# Strange relation between simulation box size and computed data

Dear LAMMPS users!

During melting process of my finite cluster which has bcc lattice, I encountered a strange link between Voronoi/atom computation outcome values and simulation box size where my cluster resides. Voronoi/atom computes the volume of unit cell for every atom with use of Wigner-Zeits idea. It seems like W-Z cell’s volume of an atom that lies on cluster’s surface, depends on distance between that atom and the closest point on the box. As a result, I obtain volume values that lies in range of 30-100 A^3 for atoms inside cluster and 10000+ A^3 for atoms that lies on the surface.

The question is how can I escape this unreasonable big values and calculate the right volume of my cluster?

Another question that bothers me is the wide range of unit cell’s volume inside the cluster at the given temperature. How it is possible that some atom has 37 A^3 (or 44, 77, 104) volume whereas other has 515 A^3 and the third has volume of 1915 A^3 at the same time?

I will be glad to any suggestions and clues!
Ruslan

The surface issue is a limitation of the Voronoi method, not of
LAMMPS. The question of the range of volumes for atom inside the
cluster is also a feature of the Voronoi method.

Dear LAMMPS users!

During melting process of my finite cluster which has bcc lattice, I
encountered a strange link between Voronoi/atom computation outcome values
and simulation box size where my cluster resides. Voronoi/atom computes the
volume of unit cell for every atom with use of Wigner-Zeits idea. It seems
like W-Z cell's volume of an atom that lies on cluster's surface, depends
on distance between that atom and the closest point on the box. As a
result, I obtain volume values that lies in range of 30-100 A^3 for atoms
inside cluster and 10000+ A^3 for atoms that lies on the surface.

​yes, this is how voronoi tesselation works.​

The question is how can I escape this unreasonable big values and
calculate the right volume of my cluster?

​you cannot use voronoi tesselation for this purpose.​ there is not a
single good method for this. the volume of a cluster is not a very well
defined property. you should look into tools that are used to estimate
molecular surfaces and volumes via probe particles.

axel.