Author Topic: Parallel run for TDDFT with COSMO  (Read 9614 times)

asaduzza

  • Newbie
  • *
  • Posts: 1
  • Karma: +0/-0
Parallel run for TDDFT with COSMO
« on: September 09, 2010, 05:57:21 PM »
Hi
We have recently bought the licence of TURBOMOLE (version 6.2) and tried to calculated the excitation energy including the COSMO. While trying to run the TDDFT calculation including solvation in a 4 processors (MPI), I got the following message that stops the run

 
Quote
MODTRACE: no modules on stack

 COSMO escf implementation under construction!
 escf ended abnormally

It seems from the message that COSMO is not implemented in the escf calculation although in the forum it is said the opposite. So far I remember, only hybrid functional calculations with COSMO is not parallelized. I, however, used PBE functional with RI. Any help would be much appreciated.

christof.haettig

  • Global Moderator
  • Sr. Member
  • *****
  • Posts: 291
  • Karma: +0/-0
    • Hattig's Group at the RUB
Re: Parallel run for TDDFT with COSMO
« Reply #1 on: November 18, 2010, 06:10:16 PM »
Up to now there is no COSMO for TDDFT  in TURBOMOLE (only for DFT ground states).

Where in the forum did you find a message telling that COSMO could be used for TDDFT?

Christof

mpjohans

  • Full Member
  • ***
  • Posts: 26
  • Karma: +0/-0
    • .
Re: Parallel run for TDDFT with COSMO
« Reply #2 on: November 26, 2010, 01:28:05 PM »
There was one thread on the forum were it was claimed that COSMO can be used for TDDFT, which I just replied to. I know from experience that this is a source of confusion for many users, as TM doesn't (didn't use to) complain about trying to do COSMO TDDFT, it just goes on and computes normal TDDFT on top of COSMO orbitals.

Now, if this is "better than nothing" or a random number generator, I don't know, and a comment from someone who does know, would be most appreciated!

Arnim

  • Developers
  • Sr. Member
  • *
  • Posts: 253
  • Karma: +0/-0
Re: Parallel run for TDDFT with COSMO
« Reply #3 on: November 26, 2010, 04:13:27 PM »
Hi,

escf could be used in older versions with COSMO, because the keyword
was simply ignored. In this case, TDDFT is calculated on COSMO MOs, this
is the so-called slow term of the whole TDDFT/COSMO procedure. The
fast term was missing, but will be available (hopefully) soon.

In order to get a feeling about the order of magnitude we are dealing with,
here are some numbers calculated with a development version
of escf which includes the fast term:

The numbers are the 1. excitation energies of phenol (BP/SV(P) gas phase geometry)
calculated with BP/def2-SVPD (expt. value 4.51 eV).
A. = escf on gas phase MOs.
B. = escf on COSMO MOs.
C. = escf on COSMO MOs and TDDFT/COSMO fast term

#  excitation energy / eV  oscillator strength (length rep.)
A.: 0.47374991993630E+01   0.26480961196510E-01
B.: 0.47581682783343E+01   0.25674524688493E-01
C.: 0.47401835183371E+01   0.35611443906804E-01

There is just not much of a shift. There might be cases where the
effect is larger, but in practice, the largest effect will occur by going
from gas phase geometries to COSMO geometries.

Cheers,
Arnim
« Last Edit: November 26, 2010, 08:20:39 PM by uwe »

mpjohans

  • Full Member
  • ***
  • Posts: 26
  • Karma: +0/-0
    • .
Re: Parallel run for TDDFT with COSMO
« Reply #4 on: November 26, 2010, 04:33:59 PM »
Hello Arnim,

Thanks for that clarification! There is a notable difference in the intensity (as perhaps expected) with proper COSMO/TDDFT though.

    Mikael J.