Molcas Forum

federico2099

Member

Registered: 2021-07-20

Posts: 11

restart Hessian calculations with XMS-CASPT2

Dear community,
is it possible to restart a Hessian calculation performed at the XMS-CASPT2 level?
I have access to all the files in the cwd/ directory
This is the input file:

&GATEWAY
 coord=geom.xyz
 basis=ANO-S-VDZP
 RICD
 Group=Nosymm

>>> export MOLCAS_MAXITER=200

&SEWARD

&RASSCF
 FileOrb=molcas.Pt2Orb
 Spin=1
 Charge=0
 Nactel=10
 Inactive = 57
 Ras2=10
 CIRoot=3 3 1
 Rlxroot=2

&CASPT2
 XMultistate = all
 MAXiter= 300
 RLXROOT = 2
 GRDT
 CORT

 &ALASKA
 ROOT = 2

&MCKINLEY
 Perturbation=Hessian
 ShowHessian

I appreciate any help you could give me as this calculation takes roughly a month to finish

Best regards
Federico

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,079

Re: restart Hessian calculations with XMS-CASPT2

I believe if you just comment out the GATEWAY input, it could work. But better try it out with a smaller/faster calculation and make a backup.

federico2099

Member

Registered: 2021-07-20

Posts: 11

Re: restart Hessian calculations with XMS-CASPT2

Dear Ignacio,
thanks for your response. I have tried your suggestion and also including the JobOld file as below, but the calculation starts from the scratch:
Accumulate the gradient for selected numerical differentiation.
     1 of   193

and when it stopped due to wall time it was in 177 of 193

>> COPY $CurrDir/cwd/molcas.JobIph molcas.JobOld
>>> export MOLCAS_MAXITER=200

&RASSCF
 JobIph
 Spin=1
 Charge=0
 Nactel=10
 Inactive = 57
 Ras2=10
 CIRoot=3 3 1
 Rlxroot=2

&CASPT2
 XMultistate = all
 MAXiter= 300
 RLXROOT = 2
 GRDT
 CORT

 &ALASKA
 ROOT = 2

&MCKINLEY
 Perturbation=Hessian
 ShowHessian 

I've checked the manual of the version I am using (24.02) but couldn't find any alternative.

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,079

Re: restart Hessian calculations with XMS-CASPT2

Well, you also need the files in the scratch directory, in particular the RunFile. I see you removed SEWARD. That will never work, because you need to compute the integrals for every displacement.