Dear Experts,
I would like to know whether the fine electron and phonon grid should be equal for a converged result in epw calculation or it will be taken as per our convenience.Is there any physics which says nkf=nqf.For example if I take nk=4,4,4 and nq=2,2,2 then is it mandatory that nkf=nqf=8,8,8 or we can amplify both of the quantities with different amplification factors.
Thanks and regards,
Anindya Bose
Research Fellow,
IIIT Allahabad
Fine K and q grid related query
Moderator: stiwari
-
- Posts: 75
- Joined: Sat Mar 03, 2018 7:37 am
- Affiliation:
Re: Fine K and q grid related query
Hi Anindya,
The fine k/q grids need not be 8x8x8 in your example. In fact, 8x8x8 would be too small for realistic calculations. You want to make sure that the quantities you are interested in are converged relative to the fine grids. So maybe you reach convergence with a fine k-grid of 60x60x60 and fine q-grid of 30x30x30. For Si (https://doi.org/10.1209/0295-5075/109/57006), the convergence in transport properties is reached with a fine k-grid of 30x30x30 and fine q-grid of 60x60x60.
So it depends on the problem at hand. For example, the following paper (https://doi.org/10.1103/PhysRevB.95.075206) deals with calculating the mobility of GaAs. Fig.1 gives you some idea how they converge the mobility with respect to the fine k/q grids.
Cheers,
Vahid
Vahid Askarpour
Department of Physics and Atmospheric Science
Dalhousie University,
Halifax, NS, Canada
The fine k/q grids need not be 8x8x8 in your example. In fact, 8x8x8 would be too small for realistic calculations. You want to make sure that the quantities you are interested in are converged relative to the fine grids. So maybe you reach convergence with a fine k-grid of 60x60x60 and fine q-grid of 30x30x30. For Si (https://doi.org/10.1209/0295-5075/109/57006), the convergence in transport properties is reached with a fine k-grid of 30x30x30 and fine q-grid of 60x60x60.
So it depends on the problem at hand. For example, the following paper (https://doi.org/10.1103/PhysRevB.95.075206) deals with calculating the mobility of GaAs. Fig.1 gives you some idea how they converge the mobility with respect to the fine k/q grids.
Cheers,
Vahid
Vahid Askarpour
Department of Physics and Atmospheric Science
Dalhousie University,
Halifax, NS, Canada
-
- Posts: 75
- Joined: Sat Mar 03, 2018 7:37 am
- Affiliation:
Re: Fine K and q grid related query
Dear Mr Vahid,
As you said, with nk=4,4,4 and nq=2,2,2 the expected fine grid value should be nkf=60,60,60 and nqf=30,30,30 for convergence.For example if I take nk=16,16,1 and nq=8,8,1 then my approximated fine grid value will be 15 times of it nkf=240,240,1 and nqf=120,120,1 for convergence or it can be more?
Thanks and regards,
Anindya Bose
As you said, with nk=4,4,4 and nq=2,2,2 the expected fine grid value should be nkf=60,60,60 and nqf=30,30,30 for convergence.For example if I take nk=16,16,1 and nq=8,8,1 then my approximated fine grid value will be 15 times of it nkf=240,240,1 and nqf=120,120,1 for convergence or it can be more?
Thanks and regards,
Anindya Bose
Re: Fine K and q grid related query
Hi Anindya,
I said you "may" reach convergence with a fine k-grid of 60x60x60 and fine q-grid of 30x30x30. Maybe you need larger grids or smaller grids. So it is up to you to do convergence tests for different grids to ensure that your quantity of interest is converged.There is no magic factor such as 15 to go from coarse to fine grids.
Cheers,
Vahid
I said you "may" reach convergence with a fine k-grid of 60x60x60 and fine q-grid of 30x30x30. Maybe you need larger grids or smaller grids. So it is up to you to do convergence tests for different grids to ensure that your quantity of interest is converged.There is no magic factor such as 15 to go from coarse to fine grids.
Cheers,
Vahid