRFILE see below e nkap number of different tail energies E K in the valence band e ek
56. kG Bb Boke ae De OS wok eee Ge Sas Program MOSQ 200 ii A ee eed eo eal eee BI Eth Bad Program PLO T 3 2 08 eE RTE we ra BE Ses era ak ap aw Ba eae Program QPNV adresi a auy Be Sos aloe A BR ee A Ds Bo ats Program SURE naci ra e do hts pe a a KN 30 30 32 33 34 35 35 36 36 36 37 37 39 39 40 41 41 41 42 42 42 42 42 43 44 44 44 45 45 13 ADDITIONAL INPUT HUBFILE 13 1 General Settings 2 4 0 2020 ada ha ee a a E a a 13 2 Description of Correlated States ee 13 3 Table of Occupation Numbers 14 ADDITIONAL INPUT HOPFILE 141 General settings ii A A O AA RTE R T T 14 2 Description of Hopping Integrals e 15 ADDITIONAL INPUT KOVFILE 16 Acknowledgents 17 COPYRIGHT 49 53 54 54 55 55 55 57 63 63 1 INTRODUCTION The full potential linear muffin tin orbital Ref 1 FP LMTO programs described here are de signed to perform band structure total energy and force calculations within the methods of density functional theory DFT Refs 2 3 4 Main features include i Local spin density approximation LSDA avaiable in many parametrizations together with the gradient corrected density functionals GGA91 amp GGA96 ii Multiple x LMTO basis sets and many panel technique iii Total energy and force calculations for determining the equillibrium structure and phonons iv LDA U method for strongly correlated systems v Spin orbi
57. lowing values 2 Adjust radial wave functions to spin dependent potential This is what is usually done and must be used in most cases 1 Adjsut radial wave functions to spin average part of the potential This eliminates the dependence of radial wave functions from the spin index It is necessary when calculating dynamical susceptibility functions using linear response theory and the program MAGPLW currently not avaiable but coming soon 0 Do not recalculate radial wave functions The spherical potential for which the radial wave functions will be constructed must be stored in the POTFILE see below This feature should bring complete correspondence between calculated total energies and forces if there is a trouble that the calculated forces are inaccurate In fact it is useful for debugging purposes only generally the force formulae programmed are sufficiently accurate e npfr 0 no accurate atomic force calculation The output will only contain the Hellmann Feuynman forces which are normally not accurate at all when using the LMTO method due to the large incomplete basis set or Pulay corrections 11 1 atomic forces including both the Hellmann Feynmann and Pulay contributions will be evaluated The accuracy of the forces due to unself consistency of the charge density can be controlled Since evaluation of the Pulay forces is computationally demanding switching this option is recommended after the self con
58. lues for circumscribed sphere radii This procedure allows to make sure that during atomic displacements the mt spheres will never overlap For the CEL program that also means that atomic polyhedra generated for every geometry will always lie within the fixed circumscribed spheres 11 2 Program GRID Program GRID of the NMTPLW package is helpul to plot the charge densities magnetization or the full potential along some line or within some plane It is located in nmtplw lib grid directory No special configuring of the program is required just compile all the files links them to get make exe executable module The program asks for three file names INIFILE SCFFILE and STRFILE Specify output charge density file as the SCFFILE if it is necessary to plot charge density or magnetization Also specify POTFILE see ipot lt potfile gt description in the chapter MAIN CONTROL FILE INIFILE to plot the full potential An additional input is required to say whether the plot along the line or within the plane is required Specify two 3D coordinates to define a desired line in the space or specify three 3D coordinates to define a desired plane in the space A number of divisions must be also given The output file LINE DAT for line plot and GRID DAT for plane plot is formatted and contains a few columes First one two columns describe the line plane grid points The last two columns define the values of the function charge density or potential and the
59. maginary part approximately 0 03 Ry must be placed to avoid this singularity Result from VECGEN for direct reciprocal spaces Accuracy 2947387E 12 Accuracy 4712605E 17 Accuracy 1545859E 19 Min energy for using Evald s method 2 262885 Total of connecting vectors found Minimum difference between k G 2 and kappal 2 Minimum difference between k G 2 and kappa2 2 Minimum difference between k G 2 and kappa3 2 Rmax Gmax Smax 2 616219 6 592308 4 473601 7 3 Computing sphere radii 7 gt of of of is is is vectors 163 vectors 609 vectors 759 Ry 9851246 1121901E 01 1749081E 02 After preparing the structure constants the execution of the program transfers to the package of programs for finding atomic polyhedra and different sphere radii This task is controlled by the program CELLS see source file cells f Next lines in the OUTFILE contain the information about accuracy in using the Fourier transforms for the pseudoLMTOs in the interstitial region The pseudoLMTOs are defined as linear combinations of the 35 Hankel functions in the interstitials but they are smooth inside the spheres In other words pseudoLMTOs are linear combinations of pseudoHankel functions The number of plane waves using for Fourier transform of every s p or d partial wave is different the higher 1 the more cutoff is required The ratios RH S H S and GH S RH S sh
60. me exists or not if it exists it will be replaced A usefull hint here is always to put iscf 2 and set the output SCFFILE to nio scf The input SCFFILE is nio scO If the execution is sucessful just rename nio scf to nio scO i e set new input SCFFILE to the output SCFFILE If the execution fails by some reason mistake in the input bad choice of the basis set ghost bands etc do not bother to erase bad nio scf which will be created after the run correct input data files and start the job again specifying nio scO as the input SCFFILE Switch iscf 2 will automatically erase bad nio scf file tout lt outfile gt This is the file for storing current output at the iteration printing total energy forces atomic charges magnetic moments and a lot of other usefull information If zout 0 this file will not be created actually created and deleted after the execution is completed If iout 1 the OUTFILE will be created and the output information stored In case the file with the same name already exists the program will terminate to avoid unnecessary rewritting If iout 2 the OUTFILE will be created regardless if the file with the same exists or not If the file exists it will be replaced by the new OUTFILE An example how it works is the following Suppose one makes 10 iterations The nio out file is created If another 10 iterations are required another nio out file must be created If sout 2 the old OUTFILE will be replaced by the new one
61. n by Relx q Cen e kjj TKI k qj The imaginary part is defined as follows Imlx q 5 er exj ler tay 6 kjj No special configuring of the program is required Just compile all the files link them to get executable make exe The input to the program are 4 files INIFILE STRFILE BNDFILE and PNTFILE BNDFILE should contain energy bands in the irreducible part of the Brillouin zone see the description of the input param eter ibnd 2 in chapter describing INIFILE PNTFILE contains a list of q points for which the susceptibility is calculated Use special program located in nmtlib qpnt directory to create PNTFILE After the input file names are given specify output file name then the following key 1 for the imaginary part 2 for the real part 3 for the imaginary part weighted by the velocity factor vx Uk qj The last input line should contain number of q points to calculate number of lines to read from PNTFILE the Fermi energy density of states just to renormailize Im x q and the frequency set zero in most cases The output file will contain susceptibility contributions from all band transitions crossing the Fermi energy The format of the output file coincides with the BNDFILE Employ it if it is necessary to plot susceptibilities with help of the band programs 12 3 Program MESH Program MESH can expand the data from the irreducible part of the BZ to the full BZ It is for example
62. nput charge density file Due to a different representation of the charge density used in ASA CEL and RUN this file is unique for every particular NMT code e STRFILE structure data file describing the crystalline lattice Examples of INIFILEs and STRFILES are provided in nmt dat 3 MAIN CONTROL FILE INIFILE The main control file of the full potential package has extension ini Below an example of band structure calculation for NiO will be considered The INIFILE for NiO can be named as nio ini or more shortly ini It will be located in nmt dat nio nio ini An example of this file is given below k Band Structure Calculation of NiO 3 lift 1 lmto 0 nlrt 2 lrwf 0 npfr Exchange Correlation 3 1 2 3 by VBarth H Gunn L Jan W Iterative Procedure Limits and Accuracy 50 0 2 0 2 1 D 9 2 1 0 3 0 3 niter mix mag eps lbr ibr mixb mixh Atomic Data 4 3 2 1 natom nsort nspin norbs 8 0510 1 0 lattice parameter v v0 1 2 3 3 is iatom 3 0 6 0 0 0 4 0 03 1 4 0 03 nkap ekap in valence states for Nil a o RR RRR Ren rnc ncn rnc nnn nnn 28 D0 12 D0 1 1 0 0 50 58 7 z zcor lr icor ispl split mass 2 100 3 500 1 1 0 5 mt sphere rou sphere weight rloc 626 lmax t 1lmax b 1lmax v Valence states are spdf states for E 0 6 Ry 4434 main quantum numbers 0000 basis set 0000 choice of Eny 0 5 0 5 0 5 0 5 Eny 1 0 2 0 3 0 4 0 Dny spdf states for E 0 4 Ry 4434 main quant
63. nted out in the output file In the CEL version this parameter is used to find mt sphere and rou sphere radii Here weights are used for blowing up the polyhedra Since one can imagine that there are fat and thin atoms then they can be described with big and small polyhedra Again only relative fatness of every atom is important the division of the space into different polyhedra will always conserve the total volume After the polyhedra are found the program will print out the radii of inscribed and circumscribed spheres in the output file and these radii must be put back into the INIFILE In the PLW version the weights also help to find the mt sphere radii The spheres around every atom will be blowed up until they touch The speed of blowing the spheres is proportional to the weight specified This again allows to control fatness of every atom rloc localization radius This large sphere around given site gives all nearest neighbors used for builiding screened LMTOs in TB calculations The parameter has no effect in all NMT programs but can be used in some applications see e g chapter ADDITIONAL INPUT HOPFILE Both CEL and PLW programs actually work only with the sphere radii while the weights are auxilary input data The strategy for choosing the spheres is the following First one chooses the weights of different sorts of atoms from some physical arguments atomic sizes lattice geometry Then one runs a special program located at
64. of first and second density gradients is required e NOPRMAX is always 48 e NENVMAX is the maximum number for the planes of polyhedra allowed during their generation Alter natively this parameter is also used to define the maximum number of cluster sites around each atom in any tight binding calculation e NVCMAX is used for the lattice summation in the structure constants It must be at least three times larger than the input parameter nvec in the STRFILE e NTERMAX is the number of iterations after which the Broyden mixing scheme restarts Emprically found value is 15 8 5 Estimation of the needed core memory The storage of the core memory is taken by several arrays The structure constants complex 16 are dimen sioned as LMAXT 2 2 NPAT LMAXB 1 2 NPAT NKAPMAX The charge density and the potential complex 16 are both dimensioned as NRADMAX 1 NSYMMAX NPAT NSPINMAX The energy bands real 8 are stored in the array of the size NDIMMAX NSPINMAX KAMX NPANMAX The size of the hamiltonian and the overlap matrices both complex 16 is NDIMMAX NSPINMAX 2 To estimate the needed memory in bytes multiply size of the complex array by 16 and real array by 8 A very usefull option is to link the programs to get the map file Under UNIX it is xlf o o main exe bloadmap map At the end of the map file there is a total ammount of core memory required by the program Disc space no significant disc storage except f
65. omment 1 2 sqrt 3 2 0 0 Bx By Bz 11 lt which is only allowed 0 0 0 0 1 0 Cx Cy Cz 111 lt when ICALC 1 Basis in the hexagonal lattice 0 0 0 0 gt 0 0 Zn 1 2 1 2 sqrt 3 1 2 Zn Rotation of the coordinate system cos pi 4 sin pi 4 0 0 sin pi 4 cos pi 4 0 0 0 0 gt 0 0 gt 1 00 Tetragonal strain c a 1 10 conserving the volume 1 10 1 3 0 0 O 1 10 1 3 0 0 O 1 10 2 3 icalc 0 give up the calculator use standard READ 1 statement e istrain switch for taking care of the charge density plane waves expansion in the presence of strain 26 4 2 istrain 1 a sphere in reciprocal lattice which selects plane waves to expand the charge density will be distorted to some ellipsoid if the strain matrix specified below is not unit matrix This is usefull for the distortions like strains changing b a c a ratio etc because if one always chooses a sphere to select plane waves for the Fourier transform during the distortions the number of plane waves is changed The latter can in principle lead to some errors in the energy difference for two lattice configurations istrain 0 always use a sphere to select plane waves for the Fourier transform ndiv1 parameter connected with the polyhedron construction This value is changed from 4 to 1 If ndivi 4 the polyhedron construction is very fast because not too many nearest neigborous are considered as th
66. on and starting from NON spin polarized charge density Then the input charge density file contains only charge density and no spin densities the potential for spin up and spin down states is equivalent and must be splitted to push the system out of the paramagnetic solution The magnetization density will be artificially introduced after the first iteration At the following iterations split will be set to zero automatically and if the system tends to be magnetic the self consistent procedure should converge to it If one continues self consistency starting from the SPIN POLARIZED charge density then non zero splitting in INIFILE will be ignored This is done in order to perform smoth continuation of the self consistent procedure from one run to another run It might happens however that it is useful to make the splittting at the first iteration even if the input charge density is spin polarized For example the system is too far from the self consistency or the previous calculation was done non magnetic but spin orbit coupled In the latter case the charge density file containes both spin up and spin down components which are equivalent For this purpose specify ispl 2 This will suppress setting split 0 at the first iteration e mass atomic mass as in the periodic table This value is read but not used by the NMT programs e mt sphere In the ASA version this value is ignored In the CEL version this is the value of the non overlapping
67. or 3p state z zcor lr icor ispl split mass mt sphere rounded sphere weight lmax t lmax b lmax v states for E 0 6 Ry main quantum numbers basis set choice of Eny Eny Dny states for E 0 4 Ry main quantum numbers basis set choice of Eny Eny Dny states for E 1 4 Ry main quantum numbers basis set choice of Eny Eny Dny of semicore states icon lt confile gt isrf lt srffile gt O nio psi ipsi lt psifile gt O nio plw itmp lt tmpfile gt O nio bnd ibnd lt bndfile gt O nio pot ipot lt potfile gt O nio ptn iptn lt ptnfile gt O nio dos idos lt dosfile gt 2 nio scf iscf lt scffile gt 2 nio out iout lt outfile gt Other Data for NMT Pack 1 4 0 0 nff nef pole 8 8 8 2 ni n2 n3 nc 32 32 32 0 02 0 04 1 5 nffti nfft2 nfft3 epsR epsG kbz rbz Additional input files 2 nio hub ihub lt hubfile gt O nio hop ihop lt hopfile gt O nio opt iopt lt optfile gt O nio enr ienr lt enrfile gt O nio pnt ipnt lt pntfile gt The meaning of the entries is explained below 3 1 Control Parameters Band Structure Calculation of NiO 3 lift 1 lmto 0 nlrt 2 lrwf 0 npfr Exchange Correlation 3 1 2 3 by VBarth H Gunn L Jan W etc e lift 1 Calculate structural data files then stop This includes structure constants of the LMTO method used by all NMT codes and or cell surface constants
68. or the structure constants file CONFILE The structure constants are stored in the reduced form for all k points Divide the memory taken by the structure constants approximately by four and multiply by a number of k points for every panel to estimate the ammount of disc space required for the CONFILE 43 9 ERROR MESSAGES The description of the error messages in the program is supposed to be essentially extended in the future Below just a few useful hints is given Generally two kind of errors exist in the program warning messages when the program does not terminate and the error messages when the program terminates Normally warning messages mean that the program can either correct the problem itself or the problem is not important for the execution like an advise to switch on scalar relativism when atomic charge exceeds 21 The error messages always mean that the program can give a wrong result if the input files will not be corrected 9 1 Errors connected with PARAM DAT In all cases the program will terminate if any of the actual parameters exceeds the parameter in the PARAM DAT Then standard error message occurs which states which of the parameters must be increased 9 2 Errors connected with input Some input data can be easily checked like the number of atoms which is read from different input files If there is a mismatch in the input a corresponding message is printed and execution is terminated 9 3 Errors connected with
69. ow the accuracy see explanation above This accuracy is regulated by the input parameters epsR and epsG in the INIFILE x xkxkkxkk STRMSH finished CPU time 94 08000 FRR kk kK KK KA kkkKAK SCELLS started CPU time 94 08000 FKK kkk a aK CK Position 00000E 00 00000E 00 00000E 00 for Nit Basis Nplw Ecut Ry RH S H S GH S RH S Ekap 400 pil st 336 17 5 99665 99933 p 548 24 4 99754 99875 a 934 35 8 99522 99316 Basis Nplw Ecut Ry RH S H S GH S RH S Ekap 1 40 pil s 336 17 5 1 0013 1 0003 p 548 24 4 99405 99682 q 934 35 8 99682 99527 Different spheres found by the program are printed below Here input mt sphere means the MT sphere read from the INIFILE this radius is actually used by the program The following lines for information only Input as sphere means the radius of the atomic sphere calculated according to the weight read from the INIFILE Cell mt sphere is the radius of the sphere inscribed into the polyhedron The polyhedra are generated according to the weights read from the INIFILE Cell as sphere is the atomic sphere estimated according to the volume of the polyhedron Touch mt sphere is the maximal radius of the spheres when they do not overlap Circums sphere is the radius of the sphere circumscribed the polyhedron Other information gives nrad 318 number of radial points for the input MT sphere znuc 28 nuclei charge zcor 18 deep core charge npln 18 number of
70. pile the file pltodos f Also compile contents of the subdirectory GRAFOR Link these file to get executable plotdos exe The input to the program is given by the DOSFILE of the main NMT package After the DOSFILE is read you are asked which of the partial or total densities of states must be plotted The output from the program has a format understandable by the SURFER GRAPHER software It can be directly plotted using view exe plot exe utilities of this software Alternatively it can be rewritten into the postscript format using the program plotps exe see below e program plotps exe is designed to reformat the output file understandable by the SURFER GRAPHER software plt format to the standard postscript No special configuring of the program is required Just compile the file plotps f link it and get executable plotps exe e program fat exe is designed to rewrite FATFILE of the NMT package to the format understandable by the plotting facility gnubnd exec of Stuttgart s TB LMTO package No special configuring of the program is required Just compile the file fat f link it and get executable fat exe The input to the program is given by the FATFILE while two output files BNDS and EIGNS are created 12 6 Program QPNT This program is used to generate irreducible points in the Brillouin zone No special configuring of the program is required Just compile contents of the directory nmtlib qpnt link object files and get executable make exe
71. planes found for this polyhedron input mt sphere is 2 100000 input as sphere is 2 610231 cell mt sphere is 2 108595 cell as sphere is 2 605776 touch mt sphere is 2 108595 nrad 318 znuc 28 zcor 12 circums sphere is 3 486185 npln 18 xexeKKKK SCELLS finished CPU time 94 37000 FRR kk k K KK k K K 7 4 Finding full potential After printing out different sphere radii the self consistent cycle starts The self consistency is controlled by the program SCF1 see source file scf1 f At the beginning of each iteration first the full poteitnal is calculated As a result the table below is produced in the OUTFILE It should be noted that the boundary values of potential V S are given with respect to the vacuum zero i e to the energy zero of the atomic program Once the average V S is found energy zero is put there and since that the items Average potential in the interstitial region Kappa s and band energies are given with respect to it It is recommended to adjust the MT radii in such a way as to make if it is possible the boundary potential values V S above not very different for different atoms The V and P values stand for the potential and pseudopotential while RO and PD values denote density and pseudodensity M S is the magnetic moment within the MT sphere PM S is the pseudomagnetic moment has no physical meaning Notation S is for the sphere while 0 is for the atom origin ITERATION 1 kkKKAK VFULL s
72. pnt program which generates irreducible set of k points according to the input division Helpful to understand how many k points is produced by certain division of the Bril louin zone Also helpful for setting up perturbation wavevectors in linear response calculations of susceptibilities phonons etc e nmtlib atom standard program after Liberman et al to solve the free atom problem The corresponding input data files are stored in the atomdat directory In fact all free atom charge densities have been calculated already for each element and also stored in this directory This directory is only used as a library to create input charge density file All programs and data files are tared gzipped and uuencoded into 5 self extractory files named as nmtasa dec nmtcel dec nmtplw dec nmtlib dec and atomdat dec To unpack them use the following commands 1 uudecode nmtasa dec 2 gunzip nmtasa tar gz 3 tar x f nmtasa tar Repeat these steps for nmtcel dec nmtplw dec nmtlib dec and atomdat dec The directory trees will be created automatically as illustrated in Figure 1 To be able to run any of the full potential LMTO programs it is necessary to compile the source data files A few comments must be said here First the maximum size of every array such as maximum number of atoms Imax etc in the program is declared using the fortran PARAMETER statement These statements are contained in the file PARAM DAT See Section PARAME
73. r rently avaiable ienr lt enrfile gt This file is not used by the NMT programs ipnt lt pntfile gt This file is not used by the NMT programs 24 4 STRUCTURE CONTROL FILE STRFILE STRFILE describes the crystal structure An example is given below keKKKA Structure Data for NiO Main Parameters 4 1 01 0 natom b a c a 1111 ibas ibz icalc istrain 4 10 ndivi ndiv2 300 1 0 nvec_max alpha Primitive Translations 1 2 1 2 1 0 Ax Ay Az 1 2 1 0 1 2 Bx By Bz 1 0 1 2 1 2 Cx Cy Cz Basis in original Unit Cell 0 0 0 0 0 0 Nii 1 0 1 0 1 0 Ni2 1 2 1 2 1 2 10 3 2 3 2 3 2 ro Basis in distorted Unit Cell 0 0 0 0 0 0 Nil 1 0 1 0 1 0 Ni2 1 2 1 2 1 2 0 3 2 3 2 3 2 0 Strain Matrix 1 0 0 0 0 0 1 1 3 0 0 1 0 0 0 s2 1 3 0 0 0 0 1 0 3 1 3 Point Group Description File 1G cubic system Primitive translations for BZ 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 High symmetry direction settings Example 4 of directions g X 0 0 0 0 0 0 1 2 0 0 0 0 g L 0 0 0 0 0 0 1 2 1 2 0 0 Z R 0 0 0 0 1 2 1 2 0 0 1 2 R A 1 2 0 0 1 2 1 2 1 2 1 2 Settings for plobands X g L Z R A The meaning of parameters is as follows 25 4 1 Main Parameters natom total number of atoms in the unit cell b a c a orthorombicity parameters ibas ibz icale switch for atomic settin
74. ration 6 2 pi 3 1 1 1 A operation 11 4 pi 3 1 1 1 A operation 12 2 pi 3 1 1 1 A operation 7 4 pi 3 1 1 1 A operation 16 pi 1 1 0 A operation 13 pi 15 1 0 A operation 18 pi 0 1 1 A operation 17 60 pi 0 1 1 A operation 23 pi 1 0 1 A operation 21 pi 1 0 1 I operation 25 C operation 26 25 2 C operation 27 25 3 C operation 28 25 4 C operation 29 25 5 C operation 30 25 6 C operation 31 25 7 C operation 32 25 8 C operation 33 25 9 C operation 34 25 10 C operation 35 25 11 C operation 36 25 12 C operation 37 25 13 C operation 38 25 14 inversion combinations 61 C operation 39 25 15 C operation 40 25 16 C operation 41 25 17 C operation 42 25 18 C operation 43 25 19 C operation 44 25 20 C operation 45 25 21 C operation 46 25 22 C operation 47 25 23 C operation 48 25 24 KOVFILE for describing the hexagonal rotational system is listed below All the expressions must be recognizable by the CALC facility see the description of the calculator in the chapter describing STRFILE Hexagonal rotations are given after Kovalev Note the difference between the present and Kovalev numbering which is given by a second column Notations are the following e U equivalent operation e A arbitrary rotation along
75. rgies is that they allow to avoid singularities of the structure consntants connected with the free electron poles ii positive kappa basis set is formed with the first x placed in the center of gravity of the occupied band another two kappa s are placed with the step 1 Ry above i e 12 0 4 Ry 92 1 4 Ry and 37 2 4 Ry Note that for the positive kappa case small imaginary part 0 03 Ry or so must be added to avoid singularities in the structure constants Positive kappa basis set reminds an expansion over plane waves while negative kappa basis looks closer to the LCAO linear combination of atomic orbitals like representation No specific preference to either chose can be given exept that the negative kappa basis should be more stable numerically no singularities in the structure constants For the narrow energy bands semicore like one can use single kappa basis set with the tail energy equal to the kinetic energy of that orbital in the interstitial region i e k E Ving where E is the position of the narrow band and Vint is the average potential in the interstitial region When both these values are given with respect to the MT zero the potential at the MT sphere averaged over different atoms the value of Vins is usually small and a good estimate for the x is just the position of the band relative to the MT zero That is actually done in the programs all energies are given with respect to the MT zero For the ASA version usag
76. ructure lattice parameter the atomic charges and the MT sphere radii of the atoms must be known To run the program goto the directory in which the input data are stored like nmt dat nio Then issue the command nmt lib scfm make exe An input to the program is INIFILE STRFILE and the localization radius in atomic units The latter is the radius of a sphere within which the summation of tails of atomic densities to a given site will be performed Usually it is 10 a u Also you will be asked by the SCFFILE which is the output from the program Specify the name like e g nio scO The SCFFILE will be stored in the current directory 30 6 RUNNING THE PROGRAM The full potential program asks for the names of three files INIFILE SCFFILE STRFILE One has either to answer this from console or to put the answers in a file for example gt nohup nmtplw run main exe lt nio job amp where nio job is nio ini nio scO nio str or to redirect the input stream gt nohup nmtplw run main exe lt lt EOF dz gt nio ini gt nio scO gt nio str gt EOF After the program finishes the output file nio scf may be renamed to nio scO OUTFILE may also be renamed and the program can be run again After shifting the atoms changing the k point sampling changing or using another tail energies K7 using another number of semicore panels CONFILE should be recalculated if it was stored before If it is necessary to chang
77. s over three lattice vectors so as to get rouphly equiddistant mesh epsR epsG accuracy of matching the spherical Hankel functions in real and reciprocal space Always specify epsR 0 02 and epsG 0 04 keyt bzr to accelerate Fourier transforms when calcualting interstitial potential matrix el ements set keyt 1 In this case the radius of the cutoff sphere in reciprocal space is set by parameter bzr times the radius of the sphere circumscribing the Brilloiun zone Usually it is 4 6 The smaller bzr value the faster calculation the lower the accuracy Because of the permanent development of different parts of the programs there is a need to provide an additional information If this is necessary additional input files must be listed at the end of INIFILE There will be no mistake if the following lines are absent in the INIFILE Additional input files 2 0 0 0 0 gt nio hub ihub lt hubfile gt nio hop ihop lt hopfile gt nio opt iopt lt optfile gt nio enr ienr lt enrfile gt nio pnt ipnt lt pntfile gt 23 ihub lt hubfile gt Specify hub 1 if it is necessary to perform LDA U calculations See chapter ADDITIONAL INPUT HUBFILE for the details ihop lt hopfile gt Specify thop 1 if it is necessary to withdraw hopping matrix elements See chapter ADDITIONAL INPUT HOPFILE for the details iopt lt optfile gt Reserved for calculating the optical properties by the package OPT not cu
78. sets is the following if the free atom state see corresponding rat lt element gt file in the directory atomdat has an energy approximately above 2 Ry from the vacuum zero or above 1 Ry from the MT zero but note that the latter is not known until the band structure calculation is performed then this state should be treated as valence state and described in the main valence panel If the state has an energy lying between 6 and 2 Ry from the vacuum zero this state should be treated as a semicore state and must be described in one of the semicore panel All other low lying states can be treated as levels e lr 0 for non relativistic calculations 1 for scalar relativistic valence states e icor 0 frozen deep core 1 soft deep core always fully relativistic e ispl split ispl 0 without permanent splitting of the spin up and spin down components of the po tential which models in this case applying an external magnetic field ispl 1 with permanent splitting on each iteration then split is the splitting between spin up and spin down components of the potential in Ry This is done according to Vele 56V 1 1 15 where V is the splitting and o runs from 1 to 2 for spin up and spin down components respectively Note then if ispl 0 but split is not zero the latter will be used only at the first iteration Specify ispl 0 and some non zero splitting when doing spin polarized calculati
79. sistency is reached 2 more fast option for force calculations but with shorter output The accuracy of forces due to unself consistency cannot be controlled In fact forces are accurate only with PLW package They are absent in ASA code See also file nmtplw run forces f for more comments 3 2 Exchange correlation functional 0 npfr Exchange Correlation 3 1 2 3 by VBarth H Gunn L Jan W etc Iterative Procedure Limits and Accuracy 20 0 2 0 2 1 D 9 2 0 0 3 0 3 niter mix mag eps lbr ibr mixb mixh e 0 not inculded e 1 von Barth Hedin e 2 Gunnarsson Ludqvist e 3 Janak Moruzzi Williams e 4 Vosko Wilk Nussair e 5 Perdew Wang local part of GGA 1991 similar to Vosko Wilk Nussair e 6 Gaspar Kohn Sham no correlation To switch on generalized gradient approximation add 10 to the key above for GGA91 of Perdew and Wang Most recent GGA96 after Perdew et al is switched on by adding 20 Therefore valid keys are Note that in PARAM DAT file there is a parameter NGGAMAX which must be set to 3 for LDA GGA calculations e 0 1 3 4 5 6 for LSDA only e 11 12 13 14 15 16 for LSDA GGA91 e 21 22 23 24 25 26 for LSDA GGA96 3 3 Iterative Procedures Limits and Accuracies Exchange Correlation 3 1 2 3 by VBarth H Gunn L Jan W etc Iterative Procedure Limits and Accuracy 12 20 0 2 0 2 1 D 9 2 0 0 3 0 3 niter mix mag eps lbr ibr mixb mixh Atomic Data 4 3 2 1 natom nsort nspin norbs e ni
80. sphere which must lie completely inside the polyhedron In the PLW version this is the non touching muffin tin sphere radius e rou sphere In the ASA and PLW versions this value is ignored In the CEL version this is the radius of a circumscribed sphere around the atomic polyhedron or slightly larger The range between MT sphere and the polyhedron boundary represents the interstitial region of a given atom while the spherical harmonic expansions are performed for the charge density and the potential in the region between inscribed MT and circumscribed spheres Depending on the shape of the polyhedron these spherical harmonic expansions can be slowly convergent For cubic BCC or FCC lattices the charge density is well convergent with lmax 8 For the diamond structure on the other hand there are some points in the cell where one center spherical harmonic expansion cannot be performed In this case one has to introduce empty spheres to improve close packing e weight initial weight factor for blowing up atomic spheres In the ASA version this is the weight with which the atomic sphere will be blown up so that the space filling occurs Only relative values of weights of different atoms play a role as they are properly normalized inside the program according to the formula 1 3 Sr gt X Wr 2 where s is the atomic sphere radius wy is the input weight and Qee is the cell volume The actual values of the atomic sphere radii are pri
81. t coupling for heavy elements At the moment three versions of the full potential code are available which use essentially the same input files Some data in the input file are ignored for one or another version These versions are NMTASA overlapping atomic spheres potential is non spherical inside the spheres no intersti tial region Fast and dirty provides reasonably good bands but is not sufficiently acurate for phonons and distortions Version 1 60 of this package or the later ones should match to this manual NMTCEL unit cell is divided into polyhedra potential is expanded in the spherical harmonics inside the polyhedra Has intermediate computation speed and provides reasonable accuracy for many phonon elastic problems provided that the structure is reasonably close packed The description of this method has been published in Ref 5 Version 3 80 of this package or the later ones should match to this manual NMTPLW plane waves non overlapping muffin tin MT spheres potential is expanded in spherical harmonics inside the spheres and Fourier transformed in the interstitial region Pro vides the best accuracy at the price of increasing the computation time A short description of this method can be found in Ref 6 Version 2 50 of this package or the later ones should match to this manual Main parts which run independently for each version are Module which makes a starting charge density file for the band str
82. tarted CPU time 97 46000 FKK kkk aK aK gt K K Input data for Nil in the position 1 gt V up S 4008890 RO up S 1974167E 01 36 P up S 4039995 P up 0 11 49737 V dn S 3899292 P dn S 3910340 P dn 0 11 50916 M S 1 787173 PD up S 2012848E 01 PD up 0 1143829E 04 RO dn S 1860992E 01 PD dn S 1878650E 01 PD dn 0 2230109E 03 PM S 4962198 Average potential over the sphere boundaries is 3368602 Average potential in the interstitial region is 9289500E 01 Total charge in the interstitial region must be 3 494624 Total charge found via fourier transform is 3 494624 Auxilary density renormalization coefficient is 9999460 Magnetization in the interstitial region is 8180455E 09 Total magnetization found in elementary cell is 2444900E 10 7 5 Calculating energy bands After constructing the full potential the execution of the program goes to the package of program for solving the eigenvalue problem of the LMTO method It is controlled ba the program BANDS see source file bands f Information about choice of Eny is prited below Eny Dny stand for the Ep D values used for this particular l channel Cny is the estimated center of the l band Wny is its width All the estimates are done using potential parameter relations docto VFULL finished CPU time 137 8900 FRR kk kK KKK AK kkKKAKK BANDS started CPU time 137 8900 k K Kk A AK gt K K 3kappa spin up panel 1 Band Struct
83. ter max number of iterations in this run e mix starting mixing of the charge density in linear mixing scheme During the iterations towards self consistency the mixing will be optimally adjasted according to the Pratt scheme This parameter is ignored if Broyden mixing see below is switched on e mag starting mixing for the magnetization in linear mixing scheme During the iterations to wards self consistency the magnetization mixing will remain constant and will NOT be adjusted The parameter has no effect for non spin polarized calculations or if Broyden mixing see below is switched on e eps total energy convergency criterion Typically 1076 Ry The program will stop if the total energy difference between two consequent iterations is less then eps e lbr switches on the Broyden mixing If 1 then Broyden is OFF if 0 then Broyden is ON for 1 0 component of p r if 1 2 then Broyden is on for l le lbr component of rho r Recommended value is 1 since it does not take much disk space because Broyden saves p r for all previous iterations Broyden restarts every time after NTERMAX iterations Parameter NTERMAX is in PARAM DAT Usually NTERMAX 15 e ibr starts Broyden after I ibr iterations If br 0 then start immidiately If br gt 0 then first I ibr iterations will be done with the linear mixing scheme where the mixing parameters miz and mag are specified above e mixb this is initial guess for Jacobian which is closely relate
84. the charge density with every iteration approaches the self consistency the iteration weight is bigger than 1 and increased can be up to infinity If the iteration gives a bad guess for the charge density the iteration weight becomes smaller than 1 If a few say 10 consequent iterations gives the weight much small than 1 of the order 0 1 than it is advised to switch off the Broyden mixing and use linear mixing instead dolo ENERGY finished CPU time 419 8300 FRR Kk KKK KKK Input output charge transfer at the iteration gt S inp S out 1 701708 1 701708 for Nil 1 701708 1 701708 for Ni2 4833732E 01 4833723E 01 for O 4833732E 01 4833723E 01 for O Input output magnetic moment at the iteration gt M inp M out 39 1 753985 1 753985 for Nil 1 753985 1 753985 for Ni2 1958433E 11 5435652E 11 for O 1955769E 11 5435652E 11 for O I inp I out 3 500091 3 500091 3 500091 3 500091 Tk hK hK hK KKK hK KK KK KK o hK hK kkk kkk Broyden Mixing for Rho r Iter Weight 1 091346 Charge Magnetization 1 701708 1 753985 for Nil 1 701708 1 753985 for Ni2 4833727E 01 2704503E 12 for 0 4833727E 01 2717826E 12 for O Interst Charge after Broyden 3 500091 Magnetization over MT spheres 1523093E 10 The last lines show the evaluation of the mean square difference between input and output charge densities magnetizations at the iteration An example below shows that the self consistency is well rea
85. the input charge density file called SCFFILE using the Mattheiss procedure by superposing free atom densities See chapter INPUT CHARGE DENSITY FILE SCFFILE for the complete description 47 12 USING NMTLIB LIBRARY The NMTLIB library contains a few subdirectories with the general purpose application programs Normally these programs understand as input output files from any of the NMT package In contrast to that inside any of the nmt directory there exists local library located in nmt lib See chapter USING LOCAL LIBRARY for the complete description 12 1 Program ATOM Free atom calculations are performed by Liberman s atomic program Input files for all atoms are stored in the directory atomdat they are called rat lt atomname gt An example of such input file is given below for Fe Most of the input parameters are self explanatory you do not have to care about number of relativistic core orbitals see parameter JCORE in table below the NMT program will do it automatically from the known core charges of constituent atoms The eigenvalues given in the rat lt atomname gt file are not more than the starting energies for a self consistent free atom calculation so their choice is not very important In the output the atomic program produces files den lt atomname gt with the charge density tab lt atomname gt with the table of the self consistent eigenvalues and inf lt atomname gt with the intermediate results
86. tion matrix as calculated within LDA This part of the input can be skipped unless Scheme LDA U1 2 is switched on 3 stands for the correction to the LDA potential in constrained LDA calculations This part of the input can be skipped unless Scheme LDA C is switched on 4 stands for the correction to the LDA potential as given in LDA U calculation These data are not used as input and for information only Therefore they can be ommited while constructing HUBFILE 56 14 ADDITIONAL INPUT HOPFILE WARNING Due to a permanent development of this part of the program some input data may differ from realization This application is designed to calculate hopping mattrix elements for tight binding description by the NMT program The development is not yet finished although some possibility for rough evaluation is provided To be able to calculate hopping integrals a special HOPFILE located in nmt dat nio nio hop must be created Also localization sphere radii must be set up in the INIFILE See the description of the input parameter rloc in the INIFILE Below is an example of HOPFILE INPUT DATA FOR TIGHT BINDING CALCULATION GENERAL SETTINGS Bands i Scheme Ry Units eV Ry avaiable Cubic N Cubic Spherical harmonics Real N Real Complex input output Yes Yes No hoppings avaiable Append 4 Overwrite Append new hoppings DESCRIPTION OF HOPPING INTEGRALS 3 From to site Energy Overlap Nil 1 3d
87. tion will be checked for matching to the current set up such as of k points tail energies etc and the structure constants will be read Setting con 2 is advised in most cases isrf lt srffile gt file for storage the cell surface constants which are necessary to integrate over the polyhedra This file is only used by the CEL program The meaning of the key isrf is the same as for the CONFILE except that there is no k point dependence on the size of this file Setting isrf 2 is advised in most cases 20 e ipsi lt psifile gt file for storage the wave functions In fact only the coefficients in the expansion of the wave function into the LMTO basis are stored The wave functions might be necessary for applications The meaning of the key psi as above Usually ipsi 0 No restart from this file can be performed e ibnd lt bndfile gt file for storage the energy bands Setting bnd 1 allows to calculate the energy bands along high symmetry directions the latter ones are listed in the STRFILE see below To control how many k points is necessary to generate along every direction use first switch in setting the k point sampling see below in this chapter The program will automatically set of iterations equal to 1 and will stop after calculating the energy bands Here some care should be taken with the CONFILE Since the k point sampling used for high symmetry directions is different from that used for integrated over the BZ the CON
88. to a commercial product References 1 O K Andersen Phys Rev B 13 3050 1975 2 P Hohenberg and W Kohn Phys Rev 136 B864 1964 3 W Kohn and L J Sham Phys Rev 140 A1133 1965 4 For a review see also Theory of the Inhomogeneous Electron Gas edited by S Lundqvist and S H March Plenum New York 1983 S Y Savrasov and D Y Savrasov Phys Rev B 46 12181 1992 S Y Savrasov Phys Rev B 54 16470 1996 a gt al 7 For a review and complete set of references see e g V Anisimov F Aryasetiawan and A I Lichtenstein J Phys Condens Matter 9 767 1997 M S Hybertsen M Schl ter and N E Christensen Phys Rev B 39 9028 1989 00 65
89. ucture calculation from the charge densities of constituent atoms Package for the self consistent FP calculation or optionally energy bands along directions Additional programs for plotting the bands densities of states charge densities etc All programs can be obtained by contacting S Savrasov using the e mail address savrasov and mpi stuttgart mpg de About the notations in this document e all file names like nio ini main exe are boldfaced e all directory names like nmtplw lib scfm are italicised e symbol in the directory name nmt refers to either asa cel or plw In other cases it has the standard meaning e capitalized names like INIFILE STRFILE are made to shorten references to the MAIN INPUT CONTROL FILE for INIFILE STRUCTURE CONTROL FILE for STRFILE etc I apologize if the description of some parameters is short and unclear Any suggestions to improve this manual are welcomed 2 INSTALLATION In this section the directories which are used for running the programs and storing input output data are described All NMT programs have the same directory organization containing see Figure 1 e nmt run directory containing the source code f text of the program written on FOR TRAN77 object files o and executable file usually named as main exe e nmt dat directory with the input output data files Usually many subdirectories are created here according to the element compound name
90. um numbers 1110 basis set 3310 choice of Eny 0 5 0 5 0 5 0 5 Eny 1 0 2 0 3 0 4 0 Dny spdf states for E 1 4 Ry 4434 main quantum numbers 1110 basis set 3310 choice of Eny 0 5 0 5 0 5 0 5 Eny 1 0 2 0 3 0 4 Dny semicore states are 1 of semicore states 3 1 4 0 0 0 n l energy for 3p state for Ni2 PSSS SE 28 D0 12 D0 1 1 0 0 5 58 7 Zz zcor lr icor ispl split mass 2 100 3 500 1 1 0 5 mt sphere rou sphere weight rloc 626 valence states are spdf 4434 0000 0000 0 5 0 5 0 5 0 5 1 0 2 0 3 0 4 0 spdf 4434 1110 3310 0 5 0 5 0 5 0 5 1 0 2 0 3 0 4 0 spdf 4434 1110 3310 0 5 0 5 0 5 0 5 1 05 2 0 3 0 4 semicore states are 1 3 1 4 0 0 0 for 0 8 DO 2 D0 1 1 0 0 D0 16 0 1 910 3 350 1 0 616 valence states are spdf 2234 1000 2000 0 5 0 5 0 5 0 5 1 0 2 0 3 0 4 0 spdf 2234 0100 0300 0 5 0 5 0 5 0 5 1 0 2 0 3 0 4 0 spdf 2234 0100 0300 0 5 0 5 0 5 0 5 gt 1 0 2 0 3 0 4 0 semicore states are 0 Output Control Parameters 2 nio con O nio srf lmax t lmax b lmax v states for E 0 6 Ry main quantum numbers basis set choice of Eny Eny Dny states for E 0 4 Ry main quantum numbers basis set choice of Eny Eny Dny states for E 1 4 Ry main quantum numbers basis set choice of Eny Eny Dny of semicore states n l energy f
91. upancies avaiable Append Overwrite Append occupancies DESCRIPTION OF CORRELATED STATES 2 of correlated states 1 3d 0 588 0 601227 0 378773 atom nl f0 f2 f4 Slater integrals 2 3d 0 588 0 601227 0 378773 atom nl f0 f2 f4 Slater integrals 1 TABLE OF OCCUPATION NUMBERS FOR LDA U State 3d for Nil spin up dn occupation numbers are yz ZX xy x2 y2 3z2 1 REAL UP 0 9500596 0 0000275 0 0000275 0 0000241 0 0000139 yz 0 0000275 0 9500596 0 0000275 0 0000241 0 0000139 ZX 0 0000275 0 0000275 0 9500596 0 0000000 0 0000278 xy 0 0000241 0 0000241 0 0000000 0 9774884 0 0000000 x2 y2 0 0000139 0 0000139 0 0000278 0 0000000 0 9774884 3z2 1 yz ZX xy x2 y2 3z2 1 IMAG UP 0 0000000 0 0000000 0 0000000 0 0000003 0 0000002 yz 0 0000000 0 0000000 0 0000000 0 0000003 0 0000002 ZX 0 0000000 0 0000000 0 0000000 0 0000000 0 0000004 xy 0 0000003 0 0000003 0 0000000 0 0000000 0 0000000 x2 y2 0 0000002 0 0000002 0 0000004 0 0000000 0 0000000 3z2 1 yz ZX xy x2 y2 3z2 1 REAL DN 0 9440717 0 0000402 0 0000402 0 0004752 0 0002744 yz 0 0000402 0 9440717 0 0000402 0 0004752 0 0002744 ZX 0 0000402 0 0000402 0 9440717 0 0000000 0 0005487 xy 0 0004752 0 0004752 0 0000000 0 1046295 0 0000000 x2 y2 0 0002744 0 0002744 0 0005487 0 0000000 0 1046295 3z2 1 yz ZX xy x2 y2 3z2 1 IMAG DN 0 0000000 0 0000000 0 0000000 0 0000005 0 0000003 yz 0 0000000 0 0000000 0 0000000 0 0000005 0 0000003 ZX 0 0000000 0 0000000 0 0000000 0 0000000 0 00000
92. ure Calculation of E k with Eny Dny Cny Wny Et 400 for Nil 40738 30297 90466 6 4873 for 4s state center 48718 62592 2 4811 2 7362 for 4p state center 60077 3 0000 60077 39204 for 3d state fix D lkappa spin up panel 2 Band Structure Calculation of E k with Eny Dny Cny Wny Et 4 00 for Nil 3 9571 5 3387 3 9856 70398E 01 for 3p state bind E Eny Dny Cny Wny Et 4 00 for Ni2 3 7966 5 2605 3 8260 74285E 01 for 3p state bind E 7 6 Integrating over Brillouin zone After the band structure calculation finished the Brillouin zone integration starts controlled by BZINT see source file bzint f The information below contains the Fermi energy found EF number of states calculated TOS which must coincide with the total valence charge DOS stands for the density of states Numbers of fully filled bands and the number of bands crossing the Fermi level is for information only They will not overwrite the input numbers nff nef from the INIFILE Calculated plasma frequencies zero in semiconductors are also printed At the end energy bands at the point are given xxkkkxkkk BANDS finished CPU time 305 5700 FRR kk kK K KK K K xxxkkkkk BZINT started CPU time 305 5700 FKK KKK k KK k gt K gt K K Semiconducting band structure discovered gt dielectric gap value is 2190656 Ry 2 980563 eV 37 EF TOS DOS 1062796 32 00000 0000000E 00 EF TOS DOS 7062796 32 00000
93. x2 y2 Ni1 0 0 O 3d x2 y2 0 301 1 0 O 3 2p z 0 f0 0 O 2pf z 0 140 1 0 Ni1 1 3d x2 y2 0 1 2 1 2 1 2 2p z 0 162 0 0 14 1 General Settings No case sensitivity is assumed in the input parameters described below e Scheme Only Scheme Bands is avaiable at the moment e Units Electronvolts or Ry input units for the energies in this file are avaiable Specify either eV or Ry e Harmonics The input can be given either in spherical or in cubic harmonics representation e Input Output can be either real or complex e Hoppings in this file either listed below yes key or absent no key e Hoppings in this file will be either overwritten during the execution overwrite option or appended append option at the end of the HOPFILE 14 2 Description of Hopping Integrals In this part of the input you must select which of the hopping matrix elements to calculate Hopping element is defined between two or the same orbitals In the first line total number of hopping elements which is supposed to calculate must be given e To select orbital from specify a sort title then atom number as listed in STRFILE then main quantum number orbital quantum number s for 1 0 p for 1 1 etc then magnetic quantum number in brackets Values of m 3 2 1 0 1 2 3 are readable for spherical harmonics For cubic harmonics use x y z when l 1 or yz zx xy 12
94. xample both plane wave set ups are the same tt of k G terms in Fourier sums 154 Brillouin zone radius calculated 8660254 2pi a Teilor s sphere radius estimated 4 330127 2pi a Teilor s cutoff energy estimated 11 41988 Ry Plane waves old 3862 new 3862 coinciding 3862 7 2 Preparing structure constants After executing the INIT subrouitine the execution transfers to the package of programs controlled by STRMSH see source file strmsh f Here the data depending on the crystalline structure are prepared Next lines give an information about number of the point group elements found for the original and distorted lattices according to the positions of atoms for original distorted lattices listed in the STRFILE The number of k points generated for the main valence panel and semicore panels is also printed xkkkkAK STRMSH started CPU time 93 15000 k K Kk a aK CK 12 elements discovered for original lattice 12 elements found for distorted lattice 65 k points generated for main valence panel 4 k points generated for semicore panels The following messages prints non zero expansion coefficients of the spherical harmonics for the charge density which are allowed by symmetry Position 00000E 00 QO000E 00 00000E 00 for Nil LMTO basis set is expanded in spherical harmonics up to Lmax 6 Charge density is expanded in spherical harmonics up to Lmax 6 34 Non zero elements allowed by symmetry are the following 1 0 1
95. ximum number of plane waves allowed for the Fourier transform of the LMTOs in the interstitial region Usually it is significantly smaller than NPLWMAX e MBASMAX is the maximum number of plane waves allowed in evaluating interstitial potential matrix elements See the description of acceleration keys in the INIFILE MBASMAX does not depend on the number of atoms as NPLWMAX and NBASMAX Usually it does not exceeds 600 42 8 3 Parameters for cells This set of parameters is effective only within the CEL package e NSHEMAX is the maximum number of radial points in the grid between inscribed and circumscribed sphere Usually it does not exceed 60 e NMSHMAX is the maximum number of polyhedron surface grid points for evaluating the integrals over the interstitial region The order of this parameter is 5000 10000 e NPOLMAX is the maximum number of Chebyshev polynoms allowed for expanding the radial behavior of the interstitial quantities Usually it is 10 8 4 Frequently used settings This set of parameters remains unchanged in most cases e NPRECISION is set to 2 double precision since some standard library routines are called differently depending whether single or double precision is used e NGGAMAX can be either 1 or 3 If NGGAMAX I1 no inclusion of GGA corrections to the LDA functional is possible To be able to make both LDA GGA calculations NGGAMAX should be set to 3 This however affects allocated memory since storage
96. y2 322 1 when l 2 or 2 512 3 y 5y2 3 2 522 3 y 12 22 2 12 y2 x y2 22 xyz when l 3 See also file nmt run cubharm f for definition of cubic harmonics 57 e To select orbital to site use the same notations as above except for the position of atom in the unit cell Specify a vector which connect site From with site to The format to set this vector is given in the example above Any of expressions understandable by calculator can be used to set coordinates For the description of calculator see section STRUCTURE CONTROL FILE e Energy is the hopping matrix element It is not input parameter to the main program but it will be overwritten or appended during the calculation e Overlap is the overlap matrix element between these two orbitals for orthogonal representation it is either 1 for the same orbital and 0 otherwise It is not input parapeter The HOPFILE can be used as input file to the package FTBHUB which is not automatically distributed with the NMT package but can be obtained upon request The FTBHUB program performs non orthogonal tight binding calculations with the hopping integrals read from HOPFILE and the Coulomb on site interactions read from HUBFILE 58 15 ADDITIONAL INPUT KOVFILE If it is necessary to built your own rotational system the file describing rotational operations KOVFILE must be created Two examples of this file are given below for cubic and hexagonal groups Note that both cu

FULL-POTENTIAL LMTO PROGRAMS ”NMT” USER's MANUAL

Contents

Download Pdf Manuals

Related Search

Related Contents

FULL-POTENTIAL LMTO PROGRAMS &rdquo;NMT&rdquo; USER's MANUAL

Contents

Download Pdf Manuals

Related Search

Related Contents

FULL-POTENTIAL LMTO PROGRAMS ”NMT” USER's MANUAL