[Wien] dstart.error

[Peter Blaha]

I cannot reproduce this error. init -b ...  runs fine.

The struct file you included has no symmetry operation. With this struct 
file, dstart would fail.
However, during initialization the identity is written to the file and 
this one must be used later on.
-----------------------------------

However, this structure is completely ridiculous and Wien2k calculation 
for it would take weeks and would probably need a computer with at least 
1000 cores + memory !!!

Why ?   What counts for cpu time is only partially the number of atoms, 
but the size of the cell. You have 38x38x190 bohr !

You have a vacuum of about 160 bohr !!!  and also the x-y supercell for 
ONE Tl-OH molecule is MUCH too large (they are separated by 40 bohr !!).
In addtion, your distances (some of the O-H) are ridiculous small, 
forcing very small RMTs.
You passivate only one of the SiO2 surfaces by H, but then put the Tl-OH 
on the passivated surface ???

I don't know from where the structure is (except from ASE) ?? but this 
one would be very expensive in WIEN2k (and I'm pretty sure with any code 
this is very hard). Have you ever visualized it ???
------------------------

Wien2k offers very powerful tools to generate a "good" model of SiO2/Tl-OH.
- Start with the bulk SiO2, relax the volume and positions. Use optimal 
lattice parameters.
- Create the corresponding surface. For simple surfaces use   x 
supercell, for complicated ones use the structeditor (you need octave !) 
- A typical vacuum is 30 bohr, maybe if you later plan to add Tl-OH, use 
40 bohr.
- Decide whether you want to passivate the surface with H and add them 
"manually" for each surface O-atom). In WIEN2k it is usually much more 
efficient to have inversion symmetry, so I would passivate either both 
or none of the surfaces.
- Relax the surface.
- Now decide on the xy-supercell to isolate the Tl-OH molecule. 
Typically, very often even a 2x2 supercell is sufficient. (x supercell)
- Now add the Tl-OH "manually" and relax it.

PS: How to add atoms manually ???
open the structure in xcrysden. Select the site where you want to add 
the atoms. Identify their coordinates using xcrysden (use eg "distances" 
and click on two atoms - select fractional coordinates). With this you 
can easily get a good guess about the coordinates of our "add-atoms".

Adding this atom can be done in w2web (for beginners) or directly 
editing the struct file:
Copy an "atom-block" and paste it at the end. Modify coordinates, name, 
ZZ, and eventually R0.
Don't forget to increase the number of atoms at the top of the struct 
file and to set the number of symm.ops to zero.

Start with a "small" model and fully relax it.
Later on you can check your results by increasing the model (with x 
supercell you can add vacuum, or make another 2x2 supercell and remove 3 
of the Tl-OH molecules to have a smaller coverage.

Best regards
Peter Blaha


-------------------------------------------------------
Betreff:
dstart.error
Von:
"Ilias Miroslav, doc. RNDr., PhD." <miroslav.ilias at umb.sk>
Datum:
28.08.2020, 13:23
An:
"wien at zeus.theochem.tuwien.ac.at" <wien at zeus.theochem.tuwien.ac.at>

Hello,

I prepared this structure (307 atoms)- from xyz via ase,  and it passed 
until dstart , giving:

  'ROTDEF' - no symmetry operation found.
  'ROTDEF' - for jatom, index   1   1 

  'ROTDEF' - atomposition of jatom   0.9979360   0.0003974   0.0000462 

  'ROTDEF' - atomposition of index   0.9979360   0.0003974   0.0000462

Any help, please ?

Miro
-- 
--------------------------------------------------------------------------
Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-1-58801-165300             FAX: +43-1-58801-165982
Email: blaha at theochem.tuwien.ac.at    WIEN2k: http://www.wien2k.at
WWW: 
http://www.imc.tuwien.ac.at/tc_blaha-------------------------------------------------------------------------