Thermostatting During Uniaxial Tension Simulation with Internal Crack and Velocity Bias

LAMMPS LAMMPS General Discussion
1

Hello,

I’m running a uniaxial tension simulation in LAMMPS where the sample contains an internal crack (created by deleting atoms). I’ve divided the system into three groups:

  1. Mobile group: This is the main deforming region, which includes the crack. I applied a velocity ramp in the y-direction to this group.
  2. Lower boundary group: This group is fixed in space.
  3. Upper boundary group: This group is given a constant velocity in the y-direction to apply the tension.

I’m using fix nvt to maintain temperature in the mobile group. Since I’m applying a velocity ramp, I learned that I should use compute temp/ramp and then apply fix_modify to avoid including the bias in the temperature calculation.

My main question is:

  • Should I apply any thermostat (like fix nvt) to the upper and lower boundary groups?

I’m currently assuming that only the mobile group should be thermostat-controlled, and the upper/lower boundaries should either be fixed or moved without temperature control.

Any clarification or best practices would be appreciated. Thanks in advance!

2

Hello,

I don’t think there is a right answer to that question, and if there is, this forum is probably not the best place to obtain it.

Also, when I read the question:

I wonder if you are asking whether you should apply a Nose-Hoover thermostat, or if you’re asking whether to integrate the equations of motion at all (versus leaving the atoms as a frozen block). But regardless, these choices can only be seriously answered by you, an expert, or a tutor.

Best,
Simon