Inquiry on "1-3, 1-4 Bond Count Inconsistent Error"

Dear Dr. Axel Kohlmeyer,

Hello Dr. Kohlmeyer. Thank you for your many advice.

I would like to ask a question in regards to an error that I am obtaining in LAMMPS when I simulate a stress strain curve of a polymer matrix. I have attached the short version of my LAMMPS code that elongates my simulation box and also my input data file that the attached LAMMPS code reads. The input data file contains information on 25 chains where each chain consists of 50 atoms. Therefore, the simulation box contains a total of 1250 atoms and this simulation box is subjected to elongation to simulate a stress strain curve.

The attached data file and LAMMPS code works fine. However, when I increase the number of chain to 250 and the number of atoms on each chain to 500, I am obtaining the following error:

ERROR on proc 112: 1-3 bond count is inconsistent (…/special.cpp:287)
application called MPI_Abort(MPI_COMM_WORLD, 1) - process 112
ERROR on proc 45: 1-3 bond count is inconsistent (…/special.cpp:287)
ERROR on proc 61: 1-3 bond count is inconsistent (…/special.cpp:287)
application called MPI_Abort(MPI_COMM_WORLD, 1) - process 45
application called MPI_Abort(MPI_COMM_WORLD, 1) - process 61
ERROR on proc 67: 1-3 bond count is inconsistent (…/special.cpp:287)
application called MPI_Abort(MPI_COMM_WORLD, 1) - process 67
ERROR on proc 93: 1-3 bond count is inconsistent (…/special.cpp:287)
application called MPI_Abort(MPI_COMM_WORLD, 1) - process 93

According to the previous LAMMPS documentation, there must be something wrong with my defined bond topologies. Looking at the previous mailing list, it might be due to bad formatting. However, the short attached code that I have written works fine, and I do not see any bad formatting in my data file that contains 250 chains each consisting 500 atoms.

I would greatly appreciate if I could request an advice in resolving this bond count inconsistency issue. Thank you

Sincerely,

Masato Koizumi

0_0238_rho_HET_20_HET_15 (245 KB)

Strain_Rate_Loading_10mu.in (934 Bytes)

the attached example already produces an “Inconsistent image flags” warning. this should be taken serious for the kind of system you are modeling.
reading the data file into a visualization program confirms, that there are misplaced atoms due to incorrect image flags. thus i would conclude that the problem is most likely with the software that produces the data file. the error for the larger system can have the same origin.

axel.

Dear Dr. Axel Kohlmeyer,

Hello Dr. Kohlmeyer. Thank you for your advice. I have attached another data file and the same LAMMPS code that simulates a stress strain curve of a polymer matrix. Using this data file, I do not see any “Inconsistent Image Flag” warnings when I run this short LAMMPS code. The only difference between this new data file that does not result in “Inconsistent Image Flag” and the previous data file that gave the “Inconsistent Image Flag” warning is that latter data file contains additional new bonds formed between different chains.

I would like to ask why simply adding new bonds between atoms result in inconsistent image flags.

The atoms have not moved at all - just simply new bonds were created. Therefore, from my understanding, if the former data file did not result in “Inconsistent Image Flag,” then the latter data file should also not output the warning.

Once again, thank you so much for your many advice.

Sincerely,

Masato Koizumi

0_rho_Single_HET (245 KB)

Strain_Rate_Loading_10mu.in (926 Bytes)

Dear Dr. Axel Kohlmeyer,

Hello Dr. Kohlmeyer. Thank you for your advice. I have attached another
data file and the same LAMMPS code that simulates a stress strain curve
of a polymer matrix. Using this data file, I do not see any "Inconsistent
Image Flag" warnings when I run this short LAMMPS code. The only difference
between this new data file that does not result in "Inconsistent Image
Flag" and the previous data file that gave the "Inconsistent Image Flag"
warning is that latter data file contains additional new bonds formed
between different chains.

I would like to ask why simply adding new bonds between atoms result in
inconsistent image flags.

​because this is what is being tested for, i.e. that for both atoms of a
bond, the image flags result that the closest images are being chosen.​

The atoms have not moved at all - just simply new bonds were created.
Therefore, from my understanding, if the former data file did not result in
"Inconsistent Image Flag," then the latter data file should also not output
the warning.

​that is not correct logic. you have removed bonds that connected groups of
atoms with consistent image flags inside each group, but not consistent
with each other.​ this can be easily verified through visualization. the
previous data file showed those "bad" bonds which were unusually long, now
those are gone. in short, this doesn't prove anything.

axel.

Dear Dr. Axel Kohlmeyer,

Hello Dr. Kohlmeyer. Thank you for your many advice.

I have attached a data file and the short LAMMPS code that simulates a stress strain curve of a polymer matrix. I have resolved the previous “Inconsistent Image Flag” warning such that the output file no longer produces this warning. However, I when I simulate a larger polymer matrix (using this same revised LAMMPS code) that consists a total of quarter million atoms, I start to see the warning, “Dangerous builds = 15” or a “Dangerous builds” equaling to a number greater than zero.

I understand that I need to increase the skin distance or have the neighbor list be updated more frequently.

I was curious if I could ask why increasing the number of atoms would have effect on increasing the skin distance or more frequent neighbor list. I would greatly appreciate if you could guide me to a source that discusses this.

Thank you.

Sincerely,

Masato Koizumi

0.01292_rho_Single_HET (219 KB)

Strain_Rate_Loading_10mu.in (934 Bytes)