On Tue, Mar 01, 2011, Ali M. Naserian-Nik wrote:
>
> Thank you very much for your comments. I think the structural differences at
> the end of the two test simulations are not negligible (I attach the
> corresponding PDB files). Is there any way to reduce the errors as much as
> possible?
> Also here is my input file:
>
> &cntrl
> imin=0, ntx=7, irest=1, ntb=0,
> ntc=2, ntf=2,
> cut=150, igb=1, extdiel=79, saltcon=0.13, gbsa=1,
> ntpr=100, ntwx=100, ntwr=10,
> tempi=300.0, temp0=300.0, ntt=3, gamma_ln=3, ig=-1,
^^^^^^^^^
> nstlim=40000, dt=0.001,
> nscm=10000000,
> &end
Using a random thermostat will certainly make the two short simulations
different than one longer one. Half way through, you are resetting the random
number generator in one case, and not in the other. I don't think Amber has
any good way of avoiding this problem: you have (in any case) it simulate long
enough so that the statistical properties of the resulting conformational
ensemble are independent of starting (or restarting) conditions.
....dac
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Mar 01 2011 - 05:00:04 PST
