Smooth cutoffs in D3/D4

[asrosen]

In the standalone simple-dftd3 and dftd4 packages, there has been an important update for periodic systems that fixes discontinuities that can appear due to a fixed cutoff distance. The upstream libraries may switch to enabling a smooth cutoff by default. This has two implications for VASP: 1) The built-in DFTD3 (ivdw=12) will differ from the standalone simple-dftd3. In fact, it has always differed slightly due to a different choice in VDW_RADIUS (see https://github.com/dftd3/simple-dftd3/issues/151). 2) Users upgrading simple-dftd3 or dfdtd4 and linking to VASP may get different values. This is ultimately on them.

I have opened this simply for the sake of documentation.

Details:
- https://github.com/dftd3/simple-dftd3/issues/199
- https://github.com/dftd4/dftd4/issues/377

[michael_wolloch]

Hi Andrew,

Thanks for this post. I have moved it to the "from users for users" section of the forum, because I think this is exactly the type of post that we want to encourage there.

I will take a look and update our wiki as well if necessary.

Cheers, Michael

[asrosen]

Thanks!! I saw the docs change.

Currently, the simple-dftd3 package and VASP differ in the default VDW_RADIUS as well. In my personal opinion, it would be ideal for the upstream simple-dftd3 package to switch the VDW_RADIUS to 95 Bohr from its current value of 60 Bohr because 95 Bohr seems to be the standard and was in the original implementation. 95 Bohr is also very close to the value used in VASP. I make the case for that at https://github.com/dftd3/simple-dftd3/issues/200 based on prior discussion at https://github.com/dftd3/simple-dftd3/issues/151.

[fabien_tran1]

In principle, the next version of VASP should allow these new smooth cutoffs to be specified.