[CP2K:9948] Problematic DFTB geometry optimization

[Maxime Van den Bossche]

I think it would be good to make that change in the code,
as you describe it.

I tried Broyden mixing, but the SCF cycle ends up oscillating 
without converging (both for the molecule and for the bulk), 
at least with the standard Broyden parameters and ALPHA 
values down to 0.01 (see attached zip folder). Adding smearing 
didn't help.

Best,
Maxime
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20180205/440fd9e1/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: broyden.zip
Type: application/zip
Size: 123491 bytes
Desc: not available
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20180205/440fd9e1/attachment.zip>