[CP2K-user] [CP2K:22078] DFTB3 D3(BJ) GAMMA

[Dobromir A Kalchevski]

Hello,

many thanx for the answers.

Yes, I found the DFTB3 examples in the test files and I tried those inputs, 
but to no avail. I played around with some of the settings and I think I 
found something that works, but I'm no sure it gives the right barrier.

With the settings below (the rest of the input is the same, as the uploaded 
in the first post) I get broken calculations - either the reagent breaks or 
different H atoms (one from the MM region, one from the QM) get way too 
close and the calculation stops:

DFTB3 + D3(BJ) with default D3BJ_SCALING from the test files, no 
HB_SR_GAMMA (as per the test files) and no EWALD (as per the test files) - 
eventually the reagent (QM region) breaks in pieces:
      &DFTB
        DIAGONAL_DFTB3 T
        !ORTHOGONAL_BASIS F
        SELF_CONSISTENT T
        DO_EWALD F
        DISPERSION T
        !HB_SR_GAMMA T
        &PARAMETER
          PARAM_FILE_PATH ${LIBPATH}/DFTB/3ob-3-1
          PARAM_FILE_NAME param_file
          DISPERSION_TYPE D3(BJ)
          DISPERSION_PARAMETER_FILE ${LIBPATH}/dftd3.dat
          D3BJ_SCALING 1.0 0.4 1.5 4.0
          !HB_SR_PARAM 4.0
        &END PARAMETER
      &END DFTB
----

DFTB3 + D3(BJ) with default D3BJ_SCALING from the test files + HB_SR_GAMMA 
(= 4.2, as per the original paper) and no EWALD (as per the test files) - H 
atoms get way too close (< 1 A) and the calculations stops:
      &DFTB
        DIAGONAL_DFTB3 T
        !ORTHOGONAL_BASIS F
        SELF_CONSISTENT T
        DO_EWALD T
        DISPERSION T
        HB_SR_GAMMA T
        &PARAMETER
          PARAM_FILE_PATH ${LIBPATH}/DFTB/3ob-3-1
          PARAM_FILE_NAME param_file
          DISPERSION_TYPE D3(BJ)
          DISPERSION_PARAMETER_FILE ${LIBPATH}/dftd3.dat
          D3BJ_SCALING 1.0 0.4 1.5 4.0
          HB_SR_PARAM 4.2
        &END PARAMETER
      &END DFTB
----

DFTB3 + D3(BJ) with original D3BJ_SCALING from the original paper + 
HB_SR_GAMMA (= 4.2, as per the original paper), no EWALD (as per the test 
files) - H atoms get way too close (< 1 A) and the calculations stops:
      &DFTB
        DIAGONAL_DFTB3 T
        !ORTHOGONAL_BASIS F
        SELF_CONSISTENT T
        DO_EWALD F
        DISPERSION T
        HB_SR_GAMMA T
        &PARAMETER
          PARAM_FILE_PATH ${LIBPATH}/DFTB/3ob-3-1
          PARAM_FILE_NAME param_file
          DISPERSION_TYPE D3(BJ)
          DISPERSION_PARAMETER_FILE ${LIBPATH}/dftd3.dat
          D3BJ_SCALING 1.0 .5719 .5883 3.6017
          HB_SR_PARAM 4.2
        &END PARAMETER
      &END DFTB
---

ORTHOGONAL_BASIS T always breaks the calculation from the beginning (not a 
single geometry in the output), so I guess 3ob-3-1 does not have an 
orthogonal basis or DFTB3 simply does not use one. As I can see - the 
default in CP2K is FALSE, anyway.
DO_EWALD F breaks even DFTB2 calculations - the reagent (QM region) breaks 
in pieces
---

However, the following setup seems to work, although I'm not sure I can 
trust the barrier (it is very sensitive to intermolecular interactions and 
it seems kind of high, and there is no available experimental value):

DFTB3 + D3(BJ), no D3BJ_SCALING keyword (I let cp2k determine s6, sr6, s8 
automatically, as suggested in a recent new thread in this google group) + 
HB_SR_GAMMA (=4 as usual default), with EWALD - it seems to work:
      &DFTB
        DIAGONAL_DFTB3 T
        SELF_CONSISTENT  T
        DO_EWALD  T
        DISPERSION  T
        HB_SR_GAMMA T
        &PARAMETER
          PARAM_FILE_PATH ${LIBPATH}/DFTB/3ob-3-1
          PARAM_FILE_NAME param_file
        !  UFF_FORCE_FIELD uff_table
          DISPERSION_TYPE D3(BJ)
          DISPERSION_PARAMETER_FILE ${LIBPATH}/dftd3.dat
          HB_SR_PARAM 4
        &END PARAMETER
      &END DFTB
---

Next I will try to explicitly provide the Hubbard parameters with in the 
input with DFTB3_PARAM.

Thank you again for the help.
If you can think of anything else - please - tell me.

Best Regards,
Dobromir

  
On Monday, February 2, 2026 at 11:23:58 AM UTC+2 Jürg Hutter wrote:

> Hi
>
> 1) and 2) Have a look in "tests/DFTB/regtest-vdw" . There you find some 
> examples on how to use D3 with DFTB.
> 3) The D3(BJ) parameters have to be given in the input file. The data file 
> only contains the atom specific parameters,
> not the method parameters.
> 4) Yes, this is the DFTB3 parameter. I don't know if CP2K can read the 
> parameter from the DFTB parameter file. I would have to check.
> 5) You need the keyword. According to the manual the default is not to use 
> it.
> 6) I cannot answer this question.
>
> regards
> JH
>
> ________________________________________
> From: cp... at googlegroups.com <cp... at googlegroups.com> on behalf of 
> Dobromir A Kalchevski <dobro... at gmail.com>
> Sent: Saturday, January 31, 2026 11:21 AM
> To: cp2k
> Subject: [CP2K:22073] DFTB3 D3(BJ) GAMMA
>
> Hello valued developers and users,
>
> I am trying to setup DFTB3 with D3 correction and GAMMA damping and I 
> can't find the answers to all of my questions in the forum threads. Also 
> some answers are not completely clear to me.
>
> 1) If I am to use the D3 correction - is it enough to include
> DISPERSION_TYPE D3(BJ)
> DISPERSION_PARAMETER_FILE ${LIBPATH}/dftd3.dat
>
> or should I also include the &VDW_POTENTIAL section and where if I am not 
> using DFT (&XC section)
>
> 2) does D3(BJ) work with DFTB3 or does only D3 work
>
> 3) are both the D3 and the D3(BJ) parameters in the dft3d.dat file. If not 
> - for D3(BJ) I can find only a1, a2 and s8, but I can't find anywhere the 
> fourth one. Why are the default values set to 0
>
> 4) is the keyword DFTB3_PARAM in KIND the Hubbard parameter and is it 
> necessary. The 3ob .skf files technically include it, basically does CP2K 
> read it from the .skfs.
>
> 5) is using 3ob enough for DFTB3 or do I have to include the keyword 
> DIAGONAL_DFTB3 in &DFTB
>
> 6) if HB_SR_GAMMA is the GAMMA damping for hydrogen bonds, is its use 
> recommended for general organics at the moment or it would rather worsen 
> the results, as before
>
> I am attaching an example input file for QM/MM Metadynamics.
>
> Best Regards,
> Dobromir
>
>
>
> --
> You received this message because you are subscribed to the Google Groups 
> "cp2k" group.
> To unsubscribe from this group and stop receiving emails from it, send an 
> email to cp2k+uns... at googlegroups.com<mailto:cp2k+uns... at googlegroups.com
> >.
> To view this discussion visit 
> https://groups.google.com/d/msgid/cp2k/e18f9019-605d-408b-a268-100e577c52f8n%40googlegroups.com
> <
> https://groups.google.com/d/msgid/cp2k/e18f9019-605d-408b-a268-100e577c52f8n%40googlegroups.com?utm_medium=email&utm_source=footer
> >.
>

-- 
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+unsubscribe at googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/cp2k/17f23652-327a-4472-a00e-62ffbce76edfn%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20260202/34cee525/attachment-0001.htm>