The ABINIT Users Mailing List ( CLOSED )

[paul.boulanger@umontreal.ca]

Hi Sergey,
            in version 4.6.3 we cannot use cell optimization with the PAW 
formalism. It seems the smearing of the functions is not yet implemented in 
this version. This means that we cannot define dilatmx and optcell (other 
then zero). In a previous post, Marc said these features are already 
implemented in the developer’s version and will be available from v4.6.4 
version. 

            This is apparent in a series of calculation I have done today.  
The convergence study in regards of Ecut of bulk cobalt is straightforward 
and gives no errors. But, when I modify the input file, defining optcell to 3 
and optcell to 1.05, the calculation stops in the first density calculation 
because the density becomes negative. The same kind of error found in your 
calculation, see your log file:  SCF step 23. This shows that the error 
originates from these variables.

               The strategy to adopt is to revert to good old fashion ways 
and manually change the cell parameters and minimizing the energy by a 
Murnaghan fit. Lets hope that version 4.6.4 is not to far away…
                                             paul