Molcas Forum

JCC225

Member

Registered: 2021-11-30

Posts: 2

[SOLVED] Dummy atom masses

Hi,

I've been trying to calculate a Hessian for a molecule with a Dummy atom. As I expect that the potential energy surface will only be weakly dependent on the location of that atom, I thought I could try just increasing the mass of that atom to effectively freeze it in place.

I attempted this by using the ISOTOPES feature of GATEWAY - here is a minimal representation of what I'm trying to do, using water as an example.

&GATEWAY  
basis set 
O.ano-s-MB
O               -0.135780       -0.511056        0.000000
end of basis
basis set
H.ano-s-MB
H1               1.796958       -0.119339        0.000000
H2              -1.061179        1.230394        0.000000
end of basis
basis set
X...1s / Inline
0.000 0
1 1 
1.00000
1.0
X               1.000000        0.0000000       0.0000000
end of basis

****************

Isotopes=1
4 999.0 Dalton

****************

ricd

&SEWARD

&SCF

&McKinley

This attempts to perform a numerical hessian on water with SCF, using a dummy atom X with mass 999 Dalton. The hessian has extremely large components for the X motion in the first IR frequency, as I assume the default mass of an X atom is 0.

However, in the second section where it calculates the isotopic substitutions, I get this specific error

 Masses:
 =======
  16   1   3 999


 Frequencies:
 ============
                         1         2         3         4         5         6

            Freq.    i111.88     i0.00     i0.00      0.00      0.00      0.00                                               


                         7         8         9        10        11        12

            Freq.       0.00    341.25    804.49   1414.08   2211.33   3206.06                                               


***
*** iNuclearChargeFromSymbol: error
***    unknown atom:  X
***
--- Stop Module: slapaf at Tue Nov 30 09:58:18 2021 /rc=-1 ---

I've had a look through the source material and I understand the error, I'm just unsure how to proceed on this. Could anyone help?

I thought I might be able to utilise the `MASS' keyword in MCLR? This is running in OpenMolcas v. 21.06 if that is useful.

Thanks in advance,
Joe

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,085

Re: [SOLVED] Dummy atom masses

The hessian has extremely large components for the X motion in the first IR frequency, as I assume the default mass of an X atom is 0.

The "default" mass should have no influence on the frequencies. This block should be using your input mass.

I've had a look through the source material and I understand the error

You probably don't, because the message is misleading. That should be a warning, not an error, as the code proceeds happily after that. It fails later with no specific message due to out-of-bounds array access (because the atomic number of X is 0, and not every array expects that).

I've posted a patch in https://gitlab.com/Molcas/OpenMolcas/-/ … quests/469, it should land in the master branch shortly.

JCC225

Member

Registered: 2021-11-30

Posts: 2

Re: [SOLVED] Dummy atom masses

Hi Ignacio,

Ahh, Okay, that makes more sense. Thank you for the speedy response and fix, I really appreciate it.

Thanks,
Joe