[Pw_forum] Davidson diagonalization with overlap
Hadi Arefi
hadi.arefi at tyndall.ie
Tue May 6 12:01:15 CEST 2014
Dear Qianqian,
Your atomic coordinate could be one of the reasons you are not getting the diagonal matrix! So you better send your input file completely, however sometimes if there is not a segmentation fault, changing diagonalization to ‘cg’ might solve the problem.
Best,
Hadi.
From: pw_forum-bounces at pwscf.org [mailto:pw_forum-bounces at pwscf.org] On Behalf Of Qianqian Wang
Sent: 06 May 2014 01:21
To: pw_forum at pwscf.org
Subject: [Pw_forum] Davidson diagonalization with overlap
Hi,
All users, I calculated a Ca-Si-O system. After vcrelax, I put one hydrogen into it and want to find a proper position for hydrogen in this system. But mostly, the jobs stopped at "Davidson diagonalization with overlap". I couldn't find the problem. Sometimes, it can run properly, but mostly, I couldn't get the proper results.
Maybe, the initial hydrogen positions are not good enough, but how can I get the proper results?
This is the input file:
&CONTROL
calculation = 'relax' ,
outdir = '/home/xxx' ,
pseudo_dir = '/home/xxx' ,
prefix = 'r1' ,
etot_conv_thr = 1.0D-4 ,
forc_conv_thr = 1.0D-3 ,
nstep = 500 ,
tstress = .true. ,
tprnfor = .true. ,
/
&SYSTEM
ibrav = 0,
nat = 29,
ntyp = 4,
tot_charge = +1 ,
ecutwfc = 45.D0 ,
ecutrho = 450.D0 ,
occupations = 'smearing' ,
degauss = 0.02 ,
smearing = 'methfessel-paxton' ,
nspin = 1 ,
noncolin = .false. ,
exxdiv_treatment = 'gygi-baldereschi' ,
/
&ELECTRONS
electron_maxstep = 500,
scf_must_converge = .true. ,
conv_thr = 1.D-6 ,
mixing_beta = 0.3 ,
/
&IONS
ion_dynamics = 'bfgs' ,
/
&CELL
cell_dynamics = 'bfgs' ,
press = 0 ,
/
CELL_PARAMETERS angstrom
5.101656231 0.000000000 0.000000000
0.000000000 11.2966859 0.000000000
0.000000000 0.000000000 6.787044
ATOMIC_SPECIES
Thank you very much!
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20140506/45a20b4e/attachment.html>
More information about the users
mailing list