AMBER Archive (2009)

Subject: [AMBER] vlimit

From: Taufik Al-Sarraj (taufik.alsarraj_at_utoronto.ca)
Date: Wed Oct 28 2009 - 16:16:32 CDT

Hi,
a molecular dynamics simulation just crashed giving me an error
vlimit exceeded for step 4; vmax = 41.1136
vlimit exceeded for step 5; vmax = 154.8294

based on previous posts this can either be due to an atom with mass 0
(zero) or two overlapping atoms
my system is a little large (144000 atoms) so i cannot attach the prmtop
or inpcrd files

i ran two minimizations and they ran to completion without errors

min1

System minimization -everything fixed except water and Na+

&cntrl
imin = 1,
maxcyc = 1000,
ncyc = 500,
ntb = 1,
ntr = 1,
iwrap = 1,
cut = 8
/
Hold the Surface-linker-DNA fixed
10.0
RES 1 41
END
END

Min2
System minimization -only the surface is fixed

&cntrl
imin = 1,
maxcyc = 10000,
ncyc = 5000,
ntb = 1,
ntr = 1,
restraint_wt=20.0,
restraintmask=':2',
iwrap = 1,
cut = 8
/

however when i try to run heat1 i get the vlimit error

Heating from 0 to 200K -Molecular dynamics 20ps
&cntrl
imin = 0,
irest = 0,
ntx = 1,
ntr = 1,
nstlim = 10000,
dt = 0.002,
ntc = 2,
ntf = 2,
ntt = 3,
gamma_ln = 1.0,
tempi = 0.0,
temp0 = 200.0,
ntpr = 200,
ntwx = 500,
ntwr = 1000,
ntb = 1,
restraint_wt = 10.0,
restraintmask = ':1-41',
cut = 8.0
/

i looked the *.prmtop file, i don't have any mass = 0, i do have some LJ
constants equal to zero but i hope these are for the water hydrogens

i also tried ptraj sys.prmtop script

script =
{
trajin sysdmin2.rst
checkoverlap 0.95
}

but no atoms were found

if i run the heat1 script with the following restraintmask = ':2' the
simulation runs to completion with no errors!

i have a couple of questions:
Are there other reasons that can cause a velocity error that i have
overlooked?

if i set vlimit to say 20, wouldn't this prevent the simulation from
crashing? if two atoms are too close, their velocity will be scaled
according to the vlimit value, and in few (hundred) steps the two atoms
will no longer overlap and the simulation will continue to run..

Thanks,
Taufik

_______________________________________________
AMBER mailing list
AMBER_at_ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber