Dear all users:
I am trying to run on LAMMPS (29-Jul11) a simple simulation where a silicon block is relaxed. I read this mailing list and it turns out that ReaxFF potential is not able to be included together with another potential, i.e., it is not possible to include the hybrid keyword in the pair_style command. This is clearly specified in the manual information for either reax or reax/c, and also in a couple of posts in the mailing lists. In my case I obtain the following error message 'ERROR: Incorrect args for pair coefficients' when using
I have tracked down this error, and appears in the method void PairHybrid::coeff(int narg, char **arg) of pair_hybrid.cpp. I also have reviewed the files pair_reax/c.cpp and pair_reax/c.h, and tried to modify the member function void PairReaxC::coeff( int nargs, char **args ) in order to accept one more argument, but this trick was unsuccessful.
Therefore, I have two questions:
1. Where should I start in order to enable reax/c to be part as a hybrid style? In the manual it says "Some additional work still need to be done to enable this".
2. In addition, find attached a file that I downloaded online where Si parameters are given. I am not sure if they are correct, I am running relaxations to reproduce basic thermodynamic parameters. Then, is there any way to get a verified list of such parameters? I did not find a file in the examples/reax folder containing silicon.
Best regards,
- Fernan
ffield.reax.budzienSi (21.6 KB)
in.reaxc.hybrid (2.26 KB)
1. Where should I start in order to enable reax/c to be part as a hybrid style? In the manual it says "Some additional work still need to >be done to enable this".
This is not a simple cosmetic change. The changes needed in both
ReaxFF and ReaxFF/C are fundamental,
since neither were designed initially to work on just part of the
system. Aidan can comment further.
2. In addition, find attached a file that I downloaded online where Si parameters are given. I am not sure if they are correct, I am running > relaxations to reproduce basic thermodynamic parameters. Then, is there any way to get a verified list of such parameters? I did not >find a file in the examples/reax folder containing silicon.
This has been discussed in the mail list before. We only provide
ReaxFF potential
files for the elements we have worked with and verified. There are many ReaxFF
parameterizations out there, but it is up to users to test themfor
their systems.
Again Aidan can comment about any plans to include more potential files,
but I think he hesitates to do so.
Steve
I concur with Steve's responses. Implementing hybrid style for either reax
or reax/c is not straightforward. Perhaps you can achieve your goals by
some other means. We do not attempt to validate ReaxFF files that people
send us. We only vouch for the correctness of the existing files and
settings that we provide with the LAMMPS distribution.
Aidan
Hi all:
In the case of silicon, van Duin et al (2003) specified that only general contributions are needed (bond, overcoordination, vdWaals, valence, and charge). I suppose that I will have to write down my own version of ReaxFF fitted for silicon. When you said 'by some other means', do you have any other recommendation than doing so? I have not seen any related message in this forum.
Thank you,
- Fernan
Originally you said you wanted to model Si. If that is all you
want to do, then I don't see why you would need pair_style hybrid
and multiple potentials. You just want ReaxFF for Si, as I see it.
I suppose that I will have to write down my own version of ReaxFF
fitted for silicon.
If what you mean by this is that you need to create a ReaxFF
potential file for Si in the format LAMMPS can use, then I agree.
If you are successful and validate it in some manner (e.g. with
help from Adri), then we can release that file with LAMMPS,
with your name on it.
Steve