DFPT (IBRION=8) does not satisify EDIFF convergence criteria
Posted: Mon Jul 09, 2012 9:16 pm
I am using IBRION=8 for some phonon calculations and have repeatedly noticed inconsistencies in the break condition for the electronic self-consistent loop.
The initial wavefunction optimization consistently satisfies my specified EDIFF criteria; however, the linear response calculations for each degree of freedom typically do not meet the EDIFF criteria. For example, with EDIFF=1e-6 the linear response steps may end with only dE/deps=1e-3. The maximum number of electrionic sc steps has not yet been reached and no error is given.
Are there other parameters I need to specify? What can I do to ensure reliable results?
My input includes:
ISTART = 1
LDAU = .True.
LDAUTYPE = 2
LDAUL = 3 -1
LDAUU = 5 0
LASPH = .True.
LMAXMIX = 6
PREC = accurate
ENCUT = 500
LREAL = .FALSE.
ISMEAR = 0
SIGMA = 0.05
ALGO = Fast
EDIFF = 1E-06
NELM = 120
NELMIN = 8
IBRION = 8
ISPIN = 2
LORBIT = 11
The initial wavefunction optimization consistently satisfies my specified EDIFF criteria; however, the linear response calculations for each degree of freedom typically do not meet the EDIFF criteria. For example, with EDIFF=1e-6 the linear response steps may end with only dE/deps=1e-3. The maximum number of electrionic sc steps has not yet been reached and no error is given.
Are there other parameters I need to specify? What can I do to ensure reliable results?
My input includes:
ISTART = 1
LDAU = .True.
LDAUTYPE = 2
LDAUL = 3 -1
LDAUU = 5 0
LASPH = .True.
LMAXMIX = 6
PREC = accurate
ENCUT = 500
LREAL = .FALSE.
ISMEAR = 0
SIGMA = 0.05
ALGO = Fast
EDIFF = 1E-06
NELM = 120
NELMIN = 8
IBRION = 8
ISPIN = 2
LORBIT = 11