On Wed, Oct 4, 2017 at 12:16 PM, R zu <rzu512.gmail.com> wrote:
> I am sorry about the point of Berendsen barostat.
>
> One comparison of the barostats says it doesn't give the correct average.
>
It gives the correct average pressure, but the wrong average isothermal
compressibility (which is a function of the size of the fluctuations, which
are incorrect with the Berendsen barostat).
>
> http://pubs.acs.org/doi/pdf/10.1021/acs.jctc.5b00748
>
> The reference 5 and 6 link to the the flying ice cube problem, and
> collective motion artifacts.
>
Flying ice cube can be corrected by removing center-of-mass motion
periodically, which is done by default within Amber.
> I guess the problem is that it would take forever to get the correct
>
> average?
>
Assuming you mean correct average compressibility, then no. The problem
is not that it would take forever to get the correct average -- the problem
is that it doesn't sample from the isobaric-isothermal ensemble. So the
limiting distribution has an incorrect thermodynamic average for the
compressibility. So even if you wait forever, the average compressibility
will still be wrong.
HTH,
Jason
--
Jason M. Swails
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Oct 04 2017 - 13:00:03 PDT
