Segmentation Fault 11 on Modified oxView Topology

LAMMPS LAMMPS General Discussion
1

To anyone who helped me on my last problem with this, it turns out there was no instability issue after further investigation (as some of you may have suspected). Now, I have a real problem.

I am using LAMMPS 2Aug-2023, and trying to extend the link bond (the bond that links each beam of my 2-beam device) of my structure such that I have several origami of varying length. However, I will receive a Segmentation Fault 11 when nothing changes except this linker length is increased.

image
image928Ă—318 76 KB

An image of what I would like, as well as incrementally smaller and perhaps larger linkers.

This leads me to believe I have a GIGO problem, but I must take these steps of linker elongation to later do the analyses I need to do.

You can reproduce the issue if you like by using this script, replacing read_data file with this “output10_xt0.dat.lammps.txt” data file.

I was wondering if there were any “LAMMPS input tips” to make the simulation behave, such as the following which I’ve tried:

-changing the neighbor list skin distance

-“super-damping”: making my damp parameter lower than it already is, beyond 0.01

-changing the energy minimization parameters

output10.dat.lammps.txt (6.9 MB)
output10xt0.dat.lammps.txt (6.9 MB)
input2.lammps.txt (2.3 KB)

2

Your “working” input has loads of warnings:

WARNING: FENE bond too long: 1 33 32 1.6472519165498052 (src/CG-DNA/bond_oxdna_fene.cpp:233)
WARNING: FENE bond too long: 1 187 186 1.0536078984391466 (src/CG-DNA/bond_oxdna_fene.cpp:233)
WARNING: FENE bond too long: 1 324 323 1.0067544691507921 (src/CG-DNA/bond_oxdna_fene.cpp:233)
WARNING: FENE bond too long: 1 472 471 1.0073413001795313 (src/CG-DNA/bond_oxdna_fene.cpp:233)
WARNING: FENE bond too long: 1 574 573 1.0091894269061588 (src/CG-DNA/bond_oxdna_fene.cpp:233)
WARNING: FENE bond too long: 1 588 587 1.0091911778321336 (src/CG-DNA/bond_oxdna_fene.cpp:233)
WARNING: FENE bond too long: 1 630 629 1.0578004592979189 (src/CG-DNA/bond_oxdna_fene.cpp:233)
WARNING: Too many warnings: 1877 vs 100. All future warnings will be suppressed (src/thermo.cpp:478)

As you can see, from here on warnings are suppressed, but that does not mean, there are none.

This suggests that your initial geometry could be improved.

I cannot say whether this is also the reason for the segmentation fault. I cannot find a specific bug at the location where the segfault happens. I suggest you file a proper bug report issue on github at: Issues · lammps/lammps · GitHub and we notify the CG-DNA developer to have a closer look.

1 Like
3

“FENE bond too long” doesn’t seem like a valid simulation whatsoever to me.

FENE stands for finite extensible nonlinear elastic. The bond potential returns garbage forces when the bond exceeds its maximum extent (R0).

Your bond_style defines the maximum bond length as 0.7564. You clearly have one bond, and then many bonds, exceeding that length.

You need to define a different bond type for any bond that will exceed the FENE bond length limit you’re working with. And you need to make sure you absolutely, absolutely understand what you’re doing here. oxDNA is very different from most MD potentials (as it needs to be to accomplish its task) and you need to take a lot of personal responsibility for avoiding GIGO.

1 Like