AMBER Archive (2002)

Subject: Re: Equilibration output Amber7

From: Thomas Cheatham (cheatham_at_chpc.utah.edu)
Date: Fri Nov 22 2002 - 11:49:55 CST


> check COM velocity, temp: 0.000002 0.00(Removed)
...
> Can anyone explain me why it is coming like this?
> my input file is as follows............

You have NSCM = 1 which will remove center of mass translation (in a
periodic system or trans/rot in a non-periodic) every 1 step. The message
you see (as discussed on the reflector a few weeks ago) states that some
velocity was removed (not a lot). But, it is interesting that it builds
to this level every step. This is likely due to the application of the
position restraints. With position restraints on part of the system there
is no need to remove center of mass translation since this is done
implicitly.

As pointed out by Ganglei, there are some other peculiarities with your
input/output. One is the Ewald error is rather large (normally in the
10**-4 to 10**-6 range).

> nsnb = 25, dielc = 78.5, ntr = 1,

You would be wise to use the automatic heuristic pairlist updating (on by
default in AMBER7) rather than specify the pairlist update frequency
(NSNB=25 above) to better conserve energy.

It is not clear why you want a dielectric constant of 78.5; if you have
explicit water in there, this water will screen charge interactions. If
you set the dielectric constant to be 78.5, you will way overscreen charge
interactions (most notably at short range).

> nstlim = 1000, nscm = 1, t = 0, dt = 0.001,

Set NTCM = 0 since NTR = 1

> tol = 0.0001,

Make the SHAKE tolerance much lower to get better energy conservation at
little expense (in overall computer time)

TOL=0.0000001

> &end
> &ewald
> nfft1 = 64, nfft2 = 64, nfft3 = 64, order = 4,
> ew_type = 0, dsum_tol = 0.00001, nbflag = 0,
> eedmeth = 5,

You are using Ewald but also setting a distance dependent dielectric
(EEDMETH=5). This is likely why the Ewald error is so large. With a
distance dependent dielectric AND a dielectric constant of 78.5 you are
seriously damping out the direct space interactions and effectively you
electrostatics is mostly reciprocal interaction. This leads to a
misbalance between the "long" and "short" range forces.

If you want to run distance dependent dielectric, turn of the Ewald
(USE_PME = 0).

Also, set NBFLAG = 1 to use the automatic pairlist update for better
energy conservation.

Let me know if you have questions,

\ Thomas E. Cheatham, III (Assistant Professor) College of Pharmacy
| Departments of Medicinal Chemistry and of University of Utah
| Pharmaceutics and Pharmaceutical Chemistry 30 South 2000 East, Room 201
| & Center for High Performance Computing Salt Lake City, Utah 84112
|
| e-mail: tec3_at_utah.edu phone: (801) 587-9652 FAX: (801) 585-9119
\ http://www.chpc.utah.edu/~cheatham Offices: BPRP295A / INSCC 418