Optimize exponents summation in "exact" mesocrystal FF algorithm (#892)
Now the summation of exponents uses the analytical expression for the geometric sum along 3rd axis (z axis by default).
So the computational complexity now is ~O(Nx * Ny)
instead of previous ~O(Nx * Ny * Nz)
, where N is the number of particles along the axis.
struct ShapeIndexes
is introduced to diminish number of variables and to mask the unused ones in some contexts.
Resolves #892 (closed)
Edited by Mikhail Svechnikov