Page 1 of 1

DFD3BJ parameters

Posted: 28 Nov 2018, 00:09
by esmuigors
Dear Dalton people,

I am confused about how to perform calculations with CAM-B3LYP and DFT-D3(BJ) correction. As I got the following error in SIRIUS:

@ Reason: ERROR: DFT-D3 Func. Dep. Paras. Unknown

I actually know the parameters (awkwardly, from Gaussian Reference), but I have no idea how to supply them to the program. There is the keyword .D3PAR, as described in the Manual:
.D3PAR READ (LUINP,*) D3_s6_inp, D3_alp_inp, D3_rs6_inp, D3_rs_18_inp, D3_s18_inp
  • keyword for specifying the s6, α, sr,6, sr,8 and s8 parameters of the DFT-D3 methods.
    Note, take care to match the parameter values to the correct version of the DFT-D3
    correction.
But these are obviously D3-ZERO coefficients (for zero dumping in the first range). I need s6, s8, a1 and a2, as they are labeled in Grimme et al.'s article! How to specify them in DALTON?

Many thanks in advance. Output file is attached.

Re: DFD3BJ parameters

Posted: 29 Nov 2018, 16:59
by a_teale
Hi. Sorry the manual entry is not complete / clear. I'll try to update this.

The .D3PAR input line can be used, the parameters for the DFT-D3(BJ) run need to be in the order

s_6, alpha, a_1, s_8, a_2

Regarding the parameters for this functional, if you want to post the reference or values here I will update the internal lists so that in a future release they don't need to be manually entered.

Re: DFD3BJ parameters

Posted: 30 Nov 2018, 20:13
by esmuigors
Thank You, prof. Teale!

The values are here:
https://www.chemie.uni-bonn.de/pctc/mul ... ctionalsbj

On the group homepage, there is also manual for DFT-D3, and it mentions that the Sherrill's group has issues their refitted version for the correction,[1] and in case of BJ-damping only the coefficients are changed. The coefficients for all functionals are available also separately in some Fortran file which apparently cannot be navigated without seeing the rest of the program (just an array of numbers).

[1] Smith, D. G. A.; Burns, L. A.; Patkowski, K.; Sherrill, C. D. J. Phys. Chem. Lett. 2016, 7, 2197.

With best regards,
Igors

Re: DFD3BJ parameters

Posted: 30 Nov 2018, 20:17
by esmuigors
One more question…

There is no such parameter α (alpha) in D3(BJ). Should I put some random number in place of it? Because obviously I cannot just leave it out of the list…

Re: DFD3BJ parameters

Posted: 08 Dec 2018, 18:28
by a_teale
You are correct that the read in here requires something to be set for alpha, but it is indeed not relevant for the D3(BJ) run. I will fix this so we have separate keywords to read in the parameters for each model. I'll update the manual accordingly too. In the meantime you can just set an arbitrary value. Thanks for pointing out the issues here.