Total energy difference between VASP5.2.12 et VASP5.3.3

[mrubes]

Dear Vasp users

I have a problem with total energy difference between versions of VASP(5.2.12) and VASP(5.3.3) on a complex containgin Cu2+ and two COO- groups . All input files are the same and 5.2.12 yields F=-.69351994E+02 and 5.3.3 version F= -.66206113E+02. I am using ISMEAR=1 and Sigma=0.1 in the INCAR file. However the band occupancies in 5.3.3 version are wierd (seem to be partially populatig 4s orbital on Cu). Thus what changed between versions. Also calculating only benzene molecule yielded same total energies between version, thus the problem sems to be connected with dublet state of Cu2+.

Thank you for your comments and thougts on this.

[struckru]

I also have the same problem. The total energy difference in my case between version of vasp 5.2.12 and vasp 5.3.3.18 is about 8.4 eV for NiMnSb cell containing 12 atoms. All the input files are exactly same too.

[admin]

Changes implemented in vasp.5.3.3 are described in www.vasp.at.
There is no reason for large differences in total energies by 5.3.3 and 5.2.12 version.

[mrubes]

Ok that's what I would think that there is no reason for that but me and strucru got different results between versions (5.2.12 and 5.3.3) for different systems. What puzzles me that I got the same total energies between versions for example for water and benzene crystal, but different for example for CuBTC mof and probably for other metals as strucru. Thus any ideal where shall I look to fix the problem.

[mrubes]

To further simlify my problem I just run simply Cu and O atom in (20x20x20) cubic cell. The numbers in 5.3.3 version are complete nonsense for Cu with either new or older potcar files. For O atom same results for both VASP versions. The problem seem to be connected with Cu POTCAR files. Any ideas?

[struckru]

of course I am aware of the changes of the new version comparing to the old one at the VASP webpage. That is why very strange that there does exist a big difference in total energy in my case (NiMnSb), although all the input files are exactly same including INCAR, POTCAR, POSCAR, and KPOINTS for the calculation with VASP 5.2.12 and VASP 5.3.3.18 in the following:

INCAR:
--------------
SYSTEM = bluk NiMnSb PAW-PBE
ISTART = 0
ICHARG = 2
ISPIN = 2
ISMEAR = -5
SIGMA = 0.2
ENCUT = 450
PREC = Accurate
NSW = 0
IBRION = -1
ISIF = 2
EDIFF = 1E-04
ALGO = Fast
LREAL = .FALSE.
-------------------

POSCAR:
------------------
NiMnSb bulk
5.90699
1.0000000000 0.0000000000 0.0000000000
0.0000000000 1.0000000000 0.0000000000
0.0000000000 0.0000000000 1.0000000000
4 4 4
Direct
0.000000000 0.000000000 0.000000000
0.000000000 0.500000000 0.500000000
0.500000000 0.000000000 0.500000000
0.500000000 0.500000000 0.000000000
0.250000000 0.250000000 0.250000000
0.750000000 0.750000000 0.250000000
0.750000000 0.250000000 0.750000000
0.250000000 0.750000000 0.750000000
0.750000000 0.750000000 0.750000000
0.250000000 0.250000000 0.750000000
0.250000000 0.750000000 0.250000000
0.750000000 0.250000000 0.250000000
----------------------

KPOINTS:
-----------------
K-Points
0
Monkhorst Pack
5 5 5
0 0 0
------------------

the exactly same POTCAR is chosen for these 2 versions (vasp 5.2.12 & 5.3.3.18), the resulting total energies are as follows:

for vasp 5.2.12:
-------------------
N E dE d eps ncg rms rms(c)
DAV: 1 0.658345367286E+03 0.65835E+03 -0.31130E+04 2320 0.145E+03
DAV: 2 -0.120656567762E+02 -0.67041E+03 -0.64905E+03 2320 0.385E+02
DAV: 3 -0.723176637462E+02 -0.60252E+02 -0.54311E+02 2674 0.144E+02
DAV: 4 -0.773985488885E+02 -0.50809E+01 -0.46722E+01 3596 0.381E+01
DAV: 5 -0.776050822040E+02 -0.20653E+00 -0.19697E+00 3426 0.775E+00 0.278E+01
RMM: 6 -0.813402470224E+02 -0.37352E+01 -0.13706E+02 2356 0.874E+01 0.174E+01
RMM: 7 -0.765136977204E+02 0.48265E+01 -0.18538E+01 2323 0.394E+01 0.122E+01
RMM: 8 -0.767111512430E+02 -0.19745E+00 -0.43568E+00 2482 0.112E+01 0.754E+00
RMM: 9 -0.768457804865E+02 -0.13463E+00 -0.21370E+00 2715 0.656E+00 0.627E+00
RMM: 10 -0.770295144022E+02 -0.18373E+00 -0.19927E-01 2590 0.250E+00 0.127E+00
RMM: 11 -0.770121340708E+02 0.17380E-01 -0.11362E-01 2827 0.132E+00 0.107E+00
RMM: 12 -0.770135266675E+02 -0.13926E-02 -0.17106E-02 2834 0.773E-01 0.464E-01
RMM: 13 -0.770140684964E+02 -0.54183E-03 -0.29962E-03 2666 0.344E-01 0.119E-01
RMM: 14 -0.770139980471E+02 0.70449E-04 -0.60870E-04 2271 0.137E-01
1 F= -.77013998E+02 E0= -.77013998E+02 d E =0.000000E+00 mag= 16.0094
----------------------------------

for vasp 5.3.3.18:
----------------------------------
N E dE d eps ncg rms rms(c)
DAV: 1 0.668601467897E+03 0.66860E+03 -0.30980E+04 2320 0.145E+03
DAV: 2 -0.619827625737E+01 -0.67480E+03 -0.65323E+03 2320 0.386E+02
DAV: 3 -0.683490300356E+02 -0.62151E+02 -0.55987E+02 2756 0.146E+02
DAV: 4 -0.736465675782E+02 -0.52975E+01 -0.48490E+01 3632 0.400E+01
DAV: 5 -0.738599508874E+02 -0.21338E+00 -0.20705E+00 3494 0.812E+00 0.335E+01
RMM: 6 -0.760504360618E+02 -0.21905E+01 -0.18115E+02 2381 0.998E+01 0.370E+01
RMM: 7 -0.702844197092E+02 0.57660E+01 -0.46987E+01 2342 0.452E+01 0.298E+01
RMM: 8 -0.690995549784E+02 0.11849E+01 -0.11544E+01 2332 0.242E+01 0.209E+01
RMM: 9 -0.689187486396E+02 0.18081E+00 -0.28673E+00 2638 0.965E+00 0.894E+00
RMM: 10 -0.684954362641E+02 0.42331E+00 -0.17707E+00 2640 0.586E+00 0.344E+00
RMM: 11 -0.684738678174E+02 0.21568E-01 -0.86346E-01 2645 0.522E+00 0.191E+00
RMM: 12 -0.685134369680E+02 -0.39569E-01 -0.19727E-01 3116 0.158E+00 0.112E+00
RMM: 13 -0.684969629995E+02 0.16474E-01 -0.30791E-02 2609 0.128E+00 0.461E-01
RMM: 14 -0.684935621991E+02 0.34008E-02 -0.90311E-03 3165 0.347E-01 0.173E-01
RMM: 15 -0.684943724273E+02 -0.81023E-03 -0.68507E-04 2371 0.120E-01 0.609E-02
RMM: 16 -0.684952744046E+02 -0.90198E-03 -0.19722E-04 1645 0.943E-02 0.427E-02
RMM: 17 -0.684956635319E+02 -0.38913E-03 -0.72346E-05 1478 0.679E-02 0.310E-02
RMM: 18 -0.684960147029E+02 -0.35117E-03 -0.52423E-05 1443 0.528E-02 0.206E-02
RMM: 19 -0.684959281074E+02 0.86596E-04 -0.25384E-05 1411 0.465E-02
1 F= -.68495928E+02 E0= -.68495928E+02 d E =0.000000E+00 mag= 15.9997
---------------------------------

As you can see, there is a big difference in total energy even with the same POTCAR. So I think POTCAR is not the problem. As for what causes the problem which seems only happen to metal system, any ideas?

[admin]

NiMnSb, PAW-PBE, 5x5x5
vasp.5.2.12: 1 F= -.77456625E+02 E0= -.77456625E+02 d E =0.000000E+00 mag= 15.9961
vasp.5.3.3: 1 F= -.77456625E+02 E0= -.77456625E+02 d E =0.000000E+00 mag= 15.9961

[mrubes]

Ok thus the results admin posts clearly indicates compilation issues. Thanks

[mrubes]

Issue resolved after updating to the latest version of ifc. The problem was with our ifc(2011.5.220) version, which when combined with openmpi lead to this behavior. The single-proccesor version yielded correct results even with ifc(2011.5.220).