Segmentation fault when using Kolmogorov/Crespi/z with REBO

Hi,

I am trying to energy minimize a system of few layer graphene. The interaction between the atoms is modeled as a combination of REBO potential and Kolmogorov/crespi/z. My simulations work when there are only 2 layers of graphene.

However, when there is more than 2 layers, I am getting a segmentation fault. I checked to see if I have made any mistakes in my code. But the documentation for Kolmogorov/crespi/z was not clear on how to use it when there are multiple layers. I just want to make sure that this is not a bug within lammps

I have attached the code (in.equil) and the data file (Gr_4layer_AB.data) along with the potential file. I am using the latest stable lammps release (Aug 11 2017).

Thanks a lot for your help.

Regards,

report_error.tar.gz (64.9 KB)

Hi,

I am trying to energy minimize a system of few layer graphene. The
interaction between the atoms is modeled as a combination of REBO potential
and Kolmogorov/crespi/z. My simulations work when there are only 2 layers
of graphene.

However, when there is more than 2 layers, I am getting a segmentation
fault. I checked to see if I have made any mistakes in my code. But the
documentation for Kolmogorov/crespi/z was not clear on how to use it when
there are multiple layers. I just want to make sure that this is not a bug
within lammps

I have attached the code (in.equil) and the data file (Gr_4layer_AB.data)
along with the potential file. I am using the latest stable lammps release
(Aug 11 2017).

​you should contact the author of this pair style how to deal with this
case. it looks to me as if you scenario is not considered, i.e. the pair
style has no provisions for handling NULL atom types. you could avoid the
segfault by using arguments "C C C C" to map all types to the only
available set of types, but i have no idea, whether that will result in the
correct behavior. again, contacting the author is your best shot at
getting an authoritative answer.

axel.​

after some more thinking, i believe the way you have to enter the potential is the following:

pair_style hybrid/overlay rebo kolmogorov/crespi/z 14.0 kolmogorov/crespi/z 14.0 kolmogorov/crespi/z 14.0
pair_coeff * * rebo CH.airebo C C C C

pair_coeff 1 2 kolmogorov/crespi/z 1 CC.KC C C NULL NULL
pair_coeff 1 3 kolmogorov/crespi/z 2 CC.KC C NULL C NULL
pair_coeff 3 4 kolmogorov/crespi/z 3 CC.KC NULL NULL C C

because you want to apply the kolmolgorov/crespi term independently for these three pairs of types and not combine them.

axel.

Axel,

Yes, I want to apply kolmogorov/crespi for each pairs of types separately.

The syntax in your last email works without any error and I believe that is the model that I want for my system.

Thanks a lot for your help!

Suggestion: Updating the documentation for kolmogorov/crespi/z with the details in your last email will be useful for other lammps users in the future.

Thanks