Author Topic: Frequencies calculation with frozen coordinates  (Read 11633 times)

gojimeo

  • Newbie
  • *
  • Posts: 9
  • Karma: +0/-0
Frequencies calculation with frozen coordinates
« on: December 05, 2007, 02:07:21 PM »
Hi all,

I have been trying to calculate analytical frequencies (with aoforce) of a large system in which there are a lot of frozen cartesian coordinates. Whereas the statpt nicely carried out the optimization in internal redundant coordinates but maintaining the selected set of cartesians frozen, the aoforce module included all the atoms in the calculation of the hessian. This resulted in a huge time-consuming task and 85 negative frequencies (!) in the output. I wonder if there was any option to exclude the frozen atoms from the hessian calculation but keeping the whole system in the wavefunction in order to be consistent with the optimization step. I have searched throught the manual , but I have not found any answer.

Thank you very much in advance,

Gonzalo

christof.haettig

  • Global Moderator
  • Sr. Member
  • *****
  • Posts: 291
  • Karma: +0/-0
    • Hattig's Group at the RUB
Re: Frequencies calculation with frozen coordinates
« Reply #1 on: January 02, 2008, 11:29:05 AM »
A frozen (fixed) atoms approximation for vibrational frequencies is presently only implemented in NumForce: If given the additional option -frznuclei, it will recognize the coordinates of the fixed atoms and exclude them from the force constant calculation. For the frequency calculation these coordinates will be decoupled from the remaining coordinates.

Christof

gojimeo

  • Newbie
  • *
  • Posts: 9
  • Karma: +0/-0
Re: Frequencies calculation with frozen coordinates
« Reply #2 on: January 03, 2008, 02:12:35 PM »
Hi,

OK, thank you very much for the reply. I realize that numerical frequencies are the only alternative. I will try it despite the high computational cost. Maybe this issue could be considered, if possible, for the next Turbomole release (analytic frequencies for partially frozen systems). It is quite usual to freeze part of a big system, for which Turbomole is fantastic and superior to any other  alternative (for me).
Regards and happy new year,

Gonzalo

gojimeo

  • Newbie
  • *
  • Posts: 9
  • Karma: +0/-0
Re: Frequencies calculation with frozen coordinates
« Reply #3 on: February 28, 2008, 10:46:23 PM »
Hi,

I just wonder if Turbomole 5.10 (it will be shipped to our lab very soon  :D) is able to perform a frozen (fixed) atoms approximation for vibrational frequencies or still not. I have been doing a lot of hard frequency calculations with NumForce (about a week each in my machines... it's just the only way for me to do these calculations... :() and I am really interested on analytical frequencies to gain some (precious) time.
Anyway, thanks for your answers. My best regards,

Gonzalo

christof.haettig

  • Global Moderator
  • Sr. Member
  • *****
  • Posts: 291
  • Karma: +0/-0
    • Hattig's Group at the RUB
Re: Frequencies calculation with frozen coordinates
« Reply #4 on: March 04, 2008, 10:06:09 AM »
No,

there are presently no plans to implement this.

Christof

mpjohans

  • Full Member
  • ***
  • Posts: 26
  • Karma: +0/-0
    • .
Re: Frequencies calculation with frozen coordinates
« Reply #5 on: September 01, 2016, 03:02:37 PM »
Bringing up old topic just because this became personally relevant, and perhaps useful for someone else as well.

You can do an aoforce calculation simulating the frozen state of the atoms by setting their mass to "infinite". So before starting aoforce, edit the control file and add very heavy masses to the frozen atoms. For example, assuming carbons 1 and 2 were frozen during optimisation:

Code: [Select]
c  1-2                                                                         \
   basis =c def2-SVP                                                           \
   jbas  =c def2-SVP                                                           \
   mass  =99999999999.9
c  3-6                                                                         \
   basis =c def2-SVP                                                           \
   jbas  =c def2-SVP

You'll get the same results as with NumForce and -frznuclei, ignoring numerical noise. The inactive modes will have (almost) zero frequencies. NumForce with -frznuclei sometimes produces some unphysical very high frequencies, these will also be "zero". The total number of vibrations, including translation and rotation, will be 3N.