[Pw_forum] How to control the convergence of hybrid functional calculations

[Wei-Bing Zhang]

Dear all


I am trying to optimize the lattice parameters of Pd with PBE0 and HSE
using QE 4.2. As done in lda/gga, I try to get the total energies with
different lattice parameters and then fit them using EOS. However, I
found the calculations can converge quickly at some lattice parameters
(6.2 6.4 6.6 7.2 7.4) but fail to converge at others(6.8 7.0 7.6). I
have let the calculations run more than 3 days and the calculations
still did not converge.  I know hybrid functional calculations are
very expensive but 3 days should be long enough for a single point
energy calculations.  I think there should be something wrong in my
calculations.

Does anyone have the similar experience with me?  The input files are
attached below. If I have missed something or there is something wrong
in it, please point out.

Thanks very much. Any suggestion about hybrid functional calculations
will be highly appreciated.

Best

Wei-Bing Zhang
------------------------------------------
 &control

    calculation='scf'

    restart_mode='from_scratch',

    tprnfor = .true.

    prefix='pbe0.6.8',

    pseudo_dir = '/afs/ipp/home/w/weibingz/code/espresso-4.2/pseudo/',

    outdir='/p5/batch/weibingz/temp/'

 /

 &system

    ibrav=2, celldm(1) =6.8, nat=  1, ntyp= 1,

    degauss=0.02,

    smearing='mp',

    occupations='smearing',

    input_dft='pbe0', nqx1 = 1, nqx2 = 1, nqx3 = 1,

    x_gamma_extrapolation = .TRUE.

    ecutwfc = 50

 /

 &electrons

    conv_thr =  1.0d-8

    mixing_beta = 0.7

 /

ATOMIC_SPECIES

 Pd 106.42   Pd.UPF

ATOMIC_POSITIONS

 Pd 0.00 0.00 0.00

K_POINTS AUTOMATIC

14 14 14 1 1 1


-- 
----------------------------------------------------------------------
Dr. Wei-Bing Zhang
Fritz-Haber-Institut der Max-Planck-Gesellschaft
Faradayweg 4-6,D-14195 Berlin-Dahlem / Germany
Phone : ++49-30-8413 4830
Fax   : ++49-30-8413 4701
E-mail: zhangwei at fhi-berlin.mpg.de, weibingzhangavh at gmail.com
----------------------------------------------------------------------