In the case that in the inputcard we declare that we want to do a FP calculation (KSHAPE=2, IRM=484 and INS=1, ICST=3) then shape functions are calculated and written in the file shapefun. This should be used as input for the FP calculation. A SMALL TIP: if the ASA potential is not good to start the FP calculation run the program a single iteration using (KSHAPE=1, IRM=484 and INS=1) and start from the resulting potential again with KSHAPE=2.
In the shapefun file we should add in the beginning two lines before using it
as input for the KKR. This is the number of different shape functions in the
file and weights for each of the shapes which should be 1.00. So if you have
2 different shape functions for your system you give
2
1.000000000000D+00 1.000000000000D+00
The format is the same
like in the rest of the file. (This feature will disappear in the near
future!)
In the FP calculation we should also take care for the muffin-tin sphere. Inside the voronoi polyhedra that defines the region of the space that belongs to an atom we have a muffin-tin sphere in which a more dense logarithmic radial mesh is used compared to the rest of the space where a linear mesh is used. This muffin-tin sphere that we declare in the inputcard as RMT for each atom should not exceed the boundaries of the corresponding voronoi polyhedra. The voronoi.exe reads in the inputcard the values of RMT that we have declared for each atom and also calculates the larger possible radius for each sphere. If our RMT exceeds the maximum possible value then the program uses the later value to construct the potential, otherwise it just uses our value. A smaller MT radius is needed as input or else you get no shapefunction update and no shapefunction file! Inside the output the program writes out explicitly these maximum values and what it has used as muffin-tin spheres. In case our RMT exceeds the maximum value, then we should use the latter one in our inputcard file. Special care should be given to the choice of the RMT when we want to do relaxations.
In the case that the atoms have different environments or different sizes (this is controlled by the parameter WEIGHT in the inputcard) then they will have different corresponding shape functions. In this case the program writes inside the file shapefun the shape functions for all the atoms one after the other and does an indexing that is written in the output file. Someone should take this indexing and use it inside the inputcard so that the KKR program knows which shape function corresponds to which atom. The parameter that we should change for each atom in the inputcard is the NTC and we use for each atom the number assigned by the voronoi program for the shape functions. Of course a similar thing occurs when we do not have the same TB-cluster for each atom due to symmetry reasons. Also in this case the voronoi.exe calculates the clusters and in the output file it writes out how many different clusters should be used and assigns number to each one of them. These number are the values of the CLS parameter inside the inputcard file as explained also above.
If we have the potential for one position of the atom then we can use this potential in combination with the voronoi.exe to produce a starting potential for another position of the atom. The problem is that in order to calculate accurately the forces and the total energies the core electrons should be calculated in the same way (same radial mesh). The core electrons are always calculated inside the muffin-tin sphere and so we should choose exactly the same RMT for all the calculations.
Finally recompile the voronoi program if you wants to change one of these parameters in the maindriver.f program
bbox : data statement changes the bounding box
for drawing atoms with povray, or rasmol. all atoms
in the box are written out in file lattice.pdb.
To visualize just type visual, a small script will give you
the rasmol picture. For more on rasmol: www.OpenRasMol.org
npoi : data number of points for the shape function
usually set to 125 and can be changed
dlt = 0.05 controls the accuracy of angular integration
for producing the shape functions.
nrad : number of points for between true MT and user defined MT
used in sub mtmesh usually set to 10.
In case the program stops complaining about dimension errors you will find the dimensions in file "inc.geometry".
There is also one known problem concerning the voronoi program. When the coordinates of the atoms are not exact, usually when square roots of numbers enter the coordinates, the program fails to construct the voronoi polyhedra. In this case change the WEIGHT of the atoms slightly.