Re: AMBER: MD problem!

From: darden <>
Date: Wed, 17 Sep 2003 10:29:50 -0500 (CDT)

Dear Pradipta
Its hard to tell without more info. However as a general policy, its best
to first minimize to relieve bumps (with a new box you can have collisions
across the box boundary), then run constant temperature to get up to 300K,
then run (using a restart from the constant temp run) constant pressure to
get up to correct density. You could for example run ~100 steps
conjugate gradient minimization (keep shake on if you like although this
will prevent minimization to a tight minimum--its ok you just want to
relieve bumps). Next run NVT for ~10-20ps with default temperature
control, this will get you to room temperature and listtot will not change
much. Finally turn on NPT and run for ~10-20ps to get density right.

If the latter fails due to too fast a box change, which is unlikely, you
can always as a last resort run a series of short NVT runs, each time
with a slightly smaller box to get to the right density.

This latter is a little more involved:

Once you have an inpcrd with box coordinates in the last line (you will
see 6 numbers; 3 identical that look like a box size (~20 for 256 waters)
followed by
90.0000000 90.0000000 90.0000000
a) run 1 step of NPT to get the density. Toss the restrt file.
If density is too low slightly decrease (by ~1%) the box size numbers in
the last line, keeping them identical; leave the 90.00000 alone
Do the modification with your favorite text editor.
Run ~50 steps of NVT. If it doesn't blow up keep the restrt file and
repeat a). If it heats up a bit run the NVT longer. If it blows up make
the density change smaller. Once you get within 0.05 of correct density
you can certainly run NPT.
Notes: This latter procedure will also work for protein simulations---use
cartesian restraints on the protein to prevent it from moving--for that
matter always restrain your solute until you get close to the "right"
density (exact value uncertain but close to 1 gm/ml)
Hope this helps
Tom Darden
 On Tue, 16 Sep 2003, Pradipta Bandyopadhyay

> Hi,
> I am running short MD simulation for 256 TIP3P water molecules. My MD
> simulation crashed after 1.16 pico second. I checked that the listtot
> variable in ewald - it is increasing with time (129153 to 151652). The
> program says floating point exception after listtot becomes 151652.
> Am I doing something stupid?
> thanks.
> my last dynamics step is shown below:
> NSTEP = 1160 TIME(PS) = 1.160 TEMP(K) = 286.80 PRESS = -497.9
> Etot = -2045.9988 EKtot = 436.8448 EPtot = -2482.8436
> BOND = 0.0000 ANGLE = 0.0000 DIHED = 0.0000
> 1-4 NB = 0.0000 1-4 EEL = 0.0000 VDWAALS = 384.0900
> EELEC = -2866.9336 EHBOND = 0.0000 RESTRAINT = 0.0000
> EKCMT = 222.7207 VIRIAL = 309.7908 VOLUME = 8098.7054
> Density = 0.9457
> Ewald error estimate: 0.6705E-04

The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to
Received on Wed Sep 17 2003 - 15:53:00 PDT
Custom Search