dynamical matrix calculation

Hi all,

I have a question regarding dynamical matrix calculation using LAMMPS. In
the USER-PHONON/examples/1-1D-mono/ directory I calculated the dynamical
matrix at q=(0.5 0 0) and the dynamical matrix it produced is given below:

# q = [0.5 0 0]
3.9469 0

But since the system is 1-dimensional and also it has 1-atom basis, then
the dynamical matrix should be a single number in stead of a 1by2 matrix.
This happens all the time. If I calculate the dynamical matrix of a 2-D
system with 2-point basis, then it is a 4by8 matrix in stead of a 4by4
matrix (given below):

# q = [0.5 0 0]
842.997 0 40.0807 0 -680.244 0 -317.185 0
40.0807 0 759.033 0 -318.323 0 -284.777 0
-680.244 0 -318.323 0 843.284 0 40.7265 0
-317.185 0 -284.777 0 40.7265 0 756.014 0

I don't understand what is the reason behind it. Where does these
zero-columns come from all the time? If I want to calculate the
eigenvectors of the dynamical matrix, then what should be the actual
matrix?

Best Regards

Can you Calculate the Dynamical Matrix using LAMMPS ? WOW I had no idea you could do that .
If someone may please give me more information how to do it I would really apreciate it.

Indeed! I had no idea LAMMPS could do this. If so, it’s a tremendously useful feature, maybe it should be better publicized? :slight_smile:

Rob

Can you Calculate the Dynamical Matrix using LAMMPS ? WOW I had no idea you
could do that .
If someone may please give me more information how to do it I would really
apreciate it.

there is no USER-PHONON package in the official LAMMPS distribution,
so this has to be a custom version of the code.

axel.

Agreed - so if someone has such a package, please let
us know. We could advertise it or release it with LAMMPS.

Steve

Agreed - so if someone has such a package, please let
us know. We could advertise it or release it with LAMMPS.

FYI,
a quick search on google reveals this:

http://code.google.com/p/fix-phonon/

axel