Flying Ice Cube

Hi Users,

I believe I am experiencing the flying ice cube problem during equilibration. I’m using NPT and equilibrating a solid state protein structure. I’ve used packmol to build the initial conformation.

I’m wondering if there is a common procedure to stop the flying? Would “fix recenter” be applicable?

Thanks.

No body can tell why is your system drifting without more information. Please make sure the initial velocities are zero before velocity create and you have a large enough system. Check your structures, too.

Also I believe the “flying ice cube problem” refers to a cold object moving at a high velocity, and therefore corrections (in particular removing the translational center-of-mass velocity) have to be made to estimate the object’s temperature. It may have very little to do with your drifting.

Ray

Hi Users,

I believe I am experiencing the flying ice cube problem during
equilibration. I'm using NPT and equilibrating a solid state protein
structure. I've used packmol to build the initial conformation.

I'm wondering if there is a common procedure to stop the flying?

the common procedure for this (and for *all* problems with lammps) is
to first understand the cause and then apply a suitable solution.
reason for the "flying icecube syndrome", i.e. a growing total
translational momentum of your simulated system is that you introduce
a non-zero total translational momentum through some reason and it is
not dissipated.

a net translational momentum can be introduced through the velocity
initialization, truncation errors in the force calculation, the
timestepping. thermalization through velocity rescaling (fix
temp/rescale or fix temp/berendsen) will enhance it, dissipative
thermalization (fix langevin, pair_style dpd/tstat) can reduce it.

Would "fix recenter" be applicable?

no, of course not. it will change coordinates but not the momentum.

axel.