The ABINIT Users Mailing List ( CLOSED )

["M.T.Storr" <m.t.storr@reading.ac.uk>]

Hi,

thanks to those people who answered my questions on the number of bands in 
a calculation. The question has now been resolved and replaced by several 
further questions!
I have run through the study of ecut, then the number of kpoints for my 
system and so am now at the point of running a geometry optimisation, 
however I am running into problems with ecut/boxcut.
I have chosen an ecut of 50 Ha and a kpoint grid as follows:

#Definition of the k-point grid
kptopt 1          # Option for the automatic generation of k points, 
taking
                  # into account the symmetry
ngkpt 6 6 6       # Definition of the different grids

nshiftk 4         # repeated four times, with different shifts

shiftk 0.5 0.5 0.5
       0.5 0.0 0.0
       0.0 0.5 0.0
       0.0 0.0 0.5
                  # In cartesian coordinates, this grid is simple cubic, and
                  # actually corresponds to the
                  # so-called 4x4x4 Monkhorst-Pack grid

However when I try to run the geometry optimisisation using:

optcell 2    # Optimise Nuclear Positions only.
ionmov 3     # Broyden-Fletcher-Goldfarb-Shanno minimization (BFGS).
ntime 10     # Number of BFGS steps.
dilatmx 1.15
ecutsm 0.0

Then when the program goes through the input data and echoes the dataset 
it uses and ecut of:

 getcut: wavevector=  0.0000  0.0000  0.0000  ngfft=  40  40  40
         ecut(hartree)=     66.125   => boxcut(ratio)=   2.14285

whihc is not the 5o Ha that I requested.
However the program goes ahead and runs the Broyden Step 0, however at the 
end of this step the program then goes to Broyden Step 1 and then selects 
the Ecut as requested of 50 Ha:

 getcut: wavevector=  0.0000  0.0000  0.0000  ngfft=  40  40  40
         ecut(hartree)=     50.000   => boxcut(ratio)=   0.67925

However as can be seen the boxcut is now below 1 and I get the following 
error:

 getcut: ERROR -
  Choice of acell, ngfft, and ecut
  ===> basis sphere extends BEYOND fft box!
  Recall that boxcut=Gcut(box)/Gcut(sphere)  must be > 1.
  Action : try larger ngfft or smaller ecut.
  Note that ecut=effcut/boxcut**2 and effcut=   23.068977

The question is what am I doing wrong. I would like to use an ecut of 50 
Ha, and have therefore tried increasing ngfft however I tried 50 50 50 
and got a boxcut(ratio)=0.84940. This suggests that I would need to 
increase ngfft to a rather large value just to get the calculation to run.
Therefore I would like to know why for the Broyden step 0 the program 
appears to select its own value of ecut which when used to intialise 
variables such as ngfft etc means that at step 1 when it uses the correct 
value all the initialised variables are wrong. Any suggestions how to 
solve this would be great.

Many Thanks

Mark

-- 
Mark T. Storr
Research Fellow
School of Chemistry
University of Reading
PO Box 224
Whiteknights
Reading
Berkshire
RG6 6AD
Tel: +44 (0)118 9875123 Ext. 7415
E-mail: M.T.Storr@reading.ac.uk