AMBER Archive (2005)

Subject: Re: AMBER: Temperature regulation

From: David A. Case (case_at_scripps.edu)
Date: Tue Aug 30 2005 - 17:45:13 CDT


On Tue, Aug 30, 2005, Monica Civera wrote:
>
> we perform MD simulation with AMBER 7.0 in NVT conditions at the unusually
> high temperature of 548 K. (see input below)
>
> We notice that the average temperature during our 10 ns long run is around
> 564 +/- 4 K. It seems to be equilibrated but not around the selected value.
> The system, consisting of a 64-residues protein in TIP3P waters, was
> pre-equilibrated at 498 K. Higher temperature NTP simulations are not
> possible due to the breakdown of the entire system (evaporation?).
>
> Could the problems be in the equilibration (than in the artificial density)
> or in the thermal bath coupling constant used?
>
>
> &cntrl
> imin=0, ntx=5, irest=1, ntpr=500, ntwr=500, ntc=2,ntwx=500,
> ntf=2, ntb=1, nstlim=100000, dt=0.001, ntt=1, cut=8,temp0 = 548.,
> tautp=2.0,taup=2.0
> &end
>
> &ewald
> use_pme=0,eedmeth=4
> &end

You are using a short cutoff (8 Ang.) but are turning off PME. This means that
you will have significant heating effects from atoms moving in and out of the
cutoff region. This heating can be partially offset by a thermostat with
tautp = 2.0 psec, but not fully: you are seeing the expected competition
between unphysical heating (caused by cutoff artifacts) and the thermostat.

Although I have no good idea of what you are really trying to do here, if you
want accurate results you cannot use a non-bonded cutoff in this way. If you
are just doing an qualitative and exploratory high-T simulation, then this
might be OK; but you really lose some of interesting properties of water when
you ignore long-range interactions.

...good luck...dac

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu