[DFTB-Plus-User] Lattice Optimization in DFTb+

[Bálint Aradi]

Dear Laura,

thanks a lot for the detailed investigations!

I just made some tests for the anatase primitive unit cell and obtained
similar differences between the end results of the optimisers (in the
range of a few hundreths AA), so I can confirm your result.

However, the main reason for those discrepancies are not the optimisers,
but the Ti-O parameterisation, which contains unfortunately a lot of
noise (probably to a technical mistake made when it was created). Due to
this, a bulk has many local minima with very similar energies and
slightly differing lattice constants. (Also other artefacts, like very
strange vibrational frequencies can be occur.)

I have recently created a smoothed version of the tiorg set, where I
have removed most of the noise. We are still evaluating that set. Once
we are sure, it works fine, it will be published as usual on dftb.org.

Using the smoothed version of tiorg, I get the same optimized bulk
geometries with CG and LBFGS (I did not try an other optimisers), with a
precision of <1e-2 AA for the lattice constant, < 0.1 degree for the
lattice vector angles and < 1e-4 eV for the total energy.

If you are interested to test the smoothed set right now, please contact
me off-list, I'd be happy to share it with you.

  Best regards,

  Bálint

-- 
Dr. Bálint Aradi
Bremen Center for Computational Materials Science, University of Bremen
http://www.bccms.uni-bremen.de/cms/people/b-aradi/


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: OpenPGP digital signature
URL: <http://mailman.zfn.uni-bremen.de/pipermail/dftb-plus-user/attachments/20190122/f90250ee/attachment.sig>