Import SIESTA File
Previous  Top  Next

Dialog Box: Import SIESTA File [File menu or button in Startup or Control Window]

The SIESTA DFT program can be used to calculate interatomic forces for both molecules and crystals, and the output in the form of a Cartesian force-constant matrix can be read into VIBRATZ. These forces in turn can be converted to valence forces (bond and angle force constants), which can be refined by least-squares to improve agreement of calculated with observed frequencies. The VIBRATZ calculation can use the full symmetry to separate the vibrations into symmetry species, and the valence forces can easily be transferred to similar compounds.

Setup in VIBRATZ and export to SIESTA. Because the Cartesian forces must refer to a specific orientation as well as order of the atoms it is usually best to follow a certain procedure if a crystal to be calculated or if symmetry is to be used for a molecule. First the calculation should be set up in VIBRATZ. The full symmetry should be used, since that usually dictates the orientation of the molecule or crystal with respect to the Cartesian axes. It is not absolutely necessary to supply valence-force specifications at this point, since the orientation can be checked and the symmetry analysis carried out without forces, but if it is intended to convert Cartesian to valence forces it will be necessary to do a valence calculation at some time to verify that the specifications are correct.

Once a calculation has been done in VIBRATZ, select Export data for SIESTA in the Forces menu. This will write a complete .fdf input file for the SIESTA program; its name by default is myfile_siesta.fdf if the VIBRATZ data file is myfile.vbr. The dialog asks for the number of extra unit cells in each axis direction; this is for possible later use of the SIESTA results in VIBRATOR. If you intend to import the SIESTA results back into VIBRATZ, set all these to zero; then all the atoms in the primitive unit cell, and only those atoms (not those in a supercell) will be given in the .fdf file. The part which deals with calculation parameters for SIESTA gives only rudimentary parameters and may need to be modified to give good results. The system label within the .fdf file is simply the name of the VIBRATZ .vbr file plus "_siesta". Most files generated by SIESTA will use this for a file name.

SIESTA is a Fortran Program and is run in command-line mode, not in a window. The command line

siesta < myfile_siesta.fdf

will execute the data, but atomic-potential files for each of the input atom types must also be present (not supplied by VIBRATZ).

Importing SIESTA files into VIBRATZ. Having completed the above, the results can now be imported back into VIBRATZ. The SIESTA output files which are used by VIBRATZ are myfile_siesta.XV, which gives the crystal axes and atomic coordinates (in bohrs) and myfile_siesta.FC, which gives the values for the Cartesian force matrix. In VIBRATZ, select Import and choose SIESTA .XV file type. If importing data for a crystal, check all the Lattice translation boxes and if importing data for a molecule check none - these settings should be the same as those in the original VIBRATZ file. The Read Cartesian force matrix box must be checked to import forces.

Now you will be asked to locate the .XV and .FC files. The data in the .FC file is converted into a VIBRATZ .car file, which is in a somewhat different format (lower diagonal) with a different scaling from the .FC file (md/A instead of mv/A).

Now the data should be set up for a VIBRATZ calculation using Cartesian forces. Since there is no symmetry all the vibrations will be in the A species.

When you import SIESTA files in this way, VIBRATZ operates in a special fixed-orientation mode, in which the structure coordinate axes remain in exactly the same orientation as in the SIESTA calculation, instead of the standard VIBRATZ orientation (c parallel to z, b in the y-z plane, except for rhombohedral crystals). Any pre-calculation Rotations (Input menu) are not applied. This is to guarantee that the orientation is correct for the Cartesian forces. If you have set up the .fdf file as described above the orientation should be the same; after the first calculation it should be verified that the orientation of axes and locations of all atoms are the same as in the original VIBRATZ calculation. The presence of the special fixed-orientation mode is indicated in the Title/Axes dialog in the Input menu. It may be removed there. Actually any files generated by SIESTA may be imported, even without the setup described above, but the orientation may not be appropriate for a calculation in full symmetry.

If the valence forces (Forces menu) have been set up you can just select Convert Cartesian to internal in the Forces menu to convert forces from Cartesian to bond and angle forces. The number of valence forces must not be larger than the number of independent vibrations, otherwise the conversion is ambiguous (it is done by least-squares). If you get unreasonably large and/or negative values, you should eliminate one or more force-constant specifications, presumably the weakest first, although this depends on geometry.

To calculate using the new valence forces, select the Set File button in the Cartesian forces group in the Control Window and choose "Do not use Cartesian forces". You can restore the original full symmetry in this file, but it will also be necessary to delete non-symmetry-unique input atoms. The fixed-orientation mode can be de-selected in the Input1/Title-Axes dialog. It may be easier however, depending on how many symmetry-unique atoms there are, to write a forces file (Forces menu) and then read this file into the original full-symmetry data set. This can be done even if the orientation in the SIESTA file was different from that in VIBRATZ. You can also use the Cartesian force file (myfile-siesta.car) in this data set if the orientations were the same.