[AMBER] Problem with QMMM umbrella sampling

From: Wook Lee <wlee.chemie.uni-wuerzburg.de>
Date: Mon, 14 Feb 2011 11:06:19 +0100

Dear Amber users

I am using Amber 11, and I want to calculate PMF as some bond is broken,
so I did umbrella sampling calculation in combination with QMMMMD with
DFTB, and I got some strange results.

The input scripts are shown below.

 &cntrl
  imin=0, ntb=0
  cut=12.0,
  temp0=310.0,
  ntt=3, gamma_ln=5.0,
  nstlim=100000, dt=0.001,
  ntpr=1000, ntwx=1000,ifqnt=1,nmropt=1
 /
 &qmmm
  qmmask='.4504-4514, 4952-4962, 12043-12055',
  qmcharge=-1,
  qm_theory='DFTB',
  qmshake=0,
 /
  &wt type='DUMPFREQ', istep1=10
/
  &wt type='END'
/
DISANG=restraint
DUMPAVE=dist21

restraint file :

&rst iat = 12043, 12044, r1=1.1,r2=2.1,r3=2.1,r4=3.1,rk2=300.00,rk3=300.00

The problem is that the distribution I got from this simulation is too
broad.
It sometimes deviates even 0.5 A at maximum from the reference value.
So I did another simulation with larger biased potential (rk=600), and
it still gave similar huge deviation.

The other strange thing is when I performed the exactly same calculation
in other computer cluster,
It showed a reasonable distribution, so I think this problem may related
to the compliation.
The version of the gcc in the cluster where I got problem is 4.3.3, and
the other is 4.2.2
Both are opteron clusters.

I would really appreciate it if anyone could help me with this problem.

Best regards,
Wook


_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Mon Feb 14 2011 - 02:30:02 PST
Custom Search