Oscillatory shear/fix deform

Hello all,

I am using wiggle in fix deform to apply oscillatory shear to a box of dilute particles. I’m also using the fix nvt for time integration. To ensure the constant temperature during the oscillations I use a small tdamp (~5 time steps). Just after shear simulation starts I see a raise and then oscillation in total energy that I suspect affects the results. I might be doing some thing wrong in the settings or using not right fix for time integration. I’d appreciate if some one comments on this. Below is a copy of my script for a strain of 5% and time period of 0.1 tau.



units lj
atom_style angle
dimension 3
read_restart restart_b100.md
change_box all triclinic

neighbor 0.3 bin
neigh_modify delay 0 every 1 check yes

bond_style harmonic
bond_coeff 1 1000 1.2

angle_style harmonic
angle_coeff 1 400 180

pair_style lj/cut 2.5
pair_coeff 1 1 0 1.
pair_modify shift yes
pair_coeff 1 2 0.01 1. 1.122
pair_coeff 2 2 0.01 1. 1.122

variable tmp equal ly
variable L0 equal ${tmp}

fix 1 all nvt temp 1.0 1.0 0.00025
fix 3 all deform 1 xy wiggle 5 0.1 units box

variable strain equal xy/v_L0

compute peratom all stress/atom
compute p all reduce sum c_peratom[1] c_peratom[2] c_peratom[3] c_peratom[4] c_peratom[5] c_peratom[6]
variable stress equal (c_p[4])/(3*vol)

fix 4 all ave/time 1 5 5 v_stress file stress_ave2.dat
thermo_style custom step temp etotal press v_stress v_strain f_4
thermo 5
timestep 0.00005

Why wouldn’t you expect oscillations in energy if you are oscillating the box
size? Also I suggest you first do this w/out a thermostat, then try it with
a very mild thermostat (big time period).


Thanks Steve for your suggestion. I perfectly expect oscillations in energy proportional to box size oscillations. I meant some quick oscillations within the main shear oscillation.
For a run w/out thermostat I use fix nve. While applying shear, temperature raises constantly until the run blows up. A thermostat of 100 t/step frequency with fix nvt solves this issue. I am wondering what I should expect in rheology results with the any of the following three combinations for a same system and if there is a proper combination for polymeric (fluid) systems:

fix nvt + fix deform wiggle remap x
fix nvt + fix deform wiggle remap v
fix nvt/sllod + fix deform wiggle remap v

The difference of remap x,v are clearly described in documentation, but not combined with fix nvt. I couldn’t find an example of oscillatory shear in the example section of LAMMPS, does anyone have an example script of oscillatory shear that can share here?



There is nothing specific about fix deform wiggle and fix nvt
that connects them to each other. I.e. they operate completely
independent of each other. You can use fix nvt (or any of the

thermostats, like fix langevin) with fix temp/deform assigned
to it (see fix modify temp) and they the thermostatting
will operate slightly differently. See Section 6.16 for a discussion
of biasing in thermostats.

Fix nvt/sllod will also assume the atoms are moving at the deformation

rate (i.e. wiggling in your case) and try to coerce them into doing that

See the fix nvt/sllod doc page for details.

If any of the thermostats show markedly different results (in a statistical
sense, not an exact sense), then you are probably pushing the system
too hard and hoping a thermostat will “fix” it. Which means you’re
on thin ice to begin with.


Ditto on Steve’s comment, esp in regards to time-response to forcing. Essentially, you must assure that you are simulating close to equilibrium dynamics and not truly soft-matter. (unless thats what’s you’re going for)

Dear All,

Thanks Eric and thanks Steve for your comments.
I ran this system in nve without any deformation to determine the equilibrium dynamics (temperature). However during shear, I actually put energy into system which heats it up so it’s not possible to work without a thermostat. So using equilibrium temperature from nve run, Nosé–Hoover (nvt/sllod), velocity remapping and fix deform/wiggle I get the following result for one cycle where I see other strange oscillations apart from main stress response. This to me doesn’t seem like a non-linear response. Is it possibly an error in computing stress?



stress.pdf (15.2 KB)

I’m not sure what your Q is. There is a kinetic energy component to stress.
So if you are wiggling the box and atoms, the stress will reflect that.