Please add documentation/tutorial for utilizing DFTB parameters from other sources

[xuetaoshi]

Hi all,
I am rather new to DFTB/LATTE and this issue may not be an issue at all for more experienced users. But I find it very daunting to translate DFTB parameter files obtained from websites such as DFTB.org into whatever files needed in TBparam. In this case, DFTB.org provide Slater-Koster file (.sk) and I can't find anywhere in the manual or wiki or even Google on how to translate this file into the files LATTE reads.

Thank you very much,

Xuetao

[cawkwell]

Dear Xuetao, Chang Liu at NCSU might be able to help with this. His email address is: cliu18@ncsu.edu. Cheers, Marc Theoretical Division Los Alamos National Laboratory Los Alamos NM 87545 USA 505 665 3302

…

________________________________________ From: Xuetao Shi [notifications@github.com] Sent: Friday, November 16, 2018 1:15 PM To: lanl/LATTE Cc: Subscribed Subject: [lanl/LATTE] Please add documentation/tutorial for utilizing DFTB parameters from other sources (#94) Hi all, I am rather new to DFTB/LATTE and this issue may not be an issue at all for more experienced users. But I find it very daunting to translate DFTB parameter files obtained from websites such as DFTB.org into whatever files needed in TBparam. In this case, DFTB.org provide Slater-Koster file (.sk) and I can't find anywhere in the manual or wiki or even Google on how to translate this file into the files LATTE reads. Thank you very much, Xuetao — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub<#94>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AVubd5VtX9_8RVKZCaZYhdPs1_9V5PJaks5uvxzrgaJpZM4Ym5k0>.

[cawkwell]

Dear Xuetao, LATTE can take parameters in table format, which can be converted from the skf files. To run a LATTE calculation with table format, you will need set the following flags in the LATTE input file: PARAMPATH= <Path_to_parameter_file_folder> SCLTYPE= *TABLE* # to use table format of BI PPOTON= *2* # to use table format of repulsive potential With these flags, you will need to put three parameter files inside the PARAMPATH folder: * electrons.dat - defining the atomic parameters * bondints.table - defining the bond integrals (Hamiltonian and Overlap) * ppots.dftb - defining the repulsive potentials You can find examples of these files in the TBparam folder of this repository. And I will use these three files to explain the format of these files for you. Units used in these files are eV for energies, amu for masses, and Angstrom for lengths, which are different from skf files' atomic units. <electrons.dat> LINE 1: Noelem= <number of elements in the file> LINE 2: title line LINE 3 to 2+Noelem: defining elements * The meaning for element, basis, and mass are self-explanation. * "Numel" means the number of electrons in the valence shell for the atom. * "Es", "Ep", "Ed", and "Ef" are the on-site energy for s, p, d, and f orbitals. * "HubbardU" is the Hubbard U for the atom. In skf file, Hubbard U is defined for each orbital. However, unless you specified to use orbital based Hubbard U, DFTB+ will always use Us. As a result, HubbardU in electrons.dat is Us in an skf file. * "Wss", "Wpp", "Wdd", and "Wff" are spin parameters for each orbital, which are not included in skf files. If do not need spin-polarized calculation, these values can be simply set as zeros. <bondints.table> LINE 1: NOINT= <number of BI tables (orbital pairs)> LINE 2: Element1 Element2 BItpye LINE 3: npt, Number of points in this BI table LINE 4 to 4+npt: ri S(ri) H(ri) ...BI table for next pair... * BItpye are in a format like "pds", where the first and second letters represent the angular momentum of orbitals, and the last letter represents the symmetry of bond. "s" represents sigma; "p" represents pi, and "d" represents delta. <ppots.dftb> LINE 1: nppot, number of repulsive potential tables (element pairs) LINE 2: Element1 Element2 LINE 3: npt, Number of points in this ppot table LINE 4 to 4+npt: ri Vrep(ri) ...PPOT table for next pair... For each BI table and PPOT table, make sure to append a point with the function value as zero if skf file doesn't have zero at the end. The distance for the last zero element will serve as the cut-off radius for the pair. And if the tail of a pair is not smoothly going to zero, there will be some discontinuity of properties calculated around at distance. I have a python script to do the conversion, which is attached to this email. To use it, you need to put all skf files for converting inside one folder, and type "python DLtab.py <skf folder> <LATTE parameter folder>". Best regards, Chang

…

On Fri, Nov 16, 2018 at 3:38 PM Cawkwell, Marc Jon ***@***.***> wrote: Dear Xuetao, Chang Liu at NCSU might be able to help with this. His email address is: ***@***.*** Cheers, Marc Theoretical Division Los Alamos National Laboratory Los Alamos NM 87545 USA 505 665 3302 ________________________________________ From: Xuetao Shi ***@***.*** Sent: Friday, November 16, 2018 1:15 PM To: lanl/LATTE Cc: Subscribed Subject: [lanl/LATTE] Please add documentation/tutorial for utilizing DFTB parameters from other sources (#94) Hi all, I am rather new to DFTB/LATTE and this issue may not be an issue at all for more experienced users. But I find it very daunting to translate DFTB parameter files obtained from websites such as DFTB.org into whatever files needed in TBparam. In this case, DFTB.org provide Slater-Koster file (.sk) and I can't find anywhere in the manual or wiki or even Google on how to translate this file into the files LATTE reads. Thank you very much, Xuetao — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub< #94>, or mute the thread< https://github.com/notifications/unsubscribe-auth/AVubd5VtX9_8RVKZCaZYhdPs1_9V5PJaks5uvxzrgaJpZM4Ym5k0 >.

[cnegre]

Dear Xuetao, We still don't have the skf reader fully implemented yet. We are working on it and expect to have it done soon. We will post it on the github readme file as soon as it is ready.

…

On Fri, Nov 16, 2018 at 1:15 PM Xuetao Shi ***@***.***> wrote: Hi all, I am rather new to DFTB/LATTE and this issue may not be an issue at all for more experienced users. But I find it very daunting to translate DFTB parameter files obtained from websites such as DFTB.org into whatever files needed in TBparam. In this case, DFTB.org provide Slater-Koster file (.sk) and I can't find anywhere in the manual or wiki or even Google on how to translate this file into the files LATTE reads. Thank you very much, Xuetao — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#94>, or mute the thread <https://github.com/notifications/unsubscribe-auth/APQltYiVlETw6YlGDR5r1m9jkLXO8i11ks5uvxzrgaJpZM4Ym5k0> .

-- Dr. Christian F. A . Negre LOS ALAMOS NATIONAL LABORATORY T-1 Division, Mail Stop B221 Los Alamos, NM 87545

[xuetaoshi]

Thanks to the help from all of you, I have used the Python script provided by Chang Liu and successfully converted .skf file from DFTB.org into parameters useable by LATTE. The calculations run with no problem though I cannot verify the results are absolutely correct. I understand that the built-in converter is in progress, but I would suggest putting the aforementioned Python script into the repository anyway in case anyone needs it before said converter gets fully implemented.

[bqluanibm]

I had the luck to use the code DLtab to convert some skf paramenter files on DFTB.org, but I cannot use the DLtab.py to convert matsci-0-3 (https://dftb.org/parameters/download/matsci/matsci-0-3-cc) skf files to latte parameter files. The header in Ti-Ti (or any pair of the same element) seems new. Especially, the string "T" in the second line cannot be recognized by DLtab.py. Can someone help? Thanks!