hi, Admin
I am using HSE to calculate DOS, but it is very slow comparing with LDA+U. I wonder if there is any tricks to improve the calculation efficiency. I am currently calculating 8 atoms on 70 nodes; it needs about 50 hours on total nodes to finish one electronic step.
Calculation DOS with HSE
Moderators: Global Moderator, Moderator
-
cocopengy
- Newbie
- Posts: 2
- Joined: Tue Jul 27, 2010 1:33 am
Calculation DOS with HSE
Last edited by cocopengy on Tue Jul 27, 2010 1:47 am, edited 1 time in total.
-
cocopengy
- Newbie
- Posts: 2
- Joined: Tue Jul 27, 2010 1:33 am
Calculation DOS with HSE
Below is the INCAR file. I am running on 14 nodes each of which has 5 processors.
SIGMA = 0.2
ISIF = 2
IBRION = -1
ISMEAR = 0
EDIFF = 1E-4
EDIFFG = -0.001
ENCUT = 600 eV
ISPIN = 2
MAGMOM = 2 -2 -2 2 4*0
LORBIT = 10
NEDOS = 500
AMIX = 0.2
BMIX = 0.00001
AMIX_MAG = 0.8
BMIX_MAG = 0.00001
ISTART = 1
LHFCALC = .TRUE.
HFSCREEN = 0.2
NBANDS = 80
ALGO = All
IALGO = 53
TIME = 0.4
ENCUTFOCK = 0 ! omit flag for high quality calculations
NKRED = 2 ! omit flag for high quality calculations
SIGMA = 0.2
ISIF = 2
IBRION = -1
ISMEAR = 0
EDIFF = 1E-4
EDIFFG = -0.001
ENCUT = 600 eV
ISPIN = 2
MAGMOM = 2 -2 -2 2 4*0
LORBIT = 10
NEDOS = 500
AMIX = 0.2
BMIX = 0.00001
AMIX_MAG = 0.8
BMIX_MAG = 0.00001
ISTART = 1
LHFCALC = .TRUE.
HFSCREEN = 0.2
NBANDS = 80
ALGO = All
IALGO = 53
TIME = 0.4
ENCUTFOCK = 0 ! omit flag for high quality calculations
NKRED = 2 ! omit flag for high quality calculations
Last edited by cocopengy on Tue Jul 27, 2010 2:10 am, edited 1 time in total.