Delete atom in pair_style hybrid peri/lps lj/cut

LAMMPS LAMMPS Mailing List Mirror
1

Hello,

I have a question about the error I get after the deletion of an atom type for a peridynamics model using peri/LPS.

I am testing expansion into a material and the geometrical model is in the input file where it is identified the coordinate and the atom type.

I have three different type of atom and after the expansion, I delete the type 3.

The simulation is fine If I do not delete atoms, but adding the deletion I get this error : “ Divide by 0 in influence function of pair peri/lps”.

Am I missing something or am I doing something wrong in the input file or in the script? Any help is greatly appreciated.

Many thanks

Danilo

2

Hello,

[...]

Am I missing something or am I doing something wrong in the input file or in the script? Any help is greatly appreciated.

impossible to say from the limited information provided.

it certainly sounds like an usual model and it is questionable if the
hybrid approach will give you meaningful physics. not all inputs using
pair style hybrid that run without crashing are meaningful, there are
recurring discussions on this list about the implicit errors from
using hybrid to model alloys with different eam styles, for example.

axel.

3

Since PD atoms have bonds to other atoms, are you doing the deletion
in a way that leaves a dangling bond? The options in delete atoms
that refer to bonds know nothing about PD bonds. Those are special
“bonds” internal to PD.

Steve

4

Hello Steve and Axel,

Thanks for getting back to me.

I would like to delete not only atoms but also the bonds generated and linked to these atoms deleted.

“ variable sigmaF equal 1.05*${lat_const}

pair_style hybrid peri/lps lj/cut ${sigmaF}

pair_coeff 1 1 peri/lps {Ka} {Ga} {horizon} {saa} 0.25

pair_coeff 1*2 2 peri/lps {Km} {Gm} {horizon} {smm} 0.25

pair_coeff * 3 lj/cut 3.e6 ${sigmaF} “

I got the error related with peridynamic when I delete the type 3, but this type shouldn’t have any bond.

Kind regards

Danilo

5

Hello Steve and Axel,

Thanks for getting back to me.

I would like to delete not only atoms but also the bonds generated and linked to these atoms deleted.

you cannot. those are implicitly defined and handled inside the pair
style. as I mentioned before, I have doubts that your model is
physically sound.

“ variable sigmaF equal 1.05*${lat_const}

pair_style hybrid peri/lps lj/cut ${sigmaF}

pair_coeff 1 1 peri/lps \{Ka\} {Ga} \{horizon\} {saa} 0.25

pair_coeff 1*2 2 peri/lps \{Km\} {Gm} \{horizon\} {smm} 0.25

pair_coeff * 3 lj/cut 3.e6 ${sigmaF} “

I got the error related with peridynamic when I delete the type 3, but this type shouldn’t have any bond.

that assumes that the peridynamics pair style is programmed to
properly manage the case of a hybrid pair style, which is not
automatic.

that said, if you delete *all* type 3 atoms, the logical next step
would be to redefine the pair style to not use hybrid, but only
peri/lps and then just assign some dummy parameters to the no longer
existing type 3 atoms. that should force the pair style to
re-initialize and you *will* have a definitely more physically
meaningful model afterwards.

axel.