hi , i installed the new version elk-3.1.12 . i tried task 205 but i had a bad results. so i run the example of Si-phonon-DFPT. but i obtained a dipersion (PHDISP.OUT) very different from that in the example with a negative frequencies. please can someone explain to me what is the probleme.??
ii have not modified the file elk.in.
THKS!
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
maybe already quite late by I ran into the same trouble as you. A solution is to switch to frozen phonon and replace task 205 with 200. This will result in a band structure close to the provided example, but take more computational time. Another option is to increase the k-points. I tried now 16x16x16 and the resulting curve significant improved. However, I still find between the q-points some negative frequencies, which maybe a result of the interpolation. I try now with an increased mesh again.
best
Michael
PS:
It seems again that the frozen phonon approach is more solid than DFPT. But maybe I am also missing some other parameters that are important for DFPT ?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I am very probably wrong, but I took the comments at the top of this elk.in to suggest that one needs to run lots of calculations (preferably in parallel), the automatically averaged result of which should be similar to (converge towards) the provided reference output (?)
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
The comments on top just explain how calculation of phonons work. To compute phonons one requieres to construct the force constant matrix, means the repulsive force an atom experience if it becomes displaced. From the force constant matrix, we can obtain the dynamical matrix and solve a harmonic oscillators like problem to find the phonon frequencies of a material. The main challenge in these calculations is the computation of the force constants which can either be done by displacing the atoms and calculate the force duo to the displacement or by density functional pertubation theory, So there is no avaerging.
best
Michael
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
hi , i installed the new version elk-3.1.12 . i tried task 205 but i had a bad results. so i run the example of Si-phonon-DFPT. but i obtained a dipersion (PHDISP.OUT) very different from that in the example with a negative frequencies. please can someone explain to me what is the probleme.??
ii have not modified the file elk.in.
THKS!
Dear Nassima,
maybe already quite late by I ran into the same trouble as you. A solution is to switch to frozen phonon and replace task 205 with 200. This will result in a band structure close to the provided example, but take more computational time. Another option is to increase the k-points. I tried now 16x16x16 and the resulting curve significant improved. However, I still find between the q-points some negative frequencies, which maybe a result of the interpolation. I try now with an increased mesh again.
best
Michael
PS:
It seems again that the frozen phonon approach is more solid than DFPT. But maybe I am also missing some other parameters that are important for DFPT ?
I am very probably wrong, but I took the comments at the top of this elk.in to suggest that one needs to run lots of calculations (preferably in parallel), the automatically averaged result of which should be similar to (converge towards) the provided reference output (?)
The comments on top just explain how calculation of phonons work. To compute phonons one requieres to construct the force constant matrix, means the repulsive force an atom experience if it becomes displaced. From the force constant matrix, we can obtain the dynamical matrix and solve a harmonic oscillators like problem to find the phonon frequencies of a material. The main challenge in these calculations is the computation of the force constants which can either be done by displacing the atoms and calculate the force duo to the displacement or by density functional pertubation theory, So there is no avaerging.
best
Michael