[gmx-users] Implementation of the v-rescale integrator

[Mario Fernández Pendás]

Dear users,

I am looking at the function vrescale_resamplekin() (in coupling.c)
that generates a new value for the kinetic energy according to
equation (7) from the original paper by Bussi et al in JCP (2007).

Maybe this is a stupid question or I am missing something, but I have
a doubt about the last "factor" term that appears multiplying inside
of the square root in the calculation of the new kinetic energy:
2.0*rr*sqrt(kk*sigma/ndeg*(1.0 - factor)*factor)

Is this here for ensuring that this whole term is zero in case tau is
very small? Is it here for numerical reasons?

Because, in the original formulation in the paper, the value that
appears is tau and this should be calculated as (1.0 - factor) instead
of as (1.0 - factor)*factor.

Thank you very much.

Best regards,
Mario