Non-gaussian and van hove, msd/nongauss, com/chunk

Hi all,

I am calculating the non-gaussian parameter and van hove function for molecules, i.e., my task should be:
Calculating the center of mass for some molecules, and then use the COMs to get the non-gaussian parameter and van hove function, both distinct part and self part.

My questions are:

1. Is there a module in lammps to get non-gaussian parameters for chunks?

“compute msd/nongauss” is very useful for atoms and it might be ok to represent small molecules by their center atoms. But I still want to know if there is an existing module calculating non-gauss for the centers of mass of molecules?

1. How “compute msd/nongauss” deals with periodic boundary conditions?

The manual page says “including all effects due to atoms passing through periodic boundaries”. To my understanding, it means the unwrapped coordinates are used. Is this the way it works?

1. How to export centers of mass in unwrapped form or with image flags?

To write post-processing myself, I hope to fully use the modules in lammps. Does the “compute com/chunk” output the wrapped or unwrapped coordinates?
To get non-gaussian parameter and van hove function, I believe that unwrapped data should be used. The manual page says the com/chunck: “The coordinates of an atom contribute to the chunk’s center-of-mass in “unwrapped” form, by using the image flags associated with each atom.” But the output only has 3 columns of x,y,z. Are these coordinates unwrapped or wrapped? How can I get the unwrapped COMs or image flags for the chunks?

Maybe too many questions. Appreciate your patience.

Luo

Steve

Hi all,

I am calculating the non-gaussian parameter and van hove function for molecules, i.e., my task should be:
Calculating the center of mass for some molecules, and then use the COMs to get the non-gaussian parameter and van hove function, both distinct part and self part.

My questions are:

1. Is there a module in lammps to get non-gaussian parameters for chunks?

No - someone would have to add a compute for this.

“compute msd/nongauss” is very useful for atoms and it might be ok to represent small molecules by their center atoms. But I still want to know if there is an existing module calculating non-gauss for the centers of mass of molecules?

1. How “compute msd/nongauss” deals with periodic boundary conditions?

The manual page says “including all effects due to atoms passing through periodic boundaries”. To my understanding, it means the unwrapped coordinates are used. Is this the way it works?

yes

1. How to export centers of mass in unwrapped form or with image flags?

To write post-processing myself, I hope to fully use the modules in lammps. Does the “compute com/chunk” output the wrapped or unwrapped coordinates?

They should be output in unwrapped coords. You can test this with a simple system.

To get non-gaussian parameter and van hove function, I believe that unwrapped data should be used. The manual page says the com/chunck: “The coordinates of an atom contribute to the chunk’s center-of-mass in “unwrapped” form, by using the image flags associated with each atom.” But the output only has 3 columns of x,y,z. Are these coordinates unwrapped or wrapped? How can I get the unwrapped COMs or image flags for the chunks?

I think the answer is the same as above. What the regular com/chunk uses to do its computation and also
outputs is unwrapped coords. So if you want to write a variant for a non-Gaussian verison, you
could just follow the coding in the original.