Reax/C and adding atoms

The problem is most likely resulted from a bad neighbor list, but I am still unclear what the solution is.

Meanwhile, you can use pair_style reax, which works well with your script.

Hi Ray,

Are there any updates on the issue? Just checking in (for as long as required I’ll just use the old implementation).

Kristof

Adding/creating isolated gas phase molecules in a very big box is straining reax/c’s memory allocation. If you have a real simulation problem (deposition of gas molecules onto a surface, etc) and run into problem, I would be more willing to track it down.

Ray

Dear Kristof,

I think I have found a workaround. Add these four lines before the second run command:

pair_style none
pair_style reax/c lammps_control safezone 15 mincap 1000
pair_coeff * * ffield.reax.cho C H
fix 9 all qeq/reax 1 0.0 10.0 1.0e-6 reax/c

Best wishes,
Michal

2014-10-07 17:44 GMT+02:00 Ray Shan <[email protected]>:

Adding/creating isolated gas phase molecules in a very big box is
straining reax/c's memory allocation. If you have a real simulation
problem (deposition of gas molecules onto a surface, etc) and run into
problem, I would be more willing to track it down.

Of course the supplied example input is not the "real" system I'm studying,
which indeed consists of deposition on a surface, but then that of a
nanoparticle. This of course means I have to use a ratger large box to
correctly model an isolated gas-phase nanoparticle.

Hi Michal,

Thanks for taking the time to help! I also tried this option, and although it might work for the example input, it does not for my “real” system, even when using ridiculous large values for the mentioned options. I’ll fiddle a bit more with it, though (for now, this kind of simulations is luckily still in the “early trying” phase).

Kristof

2014-10-07 17:44 GMT+02:00 Ray Shan <[email protected]>:

Adding/creating isolated gas phase molecules in a very big box is
straining reax/c's memory allocation. If you have a real simulation
problem (deposition of gas molecules onto a surface, etc) and run into
problem, I would be more willing to track it down.

Of course the supplied example input is not the "real" system I'm
studying, which indeed consists of deposition on a surface, but then that
of a nanoparticle. This of course means I have to use a ratger large box to
correctly model an isolated gas-phase nanoparticle.

If your real system simulation runs into problem, please post it here and I
can take a look.

Ray

2014-10-14 17:24 GMT+02:00 Ray Shan <[email protected]>:

2014-10-07 17:44 GMT+02:00 Ray Shan <[email protected]>:

Adding/creating isolated gas phase molecules in a very big box is
straining reax/c's memory allocation. If you have a real simulation
problem (deposition of gas molecules onto a surface, etc) and run into
problem, I would be more willing to track it down.

Of course the supplied example input is not the "real" system I'm
studying, which indeed consists of deposition on a surface, but then that
of a nanoparticle. This of course means I have to use a ratger large box to
correctly model an isolated gas-phase nanoparticle.

If your real system simulation runs into problem, please post it here and
I can take a look.

The relevant part of the input is attached (also added the ffield file as
this one isn't included in the LAMMPS distribution)

Kristof

ch3.txt (219 Bytes)

in.ch3 (1.67 KB)

ffield.reax.CHNi (5.85 KB)

Loop time of 37.6001 on 8 procs for 10000 steps with 126 atoms

Your script ran fine - of course I had to correct some mistakes and make some changes, but otherwise it works.

How did it not work for you and what is the error you saw?

Ray

Interesting. What were those mistakes? I saw a bondchk error when I tried it.

Kristof

Use safezone and mincap. Get rid of the control file. Look at your group definitions and see if they make any sense. Look at what molecule you are actually depositing.

Since there is not a problem with reax/c and fix deposit for real simulations, I consider this thread closed.

Ray