[lammps-users] langevin thermostat!

Dear support, I’m baffled about 24 factor in the espression

gfactor2[i] = sqrt(atom->mass[i]) *
sqrt(24.0*force->boltz/t_period/update->dt/force->mvv2e) /

why 24? I think that is 2.


My recollection is the 24 is correct when you sample with a uniform
random number rather than a Gaussian RN.


Dear Steve, I check that a uniform distribution has variance (b-a)**2/12.

Best Regard.

Nicola Varini

2008/11/10 Steve Plimpton <[email protected]>