Molcas Forum

felixp

Member

Registered: 2015-11-06

Posts: 27

Website

MS-CASPT2 job farm

Hi, I tried to use the MS-CASPT2 job farm utility but it seems to just do normal MS-CASPT2. Below, I am sending the initial RASSCF job and one of the MS-CASPT2 jobs.

If I look at the code, it seems like the OnlyRoot variable is read from input but never actually used in the calculation:

git grep -i onlyroot
src/caspt2/readin_caspt2.f:      Integer :: OnlyRoot = 0
src/caspt2/readin_caspt2.f:      Read(Line,*,Err=9920,End=9920) Input % OnlyRoot

----

CASSCF input file:

**     MOLCAS.input generated by molcas_input.py Version 1.0
&SEWARD &END
  Title=Re-Cl
  MEDIum Cholesky
  R02O
  AMFI
  Basis Set
  Re.ano-rcc.Roos.24s21p15d11f4g2h.8s7p5d2f1g.
  Re1   0.894823041  -0.052419202   0.000000000  \Angstrom
  End Of Basis
  Basis Set
  Cl.ano-rcc.Roos.17s12p5d4f2g.4s3p1d.
  Cl1  0.509433047   2.478712142    0.000000000  \Angstrom
  End Of Basis
  Basis Set
  O.ano-rcc.Roos.14s9p4d3f2g.3s2p1d. 
  O1   1.176732428  -3.127264736    0.000000000  \Angstrom
  O2   3.066716767   0.208111583    2.195547625  \Angstrom
  O3   3.066716767   0.208111583   -2.195547625  \Angstrom
  End Of Basis
  Basis Set
  N.ano-rcc.Roos.14s9p4d3f2g.3s2p1d.
  N1   -0.878232683  -0.094185530   1.337308555  \Angstrom
  N2   -0.878232683  -0.094185529  -1.337308555  \Angstrom
  End Of Basis
  Basis set
  C.ano-rcc.Roos.14s9p4d3f2g.3s2p1d.
  C1   -3.275102535  -0.044246517   1.502075308  \Angstrom
  C2   -3.196554867  -0.052098836   2.892631590  \Angstrom
  C3   -2.100549080  -0.066124547   0.740178534  \Angstrom
  C4   -1.939619874  -0.083560540   3.496676002  \Angstrom
  C5   -0.810559610  -0.103191867   2.683418959  \Angstrom
  C6   -2.100549080  -0.066124547  -0.740178534  \Angstrom
  C7   -3.275102535  -0.044246517  -1.502075308  \Angstrom
  C8   -0.810559610  -0.103191867  -2.683418959  \Angstrom
  C9   -3.196554867  -0.052098836  -2.892631590  \Angstrom
  C10  -1.939619874  -0.083560540  -3.496676002  \Angstrom
  C11   2.265516626   0.103478225   1.361643884  \Angstrom
  C12   1.070050144  -1.967896832   0.000000000  \Angstrom
  C13   2.265516626   0.103478225  -1.361643884  \Angstrom
  End Of Basis
  Basis Set
  H.ano-rcc.Widmark.8s4p3d1f.2s.
  H1   -4.106717658  -0.034368719   3.493526714  \Angstrom
  H2   -1.823888852  -0.091815265   4.580355943  \Angstrom
  H3    0.188924542  -0.125931304   3.115733939  \Angstrom
  H4   -4.248406741  -0.020601489   1.016325273  \Angstrom
  H5   -4.248406741  -0.020601490  -1.016325273  \Angstrom
  H6    0.188924542  -0.125931304  -3.115733938  \Angstrom
  H7   -4.106717658  -0.034368720  -3.493526714  \Angstrom
  H8   -1.823888852  -0.091815265  -4.580355943  \Angstrom
  End Of Basis
  End Of Input

** For DKH integrals or with ions, MOLCAS SCF seems to not work properly.
*&SCF

**     ================ Singlet states ================

**     Uncomment the following line in order to restart the orbitals:
*      >> LINK FORCE SingletOrbitals.RasOrb INPORB

&RASSCF
SPIN   = 1
NACTEL = 12,0,0
INACT  = 102
RAS2   = 12
CIROOT = 19,19,1
**     Uncomment the following line in order to restart the orbitals:
LUMORB
**     Uncomment the following lines in order to change the orbital order:
*ALTER
*1
*1 1 2

>> SAVE $Project.rasscf.molden SingletOrbitals.molden
>> SAVE $Project.RasOrb SingletOrbitals.RasOrb

>> SAVE $Project.JobIph JOBIPH

----

MS-CASPT2 input file:

&CASPT2
FROZEN     = 54
SHIFT      = 0.0
IMAGINARY  = 0.2
IPEASHIFT  = 0.25
MAXITER    = 120
* If using MOLCAS v>=8.1, uncomment the following line to get CASPT2 properties (dipole moments):
*PROP
MULTISTATE = 19 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
ONLY root = 3

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,085

Re: MS-CASPT2 job farm

It looks like someone forgot to set "NLYROOT", and of course there's no test for this feature.

Can you file an "issue" report in GitLab?

felixp

Member

Registered: 2015-11-06

Posts: 27

Website

Re: MS-CASPT2 job farm

Sorry for the late reply. I filed the "issue" and I also wrote an email to Luca De Vico. He agrees that something looks wrong and wants to look at it.