[gmx-users] anisotropic pressure coupling

[A. und B. Nick]

Hi,

I want to run a variable shape MD simulation on a box of 
3120 CH2 groups belonging to 20 chains (box size=4,5nm).
The structure was preequilibrated with MC and I run about
1,5 ns MD with isotropic boundary conditions for further
equilibration.

After switching to anistropic pressure coupling I always 
run into problems and the job crashes after a while.
The error messages are: "too skrewed box" when I use
LINCS for bond contraints or I recieve a segmentation
fault if I run the job without any contraints.

Does anyone know how to overcome this problem?

I tried many combinations of temperature and pressure
coupling and I have included a typical input-file below
(the small tau_p should guarantee small temperature
variations, the ref_p parameter accounts for the
long range correction of LJ interaction).
In all simulations I always obtain a <strange?> behaviour
of the xy, xz and yz box parameter that continously decrease
up to about -2 before the job crashes. I have attached 
the ener.edr file belonging to the following parameter that
shows this behaviour.

Thanks in advance to everybody who can help,
Gitta

input-file:
title               =  Test
cpp                 =  /lib/cpp
integrator          =  md
tinit               =  0
dt                  =  0.001	; in ps
nsteps              =  1000000	; total 1 ns.
nstcomm             =  1        ; 

nstxout             =  500
nstvout             =  500
nstfout             =  0
nstlog              =  500
nstenergy           =  500
nstxtcout           =  1000

nstlist             =  10
ns_type             =  simple
pbc                 =  xyz
rlist               =  1.0

tcoupl              =  berendsen
tc-grps		  =  System
tau_t               =  0.01
ref_t               =  300

Pcoupl              =  berendsen
Pcoupltype          =  anisotropic
tau_p               =  10
compressibility     =  6e-5 6e-5 6e-5 6e-5 6e-5 6e-5 
ref_p               =  624  624  624  624  624  624   

gen_vel             =  no
gen_temp            =  300.0
gen_seed            =  173529

constraints          =  none
constraint_algorithm =  lincs
unconstrained_start  =  yes
lincs_order          =  4
lincs_warnangle      =  30


-- 
Dr. Birgitta Nick
DWI an der RWTH Aachen
Veltmanplatz 8
52062 Aachen