NEB with almost equal initial and final states

Dear LAMMPS users,

I research the diffusion of point defects in crystal and faced the problem: how to make NEB move environment atoms if their initial and final states should be same? I look at the “sivac” example, but there only nearest neighbors with different start and final positions are moving.

Parameters that I used are:
min_style quickmin
fix qwerty all neb 0.5
neb 0.001 0.01 2000 2000 50 final data_single_atom.final

Vacancy atom moves properly, but its neighbors are frozen. Do you have any idea how to make them move?

Thanks for attention!

Sincerely,
Karen Fidanyan,
Master student at Moscow Inctitute of Physics and Technology,
Department of Molecular and Chemical Physics.

Your Q is fuzzy with respect to “move” and “state”.

To use NEB you have to define a set of atoms that

has two distinct states. The set can be all the

atoms in the system or a subset or just one atom.

The state is the coords of all the atoms in the set.

The initial vs final state have to be different, else

there is nothing for NEB to do. Whichever atoms

are in the set, why you perform NEB, they are free

to move.

Steve

Dear lammps users,

Recently I developed a new pair style for Si-C based on Vashishta's article. The bulk physical constants has been verified under the "p p p" B.C and all the parameters mentioned in the article could be reproduced very well. But the atoms on the four of eight box conners fly away when the B.C changed into other conditions.

1st question: I wonder whether there is some mistakes contain in my codes. If it is true, but why could it reproduce the physical constants just under the "p p p" B.C?

The new pair style consist of two parts: two-body part & three-body part. I have tailored the potential just to be the two-body or three-body. The test results indicated that the "etotal" is almost the same as original pair style in the two-body condition. But it is near to zero in the three-body. The test input script show below:

Dear lammps users,

Recently I developed a new pair style for Si-C based on Vashishta's
article. The bulk physical constants has been verified under the "p p p"
B.C and all the parameters mentioned in the article could be reproduced
very well. But the atoms on the four of eight box conners fly away when the
B.C changed into other conditions.

1st question: I wonder whether there is some mistakes contain in my codes.
If it is true, but why could it reproduce the physical constants just under
the "p p p" B.C?

​counter question: which is the correct physical behavior in this case?

The new pair style consist of two parts: two-body part & three-body part.
I have tailored the potential just to be the two-body or three-body. The
test results indicated that the "etotal" is almost the same as original
pair style in the two-body condition. But it is near to zero in the
three-body. The test input script show below:

##############
units metal
boundary p p p
#boundary s s s
timestep 0.0001

# Zincblende SiC
lattice diamond 4.3581
region box prism 0 1.0 0 1.0 0 1.0 0.0 0.0 0.0
create_box 2 box
create_atoms 1 box basis 5 2 basis 6 2 basis 7 2 basis 8 2
mass 1 28.086000
mass 2 12.011150

# Choose potential
pair_style vashishta
pair_coeff * * SiC.vashishta Si C

#pair_style tersoff
#pair_coeff * * SiC.tersoff Si C

# Setup neighbor style
neighbor 1.0 nsq
neigh_modify once no every 1 delay 0 check yes

# Setup output
thermo 50
thermo_style custom step etotal pe press lx ly lz vol
thermo_modify norm no
dump myDump all custom 50 dump.va-ppp id type x y z

fix 1 all nve
run 20000
##############

2nd question: Is there any mistakes in my script that would get an
unreasonable phenomenon mentioned above?

2nd counter question: where is the proof that the behavior you see is
"unreasonable"? through removing PBC, you are cutting through bonds, right?

axel.

Dear Axel,

Thanks for your quick response.

The physical behavior include in the lattice constant, cohesive energy, melting point and the elastic constants C11, C12.

The unreasonable behavior refers to that the atoms fly away in the four conners of the box under aperiodic B.C with the original pair style and the only-two-body pair. The only-three-body pair is always stable under the various B.C.

What do you maen by the “bonds”? As I know there isn’t any bond in the pair style.

Regards
Guangpeng
在 2015年10月15日,上午1:22,Axel Kohlmeyer <[email protected]> 写道:

Dear Axel,

Thanks for your quick response.

The physical behavior include in the lattice constant, cohesive energy,
melting point and the elastic constants C11, C12.

​you are not answering my question.​ a cluster is not the same thing

The unreasonable behavior refers to that the atoms fly away in the four
conners of the box under aperiodic B.C with the original pair style and
the only-two-body pair. The only-three-body pair is always stable under the
various B.C.

What do you maen by the “bonds”? As I know there isn't any bond in the
pair style.

​yes, they are, only they are handled implicitly. but i wasn't talking
about the model, i was talking about the real thing. the real structure.

let me take a very simple example. an ethane molecule: CH3-CH3. if you
split it in half, you have two methane radicals. the ethane molecule is a
stable entity. the methane radicals much less so. just pick up a chemistry
text book. what you do, if you remove PBC, is essentially the same thing.

overall, it looks to me, that you are missing some fundamental
understanding of the underlying physics of the system you are studying.

axel.

Sorry, Axel. I didn’t understand your idea well before.

The material used to simulate is single crystal silicon carbide(3C-SiC) which is a classic covalent compound with the zink blende structure.

Guangpeng
在 2015年10月15日,上午2:21,Axel Kohlmeyer <[email protected]> 写道:

Sorry, Axel. I didn't understand your idea well before.

The material used to simulate is single crystal silicon carbide(3C-SiC)
which is a classic covalent compound with the zink blende structure.

​that is essentially what i assumed. you are still failing to provide a
*physical* justification for why what you observe is "unreasonable". it
just flies in the face of my understanding of covalent bonding, that
cutting through bonds in the way you do when you remove periodicity,
without allowing for any kind of reconstruction/recombination should
produce a stable structure that would behave like bulk.

axel.

You cannot fool people by just changing the subject of your messages. You need to understand
that there are very slim chances the enthusiasts that regularly participate in this list have the expertise to answer questions coming from all flanks. Your approach to tackling this problem is all upside down.
You are trying to get the developers to help you fix some issues with “your” code because you fear there is a subtle problem with it. You use the argument that your code passed some bulk tests in order to prove to the developers that you seem to know what you are doing. The original paper for the parametrization you cite did compute several properties. Have you checked them all already? Furthermore, that paper has been cited numerous times; have you tried finding within those references alternative SiC structures (clusters perhaps?) and their properties you could use to further validate your implementation? And why not reaching out to the guys that developed or have used such parametrization in the past for the fine print? Its beyond my comprehension what is that you plan on achieving by posting over and over the same question. The most probable outcome is that the few people that actually “may have cared” in relation to helping you will get upset/frustrated and will stop answering. And that “lammps community” that you plan on sharing your potential with will remain silent, just as it has done so far.
Good luck,
Carlos

Hi, Axel.

Actually, I want to simulate a diamond tip sliding on the SiC surface which is similar to the LAMMPS provided example “friction”. According to the existing articles, the simulating boundary conditions always set as “s s p”. So I use the same BCs in my implementation. But the SiC exploded in the initial minimizing period. Then I run the minimization only with the SiC substrate under various BCs. It just can be stable with the PBCs.

Furthermore, I have some bewilderments in the BCs setting. Under free boundary conditions, how do the atoms get to equilibrium at the boundaries( ie. the surfaces and edges of the simulation box) ? The inner atoms can interact with the neighbour atoms, but the boundary atoms only can interact with one side atoms.

Thanks in advance.

best regards

guangpeng

We are going in circles now. This is getting off-topic and has next to nothing to with lammps, but everything with how to set up simulations and the physics of surfaces/edges of covalent bound crystals and clusters.

Talk to your adviser, talk to your colleagues, study textbooks and so on. A mailing list is not the place for personal tutoring and I have no time and desire to tutor you in a research subject that I have no interest in.

Axel

Dear mr. Plimpton, thanks for your answer!

I’ve found the mistake in my postprocessing, so lammps worked correctly all that time, but I didn’t see the motion of atoms.
Sorry for such newbie question :slight_smile:

Sincerely,
Karen Fidanyan,
Master student at Moscow Inctitute of Physics and Technology,
Department of Molecular and Chemical Physics.

14.10.2015, 18:08, “Steve Plimpton” <[email protected]…24…>:

The next release of LAMMPS will have the Vashishta potential,
including the SiC parameter file. It might already be available on
github.