[molpro-user] Differences between 2010.1 and 2012.1 (UKS)

[Javier Cerezo]

Dear Molpro users,

I am comparing the performance of versions 2010.1 and 2012.1 by running 
the same job with both of them. The computation is a {UKS, B3LYP3} (I 
attached the input at the end of the message) which according to what I 
understand from the manual, should be using the same settings (namely, 
the same grid). As expected, the newer version performs better by 
achieving convergence in less SCF cycles and taking less time for each 
iteration. However, the final energy is not the same in both cases, with 
a difference, E(2012.1)-E(2010.1), of 0.0007 hartrees (~0.46kcal/mol). I 
understand that it is an small difference (at least below the "chemical 
accuracy") but I was wondering why both versions did not lead to the 
same results and if there is any change in settings of thresholds that I 
am missing.

Here you have some info about the SCF in both cases.

2010.1.-
Last iteration: 46
DDIF = 0.421D-05
GRAD=0.788D-06
Energy change = 1.44E-6
Energy=-1793.40448762

2012.1.-
Last iteration: 38
DDIF = 0.421D-05
GRAD=0.788D-06
Energy change = 6.80E-7
Energy=-1793.4037537


The same input is used, which follows

***
! Set max memory in the cluster (8GB)
memory,228,m

file,2,wavefunction.wfu

basis={
  default=cc-pVDZ
! LANL2TZ from BSE for Ni
! NICKEL       (5s,5p,6d) -> [5s,5p,4d]
s, NI , 7.6200000, 2.2940000, 0.8760000, 0.1153000, 0.0396000
c, 1.1, 1
c, 2.2, 1
c, 3.3, 1
c, 4.4, 1
c, 5.5, 1
p, NI , 23.6600000, 2.8930000, 0.9435000, 0.0840000, 0.0240000
c, 1.1, 1
c, 2.2, 1
c, 3.3, 1
c, 4.4, 1
c, 5.5, 1
d, NI , 42.7200000, 11.7600000, 3.8170000, 1.1690000, 0.2836000, 0.0898000
c, 1.3, 0.0372699, 0.1956103, 0.4561273
c, 4.4, 1
c, 5.5, 1
c, 6.6, 1
! Effective core Potentials
! -------------------------
ECP, ni, 10, 2 ;
3; !  d-ul potential
1,469.9324331,-10.0000000;
2,85.4236411,-69.4084805;
2,21.2674984,-12.0951020;
4; !  s-ul potential
0,162.1686097,3.0000000;
1,176.5333232,22.0253618;
2,68.9562010,443.0181088;
2,13.5792838,145.5696411;
4; !  p-ul potential
0,69.0181506,5.0000000;
1,275.5955596,4.9882824;
2,47.1315453,256.6945853;
2,12.9874075,78.4754450;
}

! Read geom from XYZ
geomtyp=xyz
geometry=lu167s_T_25.xyz

!Specify triplet (spin=2)
wf,spin=2,charge=0

!Activate direct methods
gdirect

! Computaion of DFT energy
{UKS,B3LYP3}
---

Thanks!

Javier


-- 
Javier CEREZO BASTIDA
PhD Student
Physical Chemistry
Universidad de Murcia
Murcia (Spain)
Tel: (+34)868887434