Hello,<br>
I am somewhat new to LAPW and have some questions about generating an equation of state for a magnetic system. I am starting with the input from the example NiO-LDA+U-INT. As a first step, I attempted to converge the calculation with respect to the k-points and the basis parameters. To do this, I increased ngridk from 4 4 4 to 6 6 6 and 8 8 8 and found that the energy had converged within 1 mRy. Then I attempted to converge the rest of the calculation by increasing rgkmax. I quickly encountered instabilities in the calculation around rgkmax = 9. <br>
I searched the list and found the suggestion that I check the eigenvalues (which did have some very negative values) and as a result set lradstp=1 and cfdamp=1.0. Doing this resulted in slow calculations, but the Eigenvalues looked sane and everything was stable. The problem is that the energy was not converging even up to relatively large rgkmax (the energy was decreasing by about 0.04 Ry / 1 rgkmax). I have a feeling that I am going about the problem incorrectly, but don't quite know where to turn. I have seen suggested the use of local orbitals, but I cannot quite figure out how to properly edit Ni.in.
Could someone please give me some pointers as to how to think about the problem and specifically how to get NiO with LDA+U to converge? I would like to calculate a full equation of state, but at this point, I'd be happy with a single calculation that I could believe! Thanks so much,<br>
Luke Shulenburger
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I think the problem is not with lradstp or cfdamp (in fact don't use cfdamp if you can avoid it). Instead set 'gmaxvr' to a larger number (say 16.0). This will probably cure the instability problems.
You can also email me (dewhurst@mpi-halle.mpg.de) for an early pre-release of version 0.9.298 which should be more stable and has improved species files.
Cheers,
Kay.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hello,<br>
I am somewhat new to LAPW and have some questions about generating an equation of state for a magnetic system. I am starting with the input from the example NiO-LDA+U-INT. As a first step, I attempted to converge the calculation with respect to the k-points and the basis parameters. To do this, I increased ngridk from 4 4 4 to 6 6 6 and 8 8 8 and found that the energy had converged within 1 mRy. Then I attempted to converge the rest of the calculation by increasing rgkmax. I quickly encountered instabilities in the calculation around rgkmax = 9. <br>
I searched the list and found the suggestion that I check the eigenvalues (which did have some very negative values) and as a result set lradstp=1 and cfdamp=1.0. Doing this resulted in slow calculations, but the Eigenvalues looked sane and everything was stable. The problem is that the energy was not converging even up to relatively large rgkmax (the energy was decreasing by about 0.04 Ry / 1 rgkmax). I have a feeling that I am going about the problem incorrectly, but don't quite know where to turn. I have seen suggested the use of local orbitals, but I cannot quite figure out how to properly edit Ni.in.
Could someone please give me some pointers as to how to think about the problem and specifically how to get NiO with LDA+U to converge? I would like to calculate a full equation of state, but at this point, I'd be happy with a single calculation that I could believe! Thanks so much,<br>
Luke Shulenburger
Hi Luke,
I think the problem is not with lradstp or cfdamp (in fact don't use cfdamp if you can avoid it). Instead set 'gmaxvr' to a larger number (say 16.0). This will probably cure the instability problems.
You can also email me (dewhurst@mpi-halle.mpg.de) for an early pre-release of version 0.9.298 which should be more stable and has improved species files.
Cheers,
Kay.
in what equation of state to do fitting in elk code?
Hi Hamza,
The utility 'eos' fits the energy-volume data. You have various options. They are:
Planet Int. 109, p1 (1998))
Cheers,
Kay.
thank you
best regard