1. Information on siesta
In particular, have a good look at the tutorials
http://www.icmab.es/siesta/tutorials
2. siesta installation
I have installed siesta (with LDA+U feature) in the tmc directory, please put the following line into your ~/.bashrc
export PATH=/export/home/tmc/programs/siesta-ldau/Src:$PATH
3. How to run siesta jobs on the Pu-cluster
You can submit siesta jobs to the queue system using the following commands:
qrun.sh -i fdf_file_name -o output_file_name -p number_ of_processes siesta
(reminding: To know the use of qrun.sh, simply run "qrun.sh" without any arguments)
4. How to get pseudopotential?
You can generate your own pseudopotential following the instructions in the siesta User Guide. But in the beginning, you can try to use PSP (used in another program abinit) provided on the siesta website.
http://www.icmab.es/siesta/Pseudos-Bases
Some of them are available in
/export/home/tmc/programs/siesta-psplib
5. How to prepare molecular/crystal structure for siesta?
I have written a Python script included in tmckit that can convert molecular or crystal structures in different formats.
Usage: py_convert [options]
Options:
-h # display this help information
-f <file > # prefix for input and output
-i <input_format> # input format, available option:
xyz: xyz format
mol: MDL MOL
w2k: wien2k struct format
-o <out_format > # output format, available option:
xyz: XYZ format
w2k: wien2k struct format
aims: FHI-aims geometry.in format
siesta: SIESTA fdf format
gjf: Gaussian job file
--debug # set debug mode
-t <m/c> # molecule or crystal, if the input is w2k structure,
the default is "c" (crystal), otherwise the default is molecule
Examples:
py_convert -f h2o -i xyz -o gjf # convert the xyz format (h2o.xyz) to Gaussian input file (gjf) format (h2o.gjf)
py_convert -f h2o -i gjf -o siesta # convert the gjf (e.g. generated from GaussView) to siesta fdf file
For molecules, you can generate the structure using GaussView, and then convert from gjf to the siesta fdf format.
For crystals, it is a little more complicated currently since py_convert currently does not support cif yet. One way to do it is to first convert cif to w2k struct format, using a wien2k utility
cif2struct cif_file_name
(Note: the cif file you get is maybe in the MS-DOS text format, so before using cif2struct, please first use "dos2unix" to change it to Unix text format.)
To use cif2struct, you have to put the wien2k installation directory into your PATH
export PATH=/export/home/tmc/programs/wien2k/v091:$PATH
Then you convert the struct into siesta fdf format.
py_convert will generate two "fdf" files, <file>.fdf and <file>-xyz.fdf. The first one is the master input file needed to run a siesta calculation with main default input parameters. Be aware that the parameters in this automatically generated file is NOT complete and may not be optimal for your systems. You may have to add new parameters or modify the values. The second file contains all structure information. You can copy the two files into a new directory, and together with the pseudopotential files, you are ready to run the calculations.
Siesta is very efficient, but we have to be aware that because of the use of local atomic-like basis set, whose choices are also very limited, the accuracy may be not as good as other programs. In addition, the use of pseudopotential also introduces additional uncertainty. So before any production calculations, you have to always make some security checks concerning the accuracy of the PSP and numerical basis.
Please let me know whether you have any questions. This can be the first version of the siesta@TMC-FAQ. Please feel free to correct/extend it, and share your contributions.
Reference: http://old.chem.pku.edu.cn/jianghgroup/codes/siesta-howto.html
0 Comments