Molcas Forum

sapi

Member

Registered: 2017-02-07

Posts: 2

casscf hessian on root 2

Dear Molcas Team,

I would like to get the hessian matrix for root 2 using &rasscf. I'm aware that this is possible only for single state wavefunction.
Anyway using this input I can't get anything:

 &RASSCF
 inactive=55
 nactel=12 0 0 
 ras2=11
 jobiph
 cirestart
 ciroot=1 2 ; 2 
 &MCKINLEY
 perturbation=hessian
 verbose

I think there is something wrong in the second root single state computation, since I already got the hessian for the first root through this input:

 &RASSCF
 inactive=55
 nactel=12 0 0 
 ras2=11
 jobiph
 &MCKINLEY
 perturbation=hessian
 verbose

With regards

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,085

Re: casscf hessian on root 2

What do you mean you can't get anything? Do you get an error?

sapi

Member

Registered: 2017-02-07

Posts: 2

Re: casscf hessian on root 2

For this input:

&RASSCF
 inactive=55
 nactel=12 0 0 
 ras2=11
 jobiph
 cirestart
 ciroot=1 2 ; 2 
 &MCKINLEY
 perturbation=hessian
 verbose

I get this kind of output:

--- Stop Module:  rasscf at Tue Feb  7 11:36:05 2017 /rc=0 ---
--- Module rasscf spent 10 minutes and 42 seconds
***
--- Start Module: mckinley at Tue Feb  7 11:36:05 2017

()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()

                                             &MCKINLEY

                                   only a single process is used
                       available to each process: 1.8 GB of memory, 1 thread
()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()

--- Stop Module:  mckinley at Tue Feb  7 11:36:05 2017 /rc= _INVOKED_OTHER_MODULE_ ---
***
--- Start Module: slapaf at Tue Feb  7 11:36:05 2017

()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()

                                              &SLAPAF

                                   only a single process is used
                       available to each process: 1.8 GB of memory, 1 thread
()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()


  Slapaf requests the computation of gradients first!
 Root:                     2

--- Stop Module:  slapaf at Tue Feb  7 11:36:05 2017 /rc= _INVOKED_OTHER_MODULE_ ---
***
--- Start Module: alaska at Tue Feb  7 11:36:05 2017

()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()

                                              &ALASKA

                                   only a single process is used
                       available to each process: 1.8 GB of memory, 1 thread
()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()

The &SLAPAF-&ALASKA sequence goes on for long time with no results, because &ALASKA computes always the same gradient (in about the same time) and &SLAPAF never starts really because "requests the computation of gradients first!". And the weird thing is that both &SLAPAF &ALASKA are absent in the input! Moreover, after &MCKINLEY, &MCLR does not starts even after several hours of &SLAPAF-&ALASKA sequence. After 4-5h I interrupted it.
I hope I have been clearer.

With regards

Last edited by sapi (2017-02-08 22:50:51)

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,085

Re: casscf hessian on root 2

Ah... so it seems you are using a development version of molcas, where this bug was inded fixed about two weeks ago. Solution: use version 8.0 or the latest development version.

As for why you see SLAPAF and ALASKA in the output, that's because a semi-numerical differentiation is invoked when you are not computing a state-specific ground state, so that's normal (but slow).