" pair distance < table inner cutoff" error in brownian dynamics simulation

LAMMPS LAMMPS Mailing List Mirror
1

Dear Axel:
Thanks for your suggestion! I used pair_write commands to check my potential table file, indeed it agrees well with the potential in the reference. I am confused about “you have chose the wrong sign somehow”, is it that the sign of force in the table file? What troubled me is that I treated the nanoparticle as an atom, but in the reference the nanoparticle has 6nm diameter, the potential range between two nanopartilces is 66A to 120A, In lammps, if I apply atomic atom_style, the atom has no size, their interaction range can be 0A to 120A, which the range of 0A-66A is absent in my table file. In browian dynamics simulation, the particles can move randomly, thus most likely the diatance between particles will be smaller than 66A, then ecounting the error “pair distance<table inner cutoff”. Actually I have tried to use sphere atom_style, which the sphere diameter is 60A(6nm), but I encounted other errors. and I wonder if I can use sphere atom_style to figure out this problem? can you give me some suggestion? thank you very much
best wishes

Xuepeng liu

2

Dear Axel:
Thanks for your suggestion! I used pair_write commands to check my
potential table file, indeed it agrees well with the potential in the
reference. I am confused about "you have chose the wrong sign somehow", is
it that the sign of force in the table file? What troubled me is that I

yes, that happens to people occasionally and then particles get
attracted instead of repelled. but i downloaded your inputs and table
file and can confirm that this is not the case in your input.

treated the nanoparticle as an atom, but in the reference the nanoparticle
has 6nm diameter, the potential range between two nanopartilces is 66A to
120A, In lammps, if I apply atomic atom_style, the atom has no size, their
interaction range can be 0A to 120A, which the range of 0A-66A is absent in
my table file. In browian dynamics simulation, the particles can move
randomly, thus most likely the diatance between particles will be smaller
than 66A, then ecounting the error "pair distance<table inner cutoff".
Actually I have tried to use sphere atom_style, which the sphere diameter is
60A(6nm), but I encounted other errors. and I wonder if I can use sphere
atom_style to figure out this problem? can you give me some suggestion?

there are two issues with your input. the first is that you have very
aggressive neighbor list settings, more suitable to reduced unit
calculations. given the diameter of your particles, you should use a
much larger skin size. since your particles can move a much larger
distance per step.

the second issue - and the reason why it works with lj/cut but not
with your table, is that you have practically no repulsion, but run a
simulation at fairly high temperature. at that point, particles will
bump into each other and have too much momentum to be repelled by the
repulsive part of the potential.

the tabulation seems to be more suitable for nanoparticles in solution
(where they rarely will come close or move as fast relative to their
diameter).

axel.

pot-compare.png

3

Dear Axel:
I am very apprecited for your uesful help! Following your suggestions, I increased the skin size to 500 bin, and timestep decreased to 0.005 ps, for the nanoparticles has angular momentun, I used “sphere atom_style”, setting 60A (6nm) diameter of nanoparcticle and mass m=1.0. and the particles in a 2d lattice of hex, eventually it did work! I am very sorry to have three more questions, First, how can I insert a large amounts of particles into system randomly(not creat atoms in lattice style),Does any commands can finish it? Second, in my primary simulation, my timestep is very large probabely due to the wrong unit conversion, in the reference, the author stated as follows:" the diameter of nanoparticle is 6nm, and in Brownian dynamics simulation, we set particle diameter d=1 and measure all distances in units of d, we set timestep is 0.005 in reduced units of d(m/KT)^(1/2) with m=1." if the temperature is 300K, how can I make unit conversion of “metal units” in this case? Third, you said “the tabulation seems to be more suitable for nanoparticles in solution(where they rarely will come close or move as fast relative to their diameter).” ,is it means that my table potential is better than lj potential in the case of nanoparticles in solution? but my table file is absent in the range of 60A-66A! and do not have stronger repulsion as lj .Do you think if it will make a big difference due to the absence of a short potential range? looking forward to your reply, thank you very much!
the latest input file is as follows, and the table file remains unchanged.

------------------------ INITIALIZATION ----------------------------

units metal
atom_style sphere
dimension 2
boundary p p p

neighbor 500.0 bin

neigh_modify every 1 delay 3 check yes

variable latparam equal 75

----------------------- Colloids DEFINITION ----------------------------

lattice hex {latparam} region box block 0 50 0 50 0 1 create_box 1 box lattice hex {latparam} orient x 1 0 0 orient y 0 1 0
create_atoms 1 box

set type 1 mass 1.0

set type 1 diameter 60.0

velocity all create 1.0 482748 dist uniform

------------------------ FORCE FIELDS ------------------------------

pair_style table linear 9001
pair_coeff * * NPaticle.table NPs_NPs 120
#pair_write 1 1 18001 r 50.0 120.0 table1_6.c NPs_NPs

EQUILIBRATION

minimize 1.0e-12 1.0e-12 100 1000

compute mytemp all temp
compute 1 all gyration

timestep 0.005

fix 1 all langevin 300.0 300.0 100.0 12345

fix 2 all nve/sphere

dump 1 all atom 500 dump.lammpstrj

thermo_style custom step c_mytemp c_1 epair etotal press

thermo 500

run 20000

best wishes
Xuepeng liu

4

Dear Axel:
I am very apprecited for your uesful help! Following your suggestions, I
increased the skin size to 500 bin, and timestep decreased to 0.005 ps, for
the nanoparticles has angular momentun, I used "sphere atom_style", setting
60A (6nm) diameter of nanoparcticle and mass m=1.0. and the particles in a
2d lattice of hex, eventually it did work! I am very sorry to have three

but i doubt that using atom style sphere has anything to do with that.
since you have point potentials w/o friction terms or dipoles, there
is no effect from that.

more questions, First, how can I insert a large amounts of particles into
system randomly(not creat atoms in lattice style),Does any commands can
finish it? Second, in my primary simulation, my timestep is very large

you can create atoms at random positions. you can use pair style soft
and fix adapt to unoverlap them. you can also directly delete
overlapping atoms. that is a FAQ. check the archives and the
documentation.

probabely due to the wrong unit conversion, in the reference, the author
stated as follows:" the diameter of nanoparticle is 6nm, and in Brownian
dynamics simulation, we set particle diameter d=1 and measure all distances
in units of d, we set timestep is 0.005 in reduced units of d(m/KT)^(1/2)
with m=1." if the temperature is 300K, how can I make unit conversion of
"metal units" in this case? Third, you said "the tabulation seems to be more

that is your job to figure out. i don't have the time to do other
people's legwork in their research.

suitable for nanoparticles in solution(where they rarely will come close or
move as fast relative to their diameter)." ,is it means that my table
potential is better than lj potential in the case of nanoparticles in
solution? but my table file is absent in the range of 60A-66A! and do not
have stronger repulsion as lj .Do you think if it will make a big difference
due to the absence of a short potential range? looking forward to your
reply, thank you very much!

this is not exactly a smart question. the lj/cut potential goes from 0
to 120 with 120 cutoff. i only plotted a subject to highlight the
effect of the repulsion. if i had plotted the entire range, you would
not have seen anything. you already saw what the effect of leaving out
some of the repulsive part when running your test. however, this is
only the interaction between nano particles, and - as i already stated
- if you put a solvent in between them, they are not likely to come
that close as in vacuum.

as for the rest of the question, i don't have time to speculate and i
don't have a crystal ball that can predict simulation results. you
have to figure it out just like everybody else. step by step. that is
the nature of the business.

axel.

5

Dear Axel:
Thank you very much for your help! I will figure all these problems out step by step.
best wishes
Xuepeng liu