computing dispersion coeffs

Find answers or ask questions regarding Dalton calculations.
Please upload an output file showing the problem, if applicable.
(It is not necessary to upload input files, they can be found in the output file.)

Post Reply
Posts: 11
Joined: 28 May 2016, 21:05
First name(s): Amit
Last name(s): Sharma
Affiliation: Wright State University
Country: United States

computing dispersion coeffs

Post by greifsw » 29 May 2016, 14:46

Hi All,
I want to calculate C6 .. C10 coefficients for diatomic molecule and I am learning how to compute these quantities with Dalton. Firstly, can Dalton calculate these coefficients and output final numbers or do one have to perform some post processing?

I am able to calculate Frequency dependent polarizabilities and Cauchy moments from C6 module and the process of obtaining C6 coeff seems to be cauchy moments -> Pade approx -> integrate -> ... Can Dalton do the above internally? The numbers for He..He (just as an example) are well known and published and I am looking to calculate those with Dalton.
The following input

Code: Select all

generates Cauchy moments and RESPONSE.C8 interface file. Is the data written in RESPONSE.C8 file quadrupole-quadrupole coefficient that appear in dispersion energy expression? What does RESPONSE.C8 file interface with?
Can anyone outline the recipe to compute dispersion coefficients with Dalton.

Posts: 299
Joined: 27 Jun 2013, 18:44
First name(s): Hans Jørgen
Middle name(s): Aagaard
Last name(s): Jensen
Affiliation: Universith of Southern Denmark
Country: Denmark

Re: computing dispersion coeffs

Post by hjaaj » 29 May 2016, 20:41

You have to do post processing with another program. This feature was originally introduced in collaboration with Hinne Hettema and Paul Wormer from the Netherlands. They used their own programs for the post processing. The implementation in Dalton and an application is described in
Frequency-dependent polarizabilities of O2 and van der Waals coefficients of dimers containing O2.
H. Hettema, P.E.S. Wormer, P. Jørgensen, H.J. Aa. Jensen, and T. Helgaker.
J. Chem. Phys. 100, 1297-1302 (1994).
-- Hans Jørgen.

Posts: 4
Joined: 04 Jul 2014, 06:29
First name(s): Leonid
Last name(s): Shirkov
Affiliation: AMU
Country: Poland

Re: computing dispersion coeffs

Post by leonid » 10 Jun 2016, 10:19

Dear Colleagues,

I am also interested in calculating dynamic polarizability (plrb) tensor alpha(l,m,l',m') for imaginary frequencies.
I need it for calculating long-range van der Waals dispersion (and induction with multipole moments) coefficients. I have searched the forum and here are some related topics, categorized and with some questions to the Experts and Developers of Dalton. As an example, let's consider the simplest case of atom-atom interaction, Ne dimer. For C6 one would need just alpha(1,0,1,0), dipole-dipole plrb, for C8 both alpha(1,0,1,0) and alpha(2,0,2,0), quadropole-quadropole plrb, and so on.

1. Damped response calculation: *ABSORP
Also discussed here:
In this case .ALPHA is just dipole-dipole plrb and there is now way to find other tensor components.
Also, I did NOT manage to calculate .BETA, dipole-dipole-dipole hyperplrb for imaginary freqs.
It is not needed for disp. coef. at this level of theory, but I was curious. What is wrong in my input? It works well with just .ALPHA. (See attachment 1). As I understand .GAMMA is not yet available.

Do I understand correctly that *ABSORP works only with HF, DFT and MCSCF levels of theory? What about SOPPA(CCSD) with *ABSORP?
It fails (see attachment 2).

2. Cauchy moments with *C6
Keywords - .C6ATM, .C6SPH etc.
Once you have the Cauchy moments, you use eq. (4) from the Ref. given by Hans Jørgen for imaginary
omega, then use Pade aproximants and integrate (calculate Cn disp. coefs).
If I use .C6ATM and .C8ATM then I get the following Cauchy moments
SM01+00 SM01+00 - needed to find dipole-dipole plrb.
SM02+00 SM02+00 - needed to find quadropole-quadropole plrb.
SM02+00 SM01+00 - dipole-quadrupole plrb. (=0 for closed shell atoms)
Using .C10ATM will give higher Cauchy moments SM04+00 SM04+00 that will be needed to find the full C12 coef.

Here are C6 and C8 coefs (a.u.) calculated with hf/avtz:
C6= 5.1675, C8=.. (could NOT get it without dynamic quadrupole-quadrupole plrb)
C6=5.1879, C8=42.7871, C10=367.1082.

With *C6 one can use SOPPA and SOPPA(CCSD).

3. Dispersion coefficients D_AB(n) found with *CCLR and .DISPCF.
Note that these are different dispersion coefs, not Cn (see 31.3 paragraph of 2016 manual). See also discussion here
The C6 value found with this method is 5.8928. It is a bit different from the one found with *C6 and SOPPA(CCSD) - 5.8414. (avtz basis set)
It looks it works only for dipole-dipole plrb, i.e. with

There is some disscussion here about how to find quadropole-quadropole plrb

but I didn't manage to.

One can also use *CCQR and *CCCR for finding beta and gama.

4. There is also module *ABALNR that might be useful.

So, resuming method 2 with Cauchy moments is the only way to find all the dispersion coefs, not only C6.

P.S. I used Dalton 2013.3, but these modules were not changed in 2016 version.
(40.01 KiB) Downloaded 253 times
(39.38 KiB) Downloaded 290 times
Last edited by leonid on 07 Jun 2017, 13:50, edited 2 times in total.

Posts: 6
Joined: 10 May 2016, 11:59
First name(s): Anthony
Middle name(s): J.
Last name(s): Stone
Affiliation: University of Cambridge
Country: United Kingdom

Re: computing dispersion coeffs

Post by ajs1 » 10 Jun 2016, 10:35

The CamCASP program can be used with Dalton to calculate molecular or distributed polarizabilities at imaginary frequency and to obtain dispersion coefficients up to C12. See

Post Reply

Who is online

Users browsing this forum: No registered users and 6 guests