[SOLVED] line minimization not converged after 13 restarts

Total energy, DFT+U, BigDFT,...

Moderator: amadon

[SOLVED] line minimization not converged after 13 restarts

Postby mverstra » Mon Feb 22, 2010 4:22 pm

Hello,

this error used to appear frequently because of the conjugate gradient method used to get the ground state wavefunctions (iscf 5). It means that the CG solver is stuck, and a quadratic fit to the energy cannot find a reasonable prediction for the next SCF step (after 13 tries) and it stops.

This has been superseded by iscf 7 (the default) which never gets into this state. Use that instead. For PAW potentials use the default iscf 17 (basically you can always leave out iscf - the default is the best).

Matthieu

hello
I ran rf1 lesson for LaCoO3
my input variables are:

# Crystalline LaCoO3(cubic) : computation of the total energy
#

#Specific to ground state calculation

kptopt 1

tolvrs 1.0d-18
iscf 5


#######################################################################
#Common input variables

#Definition of the unit cell
acell 3*7.22
rprim 1.0 0.0 0.0
0.0 1.0 0.0
0.0 0.0 1.0

#Definition of the atom types
ntypat 3
znucl 57 27 8

#Definition of the atoms
natom 5
typat 1 2 3 3 3
xred
0.0 0.0 0.0
0.5 0.5 0.5
0.5 0.5 0.0
0.5 0.0 0.5
0.0 0.5 0.5
#Gives the number of band, explicitely (do not take the default)
nband 15
#Exchange-correlation functional
ixc 1

#Definition of the planewave basis set
ecut 55.0

#Definition of the k-point grid
kptrlatt -4 4 4
4 -4 4
4 4 -4

#Definition of the SCF procedure
nstep 55
diemac 9.0

and the error in log file is:

,Min el dens= 7.6096E-05 el/bohr^3 at reduced coord. 0.0000 0.0000 0.0000
,Max el dens= 1.3258E+00 el/bohr^3 at reduced coord. 0.5600 0.5600 0.5600
rhohxc_coll : enter with option, nspden 1 1
ETOT 46 -85.312003083360 -8.606E-05 1.641E-04 1.992E-01 0.000E+00 0.000E+00
scprqt: <Vxc>= -4.3537831E-01 hartree
findmin : lambda_predict -7.9456E-02 etotal_predict -8.5312079788E+01

scfcge : ERROR -
Potential-based CG line minimization not converged after 13 restarts.
Action : read the eventual warnings about lack of convergence.
Some might be relevant. Otherwise, raise nband.

leave_new : decision taken to exit ...
1


and if I increase the nband, in 3rd program in rf1 lesson an error occurs because of this increasing.
how can I solve this problem?


thanks
Matthieu Verstraete
University of Liege, Belgium
mverstra
 
Posts: 637
Joined: Wed Aug 19, 2009 12:01 pm

Return to Ground state

Who is online

Users browsing this forum: Google [Bot] and 4 guests

cron