Re: [AMBER] High temperature at the beginning

From: case <case.biomaps.rutgers.edu>
Date: Thu, 24 Feb 2011 07:51:31 -0500

On Thu, Feb 24, 2011, A. Gomez-Sicilia wrote:
>
> I am doing simulations about protein stretching, and using a script so that
> the output is fragmented in pieces. This makes a new simulation start every
> few steps, using as input the output of the last.

> ================mdin====================
> &cntrl
> nmropt = 0, ntx = 1, irest = 0,
              ^^^^^^^^^^^^^^^^^^^^

You need to set ntx=5 and irest=1 to restart a simulation from the output of a
previous run.

> ntrx = 1, ntxo = 1,
> ntpr = 100, ntwx = 100, ntr = 1,
> ntf = 2, ntb = 0, dielc = 1.0, ntc = 2,
> cut = 12, nsnb = 10, scnb = 2.0, scee = 1.2,
  ^^^^^^^^^^

There will also be some discontinuity on restarts with this cutoff. You might
try some test runs with a larger value, then experiment to see how much of a
problem this might be.

> imin = 0, igb = 1, gbsa = 1,
> nstlim = 2000, dt = 0.001,
> temp0 = 300, tempi = 300, saltcon=.15,
> ntt = 1, ig = -1, vlimit = 20.0,
> &end
> [Some constrains]
> END

I think the basic problem is just with the printout: you have a non-restart
run, and are assigning random initial velocities. The bonds that are
constrained by SHAKE actually have kinetic energy at step 0, so appear to
contribute to the temperature. After the constraints are applied to the
velocities, which doesn't happen until later steps, the temperature comes
down. Of course, the usual advice about running test simulations with ntpr=1
is very valid here -- it will help you see a lot about what is going on.

....dac


_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Feb 24 2011 - 05:00:03 PST
Custom Search