Help with periodicity and reduction of system size

Dear Users/Developers,

I am continuing the work on amorphous materials done by of my collegues and I want to use restart files from his simulations.

The problem however is that the system is large (periodic boundary conditions in all the directions) which is unnecessary for the kind of work I wish to pursue. I could delete atoms but do not want to create free surfaces, also need to preserve the inherent randomness of the atoms.

Is there a way to solve this problem?

Thanks

Sankha Mukherjee

Not likely. If it is amorphous and the BCs are all periodic, then there is no way to cut through the box and delete some atoms while maintaining its periodicity.

Depending on what you would like to do to the system in your simulation, one thing you could do is to cut through the box (creating surfaces) and immobilize a layer of atoms with a certain thickness near the surfaces. This way the atoms in the center or those far away from the surfaces do not know what happened to the boundary. Note this approach does create free surfaces, it is just these free surfaces do not affect the dynamics of the atoms far away from them.

Ray

Ray,

Surely if the sample is truly amorphous, any subset will display the same (amorphous) properties.

So a method such as;

define a region defining the new subgroup

group all the atoms in the group

create another group which is all the rest

delete the second group

change_box boundary s s s

run 0 ## to get lammps to move the boundaries

change_box boundary p p p

would that work?

A quick check (testing for some thermodynamic property) against the original will determine if the subset is suitable

Nigel

Ray,

Surely if the sample is truly amorphous, any subset will display the same
(amorphous) properties.

So a method such as;

define a region defining the new subgroup

group all the atoms in the group

create another group which is all the rest

delete the second group

change_box boundary s s s

run 0 ## to get lammps to move the boundaries

change_box boundary p p p

would that work?

you would create a scenario with very high potential energy at the
boundaries. this would have to be removed and properly dissipated and
then the structure properly relaxed. you need to be very careful here,
or else you generate shock waves that will not easily vanish. since
the target is to create a smaller structure, it is probably faster, to
simply build a new restart from scratch.

axel.

Sure, a subset will display the same thermodynamic properties but it is the boundary (surface) that will cause the problems. Your approach would work for noble gases, i.e., atomic systems without bonds. For any other amorphous systems (solid, liquid, polymer, glass, etc), you risk cutting through bonds between atoms at the boundary hence creating dangling bonds and unusually high potential energy. Changing the free surface to periodic boundary would even worsen the problem.

Ray