fix rigid/small from LAMMPS 13

Submitted by zamir on Fri, 09/20/2013 - 18:37

Hello all,

My colleague and I are trying to port fix rigid/small from LAMMPS sep13 to the liggghts 2.3.7 environment. We are getting tripped up in the barycenter calculation in tri_mesh_i.h. If anyone can give us insight into how we need to modify tri_mesh_i.h or fix_rigid_small.cpp/h, it would be greatly appreciated.


zamir | Tue, 10/01/2013 - 23:13

Hi All,

We are in the process of porting fix rigid/small to liggghts. We have some questions for the community. If anyone can help it would be greatly appreciated. Here is some background and the question:

In Lammps, there exists support for communications object rings, that leverage packing/unpacking of buffers, exchanges, reverse/forward comm for both fixed and variable sized variants etc. In liggghts, the communications object model architecture appears to support a subset of the Lammps communication model functionality.

Question: Is the communications architecture model of Lammps more preferential then the current liggghts implementation, in terms of speed and volume of data processing/handling?

Also, is there any effort to implement the communications functionality of Lammps into liggghts.

FYI, when we arrive at a solution, we intend for the results to benefit the entire open source community.

Thank you!!!

ckloss's picture

ckloss | Thu, 10/03/2013 - 16:22

Hi Zamir,

there will be a merge of LAMMPS and LIGGGHTS - planned for October.

With respect to your questions: All the comm implementation of LAMMPS (at the time of the last merge) is in LIGGGHTS, and in LIGGGHTS there is additionally a parallelization kernel which for meshes, which is superior in terms of maintainability and extendability