Weird energy jump in relaxation
Moderators: Global Moderator, Moderator
-
- Newbie
- Posts: 25
- Joined: Tue Jan 07, 2014 11:22 am
Weird energy jump in relaxation
I am calculating the energy of a 72-atom fcc-Ni cell with a single Zr atom. (Using different scale factors and ISIF=4 to relax the cell and ions)
During the relaxation, the structure seems to converge but then suddenly jumps to a higher energy.
$ grep "F=" OSZICAR
1 F= -.39776656E+03 E0= -.39776853E+03 d E =-.397767E+03 mag= 40.7926
2 F= -.39822670E+03 E0= -.39822890E+03 d E =-.460141E+00 mag= 40.9260
3 F= -.39825537E+03 E0= -.39825761E+03 d E =-.488806E+00 mag= 40.9707
4 F= -.39831827E+03 E0= -.39832062E+03 d E =-.629032E-01 mag= 40.9742
5 F= -.39833509E+03 E0= -.39833749E+03 d E =-.797222E-01 mag= 40.9736
6 F= -.39833516E+03 E0= -.39833757E+03 d E =-.797970E-01 mag= 40.9737
7 F= -.39834431E+03 E0= -.39834672E+03 d E =-.914565E-02 mag= 40.9678
8 F= -.39834494E+03 E0= -.39834736E+03 d E =-.978246E-02 mag= 40.9661
9 F= -.39834611E+03 E0= -.39834853E+03 d E =-.117021E-02 mag= 40.9648
10 F= -.39834617E+03 E0= -.39834859E+03 d E =-.122705E-02 mag= 40.9643
11 F= -.39834636E+03 E0= -.39834878E+03 d E =-.184084E-03 mag= 40.9638
12 F= -.39829695E+03 E0= -.39829938E+03 d E =0.492170E-01 mag= 40.9613 <-- first energy jumo
13 F= -.39829720E+03 E0= -.39829962E+03 d E =-.242974E-03 mag= 40.9620
14 F= -.39721318E+03 E0= -.39721559E+03 d E =0.108377E+01 mag= 40.9322 <- second larger energy jump
Weirdly, the program stops directly after this jump:
14 F= -.39721318E+03 E0= -.39721559E+03 d E =0.108377E+01 mag= 40.9322
curvature: -0.09 expect dE=-0.650E-06 dE for cont linesearch -0.195E-06
trial: gam= 0.00000 g(F)= 0.707E-05 g(S)= 0.101E-07 ort =-0.979E-05 (trialstep = 0.168E+01)
search vector abs. value= 0.708E-05
reached required accuracy - stopping structural energy minimisation
There are some numprob-warnings during the run that may be responsible for the jump, but I find the behaviour of the ionic relaxation somewhat puzzling: Why would the algorithm assume that things have converged directly after a large increase in energy?
Input files are attached.
During the relaxation, the structure seems to converge but then suddenly jumps to a higher energy.
$ grep "F=" OSZICAR
1 F= -.39776656E+03 E0= -.39776853E+03 d E =-.397767E+03 mag= 40.7926
2 F= -.39822670E+03 E0= -.39822890E+03 d E =-.460141E+00 mag= 40.9260
3 F= -.39825537E+03 E0= -.39825761E+03 d E =-.488806E+00 mag= 40.9707
4 F= -.39831827E+03 E0= -.39832062E+03 d E =-.629032E-01 mag= 40.9742
5 F= -.39833509E+03 E0= -.39833749E+03 d E =-.797222E-01 mag= 40.9736
6 F= -.39833516E+03 E0= -.39833757E+03 d E =-.797970E-01 mag= 40.9737
7 F= -.39834431E+03 E0= -.39834672E+03 d E =-.914565E-02 mag= 40.9678
8 F= -.39834494E+03 E0= -.39834736E+03 d E =-.978246E-02 mag= 40.9661
9 F= -.39834611E+03 E0= -.39834853E+03 d E =-.117021E-02 mag= 40.9648
10 F= -.39834617E+03 E0= -.39834859E+03 d E =-.122705E-02 mag= 40.9643
11 F= -.39834636E+03 E0= -.39834878E+03 d E =-.184084E-03 mag= 40.9638
12 F= -.39829695E+03 E0= -.39829938E+03 d E =0.492170E-01 mag= 40.9613 <-- first energy jumo
13 F= -.39829720E+03 E0= -.39829962E+03 d E =-.242974E-03 mag= 40.9620
14 F= -.39721318E+03 E0= -.39721559E+03 d E =0.108377E+01 mag= 40.9322 <- second larger energy jump
Weirdly, the program stops directly after this jump:
14 F= -.39721318E+03 E0= -.39721559E+03 d E =0.108377E+01 mag= 40.9322
curvature: -0.09 expect dE=-0.650E-06 dE for cont linesearch -0.195E-06
trial: gam= 0.00000 g(F)= 0.707E-05 g(S)= 0.101E-07 ort =-0.979E-05 (trialstep = 0.168E+01)
search vector abs. value= 0.708E-05
reached required accuracy - stopping structural energy minimisation
There are some numprob-warnings during the run that may be responsible for the jump, but I find the behaviour of the ionic relaxation somewhat puzzling: Why would the algorithm assume that things have converged directly after a large increase in energy?
Input files are attached.
You do not have the required permissions to view the files attached to this post.
-
- Global Moderator
- Posts: 506
- Joined: Mon Nov 04, 2019 12:41 pm
- Contact:
Re: Weird energy jump in relaxation
Could you please share the OUTCAR and OSZCAR as well?
I would like to have a look.
In the meantime, I see that you are setting some very particular input variables:
ROPT - I think this one will not be used because you have LREAL=Auto
AMIN, AMIX, BMIX - why not the default value?
Check if removing these variables from the input fixes the numprob warnings.
If they are still present you might try not setting ALGO=Fast but ALGO=Normal.
NB: looking at the value of KSPACING I have the feeling that you could increase it without sacrificing the accuracy of the results. But I trust you did your own testing.
I would like to have a look.
In the meantime, I see that you are setting some very particular input variables:
ROPT - I think this one will not be used because you have LREAL=Auto
AMIN, AMIX, BMIX - why not the default value?
Check if removing these variables from the input fixes the numprob warnings.
If they are still present you might try not setting ALGO=Fast but ALGO=Normal.
NB: looking at the value of KSPACING I have the feeling that you could increase it without sacrificing the accuracy of the results. But I trust you did your own testing.
-
- Newbie
- Posts: 25
- Joined: Tue Jan 07, 2014 11:22 am
Re: Weird energy jump in relaxation
Sorry, the OUTCAR is too large to post it here (194 MB). I put the OSZICAR (slightly truncated at the beginning since it is also too large ) at the end. (I could put the OUTCAR somewhere in the cloud if that is helpful.)
The low k-spacing is because I am doing a convergence test; I found some surprisingly large deviations between kspacings of roughly 0.1 and 0.08.
The other parameters (AMIN/AMIX/BMIX) have improved convergence in past runs, that's why I use them.
I'm not sure about ROPT, the manual gives some default values
wiki/index.php/ROPT
but I thought those will be overridden by values in the INCAR.
Thanks for any help,
Martin.
Here is the OSZICAR
The low k-spacing is because I am doing a convergence test; I found some surprisingly large deviations between kspacings of roughly 0.1 and 0.08.
The other parameters (AMIN/AMIX/BMIX) have improved convergence in past runs, that's why I use them.
I'm not sure about ROPT, the manual gives some default values
wiki/index.php/ROPT
but I thought those will be overridden by values in the INCAR.
Thanks for any help,
Martin.
Here is the OSZICAR
Code: Select all
N E dE d eps ncg rms rms(c)
DAV: 1 0.194066512763E+05 0.19407E+05 -0.70454E+05880376 0.243E+03
DAV: 2 0.156275678166E+04 -0.17844E+05 -0.17322E+05810880 0.767E+02
...
RMM: 36 -0.397766542269E+03 0.16445E-04 0.97356E-03614259 0.129E-03 0.677E-03
RMM: 37 -0.397766558556E+03 -0.16286E-04 -0.16102E-04461701 0.783E-04 0.300E-03
RMM: 38 -0.397766558704E+03 -0.14796E-06 -0.57596E-07430042 0.474E-04 0.134E-03
RMM: 39 -0.397766558795E+03 -0.91299E-07 -0.18081E-08424835 0.260E-04
1 F= -.39776656E+03 E0= -.39776853E+03 d E =-.397767E+03 mag= 40.7926
N E dE d eps ncg rms rms(c)
DAV: 1 -0.378508978405E+03 0.19258E+02 -0.33301E+02****** 0.164E+02 0.189E+01
RMM: 2 -0.410683126008E+03 -0.32174E+02 -0.24352E+02990903 0.242E+01 0.444E+01
RMM: 3 -0.403301592250E+03 0.73815E+01 -0.12013E+01901301 0.543E+00 0.403E+01
RMM: 4 -0.399302811399E+03 0.39988E+01 -0.25978E+00819942 0.283E+00 0.190E+01
....
RMM: 23 -0.398226701611E+03 -0.23392E-05 -0.28553E-06818961 0.383E-03 0.286E-02
RMM: 24 -0.398226700148E+03 0.14622E-05 -0.12665E-06756605 0.242E-03 0.136E-02
RMM: 25 -0.398226700194E+03 -0.45198E-07 -0.23226E-07497663 0.112E-03
2 F= -.39822670E+03 E0= -.39822890E+03 d E =-.460141E+00 mag= 40.9260
N E dE d eps ncg rms rms(c)
DAV: 1 -0.396235231340E+03 0.19915E+01 -0.36836E+01****** 0.556E+01 0.403E+00
RMM: 2 -0.399210474790E+03 -0.29752E+01 -0.27446E+01942340 0.778E+00 0.995E+00
RMM: 3 -0.398504503564E+03 0.70597E+00 -0.72525E-01898006 0.138E+00 0.104E+01
...
RMM: 21 -0.398255365246E+03 0.14229E-06 -0.23622E-06517052 0.121E-03 0.594E-03
RMM: 22 -0.398255365144E+03 0.10230E-06 -0.13409E-07433866 0.879E-04 0.256E-03
RMM: 23 -0.398255365118E+03 0.26048E-07 -0.45278E-08424548 0.476E-04
3 F= -.39825537E+03 E0= -.39825761E+03 d E =-.488806E+00 mag= 40.9707
N E dE d eps ncg rms rms(c)
DAV: 1 -0.395732497214E+03 0.25229E+01 -0.44379E+01****** 0.696E+01 0.473E+00
RMM: 2 -0.399462273277E+03 -0.37298E+01 -0.34816E+01934501 0.936E+00 0.108E+01
RMM: 3 -0.398719969587E+03 0.74230E+00 -0.89286E-01861050 0.169E+00 0.130E+01
...
RMM: 17 -0.398318269643E+03 0.58428E-05 -0.56541E-06812439 0.508E-03 0.226E-02
RMM: 18 -0.398318269530E+03 0.11286E-06 -0.20823E-06806842 0.313E-03 0.202E-02
RMM: 19 -0.398318268422E+03 0.11073E-05 -0.11626E-06781512 0.250E-03 0.988E-03
RMM: 20 -0.398318268365E+03 0.57247E-07 -0.50151E-07653079 0.162E-03
4 F= -.39831827E+03 E0= -.39832062E+03 d E =-.629032E-01 mag= 40.9742
N E dE d eps ncg rms rms(c)
DAV: 1 -0.396196868308E+03 0.21214E+01 -0.36850E+01****** 0.633E+01 0.425E+00
RMM: 2 -0.399264067952E+03 -0.30672E+01 -0.28683E+01934037 0.853E+00 0.969E+00
RMM: 3 -0.398659816403E+03 0.60425E+00 -0.72638E-01855373 0.153E+00 0.117E+01
RMM: 4 -0.398394300133E+03 0.26552E+00 -0.15600E-01924152 0.701E-01 0.457E+00
RMM: 5 -0.398348045589E+03 0.46255E-01 -0.30838E-02818418 0.333E-01 0.207E+00
RMM: 6 -0.398341218380E+03 0.68272E-02 -0.49352E-03856075 0.132E-01 0.128E+00
RMM: 7 -0.398339948798E+03 0.12696E-02 -0.19112E-03815468 0.850E-02 0.108E+00
...
RMM: 18 -0.398335088247E+03 0.38213E-07 -0.14989E-06789617 0.276E-03 0.180E-02
RMM: 19 -0.398335087305E+03 0.94168E-06 -0.90711E-07745017 0.221E-03 0.808E-03
RMM: 20 -0.398335087273E+03 0.32334E-07 -0.39363E-07611218 0.141E-03
5 F= -.39833509E+03 E0= -.39833749E+03 d E =-.797222E-01 mag= 40.9736
N E dE d eps ncg rms rms(c)
DAV: 1 -0.398325426305E+03 0.96610E-02 -0.17301E-01****** 0.431E+00 0.268E-01
RMM: 2 -0.398338899015E+03 -0.13473E-01 -0.12732E-01933106 0.567E-01 0.608E-01
...
RMM: 10 -0.398335164350E+03 0.55826E-05 -0.51384E-06857078 0.554E-03 0.304E-02
RMM: 11 -0.398335163329E+03 0.10206E-05 -0.18340E-06869204 0.282E-03 0.234E-02
RMM: 12 -0.398335162622E+03 0.70740E-06 0.22852E-05779758 0.201E-03 0.144E-02
RMM: 13 -0.398335162196E+03 0.42556E-06 -0.29400E-07535760 0.124E-03 0.641E-03
RMM: 14 -0.398335162154E+03 0.42142E-07 -0.10411E-07444326 0.756E-04
6 F= -.39833516E+03 E0= -.39833757E+03 d E =-.797970E-01 mag= 40.9737
N E dE d eps ncg rms rms(c)
DAV: 1 -0.397791062696E+03 0.54410E+00 -0.95731E+00****** 0.326E+01 0.207E+00
RMM: 2 -0.398544036235E+03 -0.75297E+00 -0.71692E+00933071 0.432E+00 0.430E+00
RMM: 3 -0.398430900185E+03 0.11314E+00 -0.20585E-01824524 0.780E-01 0.611E+00
RMM: 4 -0.398366933166E+03 0.63967E-01 -0.34365E-02923179 0.344E-01 0.292E+00
RMM: 5 -0.398346781083E+03 0.20152E-01 -0.64681E-03827977 0.152E-01 0.834E-01
RMM: 6 -0.398346091497E+03 0.68959E-03 -0.10378E-03848606 0.643E-02 0.652E-01
RMM: 7 -0.398345759257E+03 0.33224E-03 -0.62474E-04811048 0.529E-02 0.595E-01
RMM: 8 -0.398345009735E+03 0.74952E-03 -0.35214E-04812534 0.395E-02 0.372E-01
RMM: 9 -0.398344840442E+03 0.16929E-03 -0.19254E-04814588 0.299E-02 0.331E-01
RMM: 10 -0.398344572394E+03 0.26805E-03 -0.17212E-03820569 0.467E-02 0.314E-01
RMM: 11 -0.398344449087E+03 0.12331E-03 -0.18800E-04914436 0.290E-02 0.258E-01
RMM: 12 -0.398344372103E+03 0.76984E-04 -0.82092E-05894368 0.195E-02 0.166E-01
RMM: 13 -0.398344317872E+03 0.54231E-04 -0.38273E-05880732 0.132E-02 0.718E-02
RMM: 14 -0.398344313150E+03 0.47218E-05 -0.90305E-06855332 0.616E-03 0.492E-02
RMM: 15 -0.398344308402E+03 0.47479E-05 -0.37948E-06806025 0.407E-03 0.182E-02
RMM: 16 -0.398344308623E+03 -0.22116E-06 -0.14176E-06733474 0.238E-03 0.174E-02
RMM: 17 -0.398344307861E+03 0.76264E-06 -0.63980E-07691452 0.184E-03 0.897E-03
RMM: 18 -0.398344307799E+03 0.61555E-07 -0.28815E-07570096 0.124E-03
7 F= -.39834431E+03 E0= -.39834672E+03 d E =-.914565E-02 mag= 40.9678
N E dE d eps ncg rms rms(c)
DAV: 1 -0.398271637678E+03 0.72670E-01 -0.12753E+00****** 0.119E+01 0.747E-01
RMM: 2 -0.398372883115E+03 -0.10125E+00 -0.96303E-01932997 0.157E+00 0.160E+00
RMM: 3 -0.398355842961E+03 0.17040E-01 -0.22294E-02818785 0.282E-01 0.215E+00
RMM: 4 -0.398347196184E+03 0.86468E-02 -0.40864E-03923458 0.120E-01 0.914E-01
RMM: 5 -0.398345266685E+03 0.19295E-02 -0.67061E-04836055 0.506E-02 0.301E-01
RMM: 6 -0.398345172034E+03 0.94651E-04 -0.11921E-04855957 0.214E-02 0.230E-01
RMM: 7 -0.398345117771E+03 0.54263E-04 -0.86381E-05811018 0.200E-02 0.205E-01
RMM: 8 -0.398345028301E+03 0.89471E-04 -0.45025E-05813041 0.138E-02 0.130E-01
RMM: 9 -0.398345014271E+03 0.14030E-04 -0.27097E-05815004 0.113E-02 0.128E-01
RMM: 10 -0.398344971748E+03 0.42523E-04 -0.53083E-05824452 0.165E-02 0.105E-01
RMM: 11 -0.398344959560E+03 0.12188E-04 -0.20959E-05918753 0.957E-03 0.843E-02
RMM: 12 -0.398344948771E+03 0.10789E-04 -0.98162E-06893258 0.658E-03 0.420E-02
RMM: 13 -0.398344945279E+03 0.34919E-05 0.31773E-05837163 0.361E-03 0.179E-02
RMM: 14 -0.398344944986E+03 0.29302E-06 -0.73942E-07695253 0.182E-03 0.122E-02
RMM: 15 -0.398344944634E+03 0.35233E-06 -0.36530E-07608439 0.133E-03 0.555E-03
RMM: 16 -0.398344944617E+03 0.16851E-07 -0.12079E-07452575 0.768E-04
8 F= -.39834494E+03 E0= -.39834736E+03 d E =-.978246E-02 mag= 40.9661
N E dE d eps ncg rms rms(c)
DAV: 1 -0.398272732606E+03 0.72212E-01 -0.12790E+00****** 0.120E+01 0.732E-01
RMM: 2 -0.398370290680E+03 -0.97558E-01 -0.96602E-01932949 0.158E+00 0.867E-01
RMM: 3 -0.398364528955E+03 0.57617E-02 -0.22825E-02817275 0.291E-01 0.286E+00
RMM: 4 -0.398348172736E+03 0.16356E-01 -0.45370E-03924011 0.126E-01 0.910E-01
RMM: 5 -0.398346370015E+03 0.18027E-02 -0.65805E-04826326 0.509E-02 0.245E-01
RMM: 6 -0.398346353601E+03 0.16414E-04 -0.11435E-04855524 0.221E-02 0.233E-01
RMM: 7 -0.398346283595E+03 0.70006E-04 -0.78434E-05811099 0.190E-02 0.202E-01
RMM: 8 -0.398346196174E+03 0.87420E-04 -0.34792E-05814395 0.124E-02 0.117E-01
RMM: 9 -0.398346183554E+03 0.12620E-04 -0.34285E-05813993 0.130E-02 0.128E-01
RMM: 10 -0.398346136003E+03 0.47551E-04 -0.54958E-05833969 0.166E-02 0.907E-02
RMM: 11 -0.398346125255E+03 0.10747E-04 -0.16257E-05920002 0.831E-03 0.690E-02
RMM: 12 -0.398346116813E+03 0.84421E-05 -0.52598E-06885554 0.587E-03 0.307E-02
RMM: 13 -0.398346114814E+03 0.19992E-05 -0.20100E-06823912 0.299E-03 0.128E-02
RMM: 14 -0.398346114830E+03 -0.16182E-07 -0.57834E-07676778 0.167E-03
9 F= -.39834611E+03 E0= -.39834853E+03 d E =-.117021E-02 mag= 40.9648
N E dE d eps ncg rms rms(c)
DAV: 1 -0.398340694305E+03 0.54205E-02 -0.95648E-02****** 0.328E+00 0.200E-01
RMM: 2 -0.398347949664E+03 -0.72554E-02 -0.71649E-02932932 0.433E-01 0.254E-01
RMM: 3 -0.398347487282E+03 0.46238E-03 -0.20849E-03818727 0.795E-02 0.761E-01
RMM: 4 -0.398346314325E+03 0.11730E-02 -0.33729E-04923147 0.340E-02 0.239E-01
RMM: 5 -0.398346190287E+03 0.12404E-03 -0.48076E-05830172 0.136E-02 0.668E-02
RMM: 6 -0.398346189382E+03 0.90417E-06 -0.85605E-06855191 0.593E-03 0.647E-02
RMM: 7 -0.398346184300E+03 0.50827E-05 -0.29550E-06807862 0.521E-03 0.562E-02
RMM: 8 -0.398346177447E+03 0.68524E-05 -0.27845E-06801154 0.355E-03 0.322E-02
RMM: 9 -0.398346176670E+03 0.77704E-06 -0.23712E-06811432 0.340E-03 0.350E-02
RMM: 10 -0.398346173220E+03 0.34503E-05 -0.41745E-06830103 0.456E-03 0.248E-02
RMM: 11 -0.398346172423E+03 0.79663E-06 -0.71851E-06843289 0.233E-03 0.188E-02
RMM: 12 -0.398346171819E+03 0.60388E-06 -0.58826E-07700547 0.180E-03 0.849E-03
RMM: 13 -0.398346129853E+03 0.41966E-04 -0.18334E-02464009 0.113E-03 0.334E-03
RMM: 14 -0.398346171175E+03 -0.41322E-04 -0.30098E-04425774 0.666E-04 0.288E-03
RMM: 15 -0.398346171662E+03 -0.48708E-06 -0.23893E-06420760 0.460E-04 0.117E-03
RMM: 16 -0.398346171666E+03 -0.40163E-08 -0.39401E-08416884 0.277E-04
10 F= -.39834617E+03 E0= -.39834859E+03 d E =-.122705E-02 mag= 40.9643
N E dE d eps ncg rms rms(c)
DAV: 1 -0.398336462892E+03 0.97088E-02 -0.17608E-01****** 0.433E+00 0.271E-01
RMM: 2 -0.398349707592E+03 -0.13245E-01 -0.12992E-01932263 0.571E-01 0.408E-01
RMM: 3 -0.398348383064E+03 0.13245E-02 -0.30929E-03820248 0.104E-01 0.932E-01
RMM: 4 -0.398346607561E+03 0.17755E-02 -0.54162E-04925953 0.432E-02 0.306E-01
RMM: 5 -0.398346401469E+03 0.20609E-03 -0.78790E-05829688 0.177E-02 0.968E-02
RMM: 6 -0.398346401574E+03 -0.10518E-06 -0.18408E-05828783 0.867E-03 0.101E-01
RMM: 7 -0.398346390251E+03 0.11323E-04 -0.13774E-05810840 0.857E-03 0.945E-02
RMM: 8 -0.398346368448E+03 0.21803E-04 -0.10105E-05814065 0.668E-03 0.479E-02
RMM: 9 -0.398346367933E+03 0.51421E-06 -0.86089E-06820247 0.618E-03 0.595E-02
RMM: 10 -0.398346358520E+03 0.94135E-05 -0.89282E-06860271 0.643E-03 0.334E-02
RMM: 11 -0.398346356945E+03 0.15748E-05 -0.18809E-06835110 0.280E-03 0.228E-02
RMM: 12 -0.398346356017E+03 0.92783E-06 -0.82045E-07754769 0.205E-03 0.110E-02
RMM: 13 -0.398346355738E+03 0.27896E-06 -0.26411E-07530526 0.116E-03 0.487E-03
RMM: 14 -0.398346355750E+03 -0.11554E-07 -0.13051E-07438486 0.828E-04
11 F= -.39834636E+03 E0= -.39834878E+03 d E =-.184084E-03 mag= 40.9638
N E dE d eps ncg rms rms(c)
DAV: 1 -0.398336968586E+03 0.93872E-02 -0.16848E-01****** 0.424E+00 0.265E-01
RMM: 2 -0.398276839428E+03 0.60129E-01 0.60333E-01932273 0.560E-01 0.376E-01
RMM: 3 -0.398298550297E+03 -0.21711E-01 -0.22013E-01820174 0.103E-01 0.939E-01
RMM: 4 -0.398297164580E+03 0.13857E-02 -0.42745E-03925671 0.427E-02 0.300E-01
RMM: 5 -0.398296996677E+03 0.16790E-03 -0.27557E-04829017 0.174E-02 0.952E-02
RMM: 6 -0.398296999974E+03 -0.32967E-05 -0.50158E-05827185 0.859E-03 0.996E-02
RMM: 7 -0.398296988276E+03 0.11698E-04 -0.15952E-05811015 0.833E-03 0.915E-02
RMM: 8 -0.398296968074E+03 0.20202E-04 -0.95815E-06813908 0.646E-03 0.478E-02
RMM: 9 -0.398296967219E+03 0.85475E-06 -0.83858E-06819832 0.610E-03 0.592E-02
RMM: 10 -0.398296957513E+03 0.97060E-05 -0.83902E-06859985 0.627E-03 0.324E-02
RMM: 11 -0.398296956056E+03 0.14568E-05 -0.22486E-06831960 0.278E-03 0.229E-02
RMM: 12 -0.398296954999E+03 0.10575E-05 -0.16825E-06767724 0.209E-03 0.110E-02
RMM: 13 -0.398296954714E+03 0.28525E-06 -0.26148E-07527980 0.115E-03 0.500E-03
RMM: 14 -0.398296954698E+03 0.15687E-07 -0.13184E-07440344 0.825E-04
12 F= -.39829695E+03 E0= -.39829938E+03 d E =0.492170E-01 mag= 40.9613
N E dE d eps ncg rms rms(c)
DAV: 1 -0.398341748098E+03 -0.44793E-01 -0.53656E-01****** 0.250E+00 0.158E-01
RMM: 2 -0.398271230092E+03 0.70518E-01 0.77140E-01932721 0.338E-01 0.894E-02
RMM: 3 -0.398296831918E+03 -0.25602E-01 -0.28490E-01841878 0.625E-02 0.253E-01
RMM: 4 -0.398297173758E+03 -0.34184E-03 -0.49996E-03927630 0.229E-02 0.881E-02
RMM: 5 -0.398297206463E+03 -0.32705E-04 -0.33708E-04852535 0.703E-03 0.519E-02
RMM: 6 -0.398297236212E+03 -0.29749E-04 -0.74620E-05811016 0.509E-03 0.120E-01
RMM: 7 -0.398297206206E+03 0.30006E-04 -0.51057E-06808664 0.409E-03 0.430E-02
RMM: 8 -0.398297202271E+03 0.39354E-05 0.96763E-05800027 0.399E-03 0.288E-02
RMM: 9 -0.398297200374E+03 0.18970E-05 -0.28684E-06813072 0.394E-03 0.286E-02
RMM: 10 -0.398297198586E+03 0.17874E-05 -0.14444E-06802175 0.258E-03 0.173E-02
RMM: 11 -0.398297198339E+03 0.24762E-06 -0.59817E-07659258 0.166E-03 0.148E-02
RMM: 12 -0.398297197879E+03 0.45987E-06 -0.49998E-07675367 0.153E-03 0.892E-03
RMM: 13 -0.398297197665E+03 0.21377E-06 -0.19297E-07470790 0.103E-03 0.394E-03
RMM: 14 -0.398297197672E+03 -0.71595E-08 -0.79639E-08420470 0.657E-04
13 F= -.39829720E+03 E0= -.39829962E+03 d E =-.242974E-03 mag= 40.9620
N E dE d eps ncg rms rms(c)
DAV: 1 -0.398342032561E+03 -0.44835E-01 -0.44815E-01****** 0.268E-02 0.406E-03
RMM: 2 -0.396754960770E+03 0.15871E+01 0.19726E+01884924 0.641E-02 0.243E-01
RMM: 3 -0.397205851234E+03 -0.45089E+00 -0.41630E+00457703 0.298E-02 0.391E-02
RMM: 4 -0.397212911352E+03 -0.70601E-02 -0.35208E-02443913 0.819E-03 0.294E-02
RMM: 5 -0.397213133991E+03 -0.22264E-03 -0.52271E-03413169 0.171E-03 0.297E-02
RMM: 6 -0.397213207200E+03 -0.73209E-04 -0.71451E-04427860 0.635E-04 0.292E-02
RMM: 7 -0.397213209130E+03 -0.19300E-05 -0.14155E-05407879 0.243E-04 0.291E-02
RMM: 8 -0.397213209467E+03 -0.33676E-06 -0.22333E-06202982 0.119E-04 0.288E-02
RMM: 9 -0.397213209563E+03 -0.96508E-07 -0.29101E-06402454 0.173E-04 0.289E-02
RMM: 10 -0.397213208730E+03 0.83292E-06 0.37658E-04437598 0.473E-04 0.279E-02
RMM: 11 -0.397213208050E+03 0.68001E-06 -0.64043E-07431390 0.391E-04 0.273E-02
RMM: 12 -0.397213205285E+03 0.27654E-05 -0.19518E-07495490 0.778E-04 0.255E-02
RMM: 13 -0.397213203214E+03 0.20709E-05 -0.32574E-07436611 0.574E-04 0.248E-02
RMM: 14 -0.397213197322E+03 0.58916E-05 -0.68484E-07622748 0.133E-03 0.224E-02
RMM: 15 -0.397213194019E+03 0.33026E-05 -0.22441E-06520864 0.870E-04 0.216E-02
RMM: 16 -0.397213191838E+03 0.21817E-05 -0.84843E-08431880 0.562E-04 0.203E-02
RMM: 17 -0.397213190171E+03 0.16663E-05 -0.14261E-07419763 0.364E-04 0.189E-02
RMM: 18 -0.397213146861E+03 0.43311E-04 -0.18881E-02406661 0.321E-04 0.184E-02
RMM: 19 -0.397213187885E+03 -0.41024E-04 -0.42302E-04415229 0.499E-04 0.152E-02
RMM: 20 -0.397213187385E+03 0.49933E-06 -0.10554E-06409548 0.213E-04 0.139E-02
RMM: 21 -0.397213186623E+03 0.76267E-06 -0.31361E-05407942 0.251E-04 0.121E-02
RMM: 22 -0.397213185757E+03 0.86540E-06 -0.12982E-06408745 0.244E-04 0.106E-02
RMM: 23 -0.397213185098E+03 0.65932E-06 -0.17551E-05412822 0.281E-04 0.883E-03
RMM: 24 -0.397213184034E+03 0.10645E-05 -0.25798E-08410441 0.226E-04 0.725E-03
RMM: 25 -0.397213183517E+03 0.51683E-06 -0.90325E-08380310 0.146E-04 0.632E-03
RMM: 26 -0.397213182953E+03 0.56330E-06 -0.15713E-08407969 0.189E-04 0.515E-03
RMM: 27 -0.397213182747E+03 0.20620E-06 -0.11234E-07 95015 0.774E-05 0.477E-03
RMM: 28 -0.397213182531E+03 0.21592E-06 0.58992E-08 98612 0.804E-05 0.432E-03
RMM: 29 -0.397213182251E+03 0.28001E-06 -0.96349E-08311834 0.110E-04 0.377E-03
RMM: 30 -0.397213181864E+03 0.38696E-06 -0.11478E-08372739 0.118E-04 0.288E-03
RMM: 31 -0.397213181687E+03 0.17748E-06 -0.58789E-08 69628 0.639E-05 0.251E-03
RMM: 32 -0.397213181588E+03 0.99186E-07 -0.28383E-08 51596 0.666E-05
14 F= -.39721318E+03 E0= -.39721559E+03 d E =0.108377E+01 mag= 40.9322
-
- Global Moderator
- Posts: 506
- Joined: Mon Nov 04, 2019 12:41 pm
- Contact:
Re: Weird energy jump in relaxation
Hi again,
Actually, I overlooked this before in your INCAR file.
You are using EDIFF=1e-7 which is very stringent especially for a supercell with 72 atoms.
Are you sure you need such tight criteria?
Try running the same calculation with EDIFF=1e-4 and EDIFFG=-1e-2 (the negative number means the convergence criteria is applied in the norm of the forces: wiki/index.php/EDIFFG).
If using a less stringent criterion is not acceptable then I would recommend trying ALGO=Normal.
Actually, I overlooked this before in your INCAR file.
You are using EDIFF=1e-7 which is very stringent especially for a supercell with 72 atoms.
Are you sure you need such tight criteria?
Try running the same calculation with EDIFF=1e-4 and EDIFFG=-1e-2 (the negative number means the convergence criteria is applied in the norm of the forces: wiki/index.php/EDIFFG).
If using a less stringent criterion is not acceptable then I would recommend trying ALGO=Normal.
-
- Newbie
- Posts: 25
- Joined: Tue Jan 07, 2014 11:22 am
Re: Weird energy jump in relaxation
Thanks for the hint; I am currently working on finding out how tight convergence criteria have to be.
However, my issue here is not so much with the jump in the energy during the electronic loop. I actually discussed this here a while ago already:
forum/viewtopic.php?t=18121
My issue here is that after the increase in the energy, the ionic relaxation is stopped instead of continued as I would expect it to be if the energy changes (and increases!) by a significant value.
However, my issue here is not so much with the jump in the energy during the electronic loop. I actually discussed this here a while ago already:
forum/viewtopic.php?t=18121
My issue here is that after the increase in the energy, the ionic relaxation is stopped instead of continued as I would expect it to be if the energy changes (and increases!) by a significant value.
-
- Global Moderator
- Posts: 506
- Joined: Mon Nov 04, 2019 12:41 pm
- Contact:
Re: Weird energy jump in relaxation
I suggest that you first try the two suggestions related to electronic convergence:
1. Use a less stringent criterium for the total energy tolerance
2. or use ALGO=Normal
Currently, you have a problem with the electronic minimization.
When that happens, your total energies, forces, and stresses might not be accurate.
If these are not accurate in two consecutive steps of the ionic relaxation this will 'confuse' the minimization algorithm and it might not be clear what ionic step needs to be taken next.
The ionic minimization algorithm checks whether the next ionic step to be taken is large enough otherwise the ionic minimization stops.
This might be what is happening (although I cannot say for sure).
The calculation you posted is too expensive for me to run locally and check.
I would need a smaller system to try to reproduce the issue.
If you see this problem on a relatively fast calculation I can run it and indicate what is the condition that provoked the ionic minimization to stop.
But that will likely not help you too much.
What is very important when performing the ionic minimization is that the electronic minimization runs without problems and the total energies, forces and stresses are accurate.
1. Use a less stringent criterium for the total energy tolerance
2. or use ALGO=Normal
Currently, you have a problem with the electronic minimization.
When that happens, your total energies, forces, and stresses might not be accurate.
If these are not accurate in two consecutive steps of the ionic relaxation this will 'confuse' the minimization algorithm and it might not be clear what ionic step needs to be taken next.
The ionic minimization algorithm checks whether the next ionic step to be taken is large enough otherwise the ionic minimization stops.
This might be what is happening (although I cannot say for sure).
The calculation you posted is too expensive for me to run locally and check.
I would need a smaller system to try to reproduce the issue.
If you see this problem on a relatively fast calculation I can run it and indicate what is the condition that provoked the ionic minimization to stop.
But that will likely not help you too much.
What is very important when performing the ionic minimization is that the electronic minimization runs without problems and the total energies, forces and stresses are accurate.
-
- Newbie
- Posts: 25
- Joined: Tue Jan 07, 2014 11:22 am
Re: Weird energy jump in relaxation
Thanks for explaining that, I wasn't aware how the ionic and electronic calculations interact.
I have rerun the job with slightly different parameters (changed MAXMIX and AMIX, kept EDIFF at its small value) and no problems.
If I find a small system with similar trouble, I'll post it.
I have rerun the job with slightly different parameters (changed MAXMIX and AMIX, kept EDIFF at its small value) and no problems.
If I find a small system with similar trouble, I'll post it.