First of, I would like to thank the Exciting team for creating the excellent alternative to commercial LAPW programs! It's great to have such feature-rich and free software!
I am new to ab initio calculations and learning how to use the program using Wien2k "primer" from http://www.fys.kuleuven.ac.be/iks/nvsf/publications/DFT_and_LAPW.pdf. Comparing the two programs, there are some thing which I would like to be clarified. In the Exciting code features it is stated that it uses FP-LAPW basis set with local orbitals. Using Wien2k terminology, is it LAPW+LO? Are APW+lo functions also available? Is there any documentation on the structure of species files and on setting up the local orbitals?
The other question is more practical. I am trying to optimize the calculation parameters (again following the same Wien2k primer). After finding the optimal mesh I am checking the convergence with respect to rgkmax. The convergence is good at rgkmax=6.0-7.0, but becomes worse for rgkmax=7.5 (the total energy keeps going down). The calculations don't converge within 200 iterations for rgkmax=8.0. What could be the problem? My material system is 4H-SiC with 8 atoms in the cell and autormt turned on with default parameters.
Thanks very much in advance for any info,
Andrei
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
The EXCITING code is similar to Wien2k, but is somewhat more flexible in that you can have any combination of (L)APWs and local orbitals for any energy and l-value. You can also match the APWs to any order of derivative (the practical limit is 6th order) for very smooth basis sets (Super-LAPW etc.). The species files are fairly self-explanatory, although in the next release (0.9.8x), there will be an example which requires a new species file (Ununhexium) as an example of how to construct them. Complete documentation will be available in a few months.
With regard to the second question, we find that if you increase rgkmax you also have to increase gmaxvr (default is 12), and possibly lmaxapw. Try gmaxvr=14 and lmaxapw=10, this should fix the problem.
Let me know if this works.
Regards, Kay.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Thank you for the explanation. Increasing gmaxvr and lmaxapw helped to restore the convergence of the self-consistent loop. Further increasing these parameters (to 16 and 12, respectively) made the convergence worse, however. Is this explained by the fact that the basis set becomes "overcomplete" and linearly dependent or by some other reason?
Thanks and regards,
Andrei
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi,
First of, I would like to thank the Exciting team for creating the excellent alternative to commercial LAPW programs! It's great to have such feature-rich and free software!
I am new to ab initio calculations and learning how to use the program using Wien2k "primer" from http://www.fys.kuleuven.ac.be/iks/nvsf/publications/DFT_and_LAPW.pdf. Comparing the two programs, there are some thing which I would like to be clarified. In the Exciting code features it is stated that it uses FP-LAPW basis set with local orbitals. Using Wien2k terminology, is it LAPW+LO? Are APW+lo functions also available? Is there any documentation on the structure of species files and on setting up the local orbitals?
The other question is more practical. I am trying to optimize the calculation parameters (again following the same Wien2k primer). After finding the optimal mesh I am checking the convergence with respect to rgkmax. The convergence is good at rgkmax=6.0-7.0, but becomes worse for rgkmax=7.5 (the total energy keeps going down). The calculations don't converge within 200 iterations for rgkmax=8.0. What could be the problem? My material system is 4H-SiC with 8 atoms in the cell and autormt turned on with default parameters.
Thanks very much in advance for any info,
Andrei
Dear Andrei,
The EXCITING code is similar to Wien2k, but is somewhat more flexible in that you can have any combination of (L)APWs and local orbitals for any energy and l-value. You can also match the APWs to any order of derivative (the practical limit is 6th order) for very smooth basis sets (Super-LAPW etc.). The species files are fairly self-explanatory, although in the next release (0.9.8x), there will be an example which requires a new species file (Ununhexium) as an example of how to construct them. Complete documentation will be available in a few months.
With regard to the second question, we find that if you increase rgkmax you also have to increase gmaxvr (default is 12), and possibly lmaxapw. Try gmaxvr=14 and lmaxapw=10, this should fix the problem.
Let me know if this works.
Regards, Kay.
Dear Kay,
Thank you for the explanation. Increasing gmaxvr and lmaxapw helped to restore the convergence of the self-consistent loop. Further increasing these parameters (to 16 and 12, respectively) made the convergence worse, however. Is this explained by the fact that the basis set becomes "overcomplete" and linearly dependent or by some other reason?
Thanks and regards,
Andrei