Hi
i have just observed that my model relaxation run continues for some iterations even after the EDIFFG = -0.050 is satisfied. is it possible as per the code?
my INCAR file is
PREC = Accurate
NPAR=16
IBRION = 2
ISIF = 3
ENCUT = 450
EDIFF = 1.0e-07
EDIFFG = -0.050
IVDW = 11
ISMEAR = -1; SIGMA = 0.2
POTIM = 0.1
ALGO = Fast
LREAL = Auto
LWAVE = .FALSE.
LCHARG = .FALSE.
ADDGRID = .TRUE.
ISYM = 1
#SYMPREC = 1e-05
AMIX = 0.01
NSW=500
NELMDL = -4
NELM = 120
--------------------------------
the summary of iterations as below (copied only last few steps)
27 Energy: -59.625307 Log|dE|: -0.926 SCF: 25 Avg|F|: 0.120 Max|F|: 0.166 Vol.: 593.8 Time: 0.39m
28 Energy: -59.705324 Log|dE|: -1.097 SCF: 27 Avg|F|: 0.088 Max|F|: 0.141 Vol.: 594.2 Time: 0.43m
29 Energy: -59.648133 Log|dE|: -1.243 SCF: 25 Avg|F|: 0.086 Max|F|: 0.178 Vol.: 594.6 Time: 0.40m
30 Energy: -59.722421 Log|dE|: -1.129 SCF: 17 Avg|F|: 0.059 Max|F|: 0.136 Vol.: 594.4 Time: 0.27m
31 Energy: -59.734137 Log|dE|: -1.931 SCF: 26 Avg|F|: 0.016 Max|F|: 0.034 Vol.: 581.4 Time: 0.41m
32 Energy: -59.746208 Log|dE|: -1.918 SCF: 21 Avg|F|: 0.034 Max|F|: 0.087 Vol.: 586.7 Time: 0.33m
33 Energy: -59.755186 Log|dE|: -2.047 SCF: 28 Avg|F|: 0.028 Max|F|: 0.038 Vol.: 591.9 Time: 0.44m
34 Energy: -59.758745 Log|dE|: -2.449 SCF: 18 Avg|F|: 0.012 Max|F|: 0.031 Vol.: 590.0 Time: 0.29m
----------------
as per my understanding the run should stop after Step 31 as it's Maximum force 0.034 is < EDIFFG. but it continued for next 3 steps and then only stopped.
May i know if anything wrong in the run?
or my understanding might be wrong. May please clarify.
Regards
Isn't it strange even after EDIFFG is satisfied the run continues?
Moderators: Global Moderator, Moderator
- SKM
- Full Member
- Posts: 125
- Joined: Wed Oct 30, 2019 5:39 am
- License Nr.: 5-516
-
- Global Moderator
- Posts: 460
- Joined: Mon Nov 04, 2019 12:44 pm
Re: Isn't it strange even after EDIFFG is satisfied the run continues?
With ISIF=3 and EDIFFG negative it needs to converge energy, forces and stress. One of them is most likely not converged.
- SKM
- Full Member
- Posts: 125
- Joined: Wed Oct 30, 2019 5:39 am
- License Nr.: 5-516