[lammps-users] how to control temperature during Molecular Dynamics??

hello all:

im using lammps for my first time and im trying to run a molecular dynamics to a peptide
using as a template the most of in.peptide script available in "lammps/examples/peptide"
my doubt is "how can i control the temperature of the MDynamics"

I have set a group with all my atoms with "group peptide type all"

but during the run the temperature keeps increasing untill the run ends..

my final objective is to run a Targeted Molecular Dynamics .. but first i need to understand and the basics of
LAMMPS
thanks to all in advance

tarajano

my input script

Tarajano,

you already have the fix nvt in your script that would control the temperature of the simulation.

Temperature control means that T will fluctuate around your target value (275K in the script below).
The dumping constant of the thermostat (100 fs in your script) is shorter than your total simulation
time, 50 fs, that’s why you do not see the temperature as a constant or oscillating around your
target value of 275K.

As a rule of thumb, the thermostat dumping constant has to be 50 to 100 times the timestep. The
dumping constant of 100 fs is most reasonable, but the timestep you are using is unnecessarily low
(“real” units of time are fs). I suggest you to use 1 fs step, and always look at average temperatures
in runs that are at least several times longer than the thermostat dumping constant.

To read more about simulations in the NVT ensamble in http://lammps.sandia.gov/doc/fix_nvt.html
(if you want to do constant pressure, look for http://lammps.sandia.gov/doc/fix_npt.html).

I hope this helps,
vale

Valeria Molinero wrote:

Tarajano,

you already have the fix nvt in your script that would control the temperature of the simulation.

Temperature control means that T will fluctuate around your target value (275K in the script below).
The dumping constant of the thermostat (100 fs in your script) is shorter than your total simulation
time, 50 fs, that's why you do not see the temperature as a constant or oscillating around your
target value of 275K.

As a rule of thumb, the thermostat dumping constant has to be 50 to 100 times the timestep. The dumping constant of 100 fs is most reasonable, but the timestep you are using is unnecessarily low ("real" units of time are fs). I suggest you to use 1 fs step, and always look at average temperatures in runs that are at least several times longer than the thermostat dumping constant.

To read more about simulations in the NVT ensamble in LAMMPS Molecular Dynamics Simulator
(if you want to do constant pressure, look for LAMMPS Molecular Dynamics Simulator).

I hope this helps,
vale

hello all:

im using lammps for my first time and im trying to run a molecular dynamics to a peptide
using as a template the most of in.peptide script available in "lammps/examples/peptide"
my doubt is "how can i control the temperature of the MDynamics"

I have set a group with all my atoms with "group peptide type all"

but during the run the temperature keeps increasing untill the run ends..

my final objective is to run a Targeted Molecular Dynamics .. but first i need to understand and the basics of
LAMMPS
thanks to all in advance

tarajano

my input script

#
units real
neigh_modify delay 2 every 1

atom_style full
bond_style harmonic
angle_style charmm
dihedral_style charmm
improper_style harmonic

pair_style lj/charmm/coul/long 8 10 10
pair_modify mix arithmetic
kspace_style pppm 1e-4

read_data 1omb_ia3.data

timestep 0.005

thermo_style multi
thermo 50

special_bonds charmm

fix 1 all nvt 275.0 275.0 100.0
fix 2 all shake 0.0001 10 100 m 1 2 3 4 5 6

group peptide type all

restart 1000 tmd.restart
dump 1 all atom 10 tmd.dump.trj

run 10000

-------------------------------------------------------------------------
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it now.
http://sourceforge.net/powerbar/db2/
_______________________________________________
lammps-users mailing list
[email protected] <mailto:[email protected]>
https://lists.sourceforge.net/lists/listinfo/lammps-users

thanks a lot Valeria for your reply..

In fact i need to simulate natural conditions T=298 P=1atm and I have followed your suggestion of using npt and i have modified the script as follows :

fix 2 all shake 0.0001 100 100 m 1 2 3 4 5 6
fix 1 all npt 298.0 298.0 50.0 xyz 1.0 1.0 50.0

but i have the feeling that the Pressure is not being controlled in a confident way,
the output shows pressure jumps in a range of 200-300 units around the setted 1 atm value,
in fact ..the output values are of the order of hundreds and thats odd to me since *units* documentation * *for "*real*" specifies that pressure are atmospheres .
anyway it is obvious that theres something missing in my script...
But i have searched in examples scripts and in the documentation of *npt *and i have found nothing suggesting what to change or add to my script .

i will preciate any help or comment
thanks in advance

tarajano

In fact i need to simulate natural conditions T=298 P=1atm and I have
followed your suggestion of using npt and i have modified the script as
follows :

fix 2 all shake 0.0001 100 100 m 1 2 3 4 5 6
fix 1 all npt 298.0 298.0 50.0 xyz 1.0 1.0 50.0

but i have the feeling that the Pressure is not being controlled in a
confident way,
the output shows pressure jumps in a range of 200-300 units around the
setted 1 atm value,

instantaneous pressure fluctuation of condensed liquids _are_ very high,
since you are looking at something not very compressible, so small
changes in the volume have large impact on the pressure and for small
volumes the fluctuations due to the movement of the molecules are
very large, too.

cheers,
  axel.

Tarajano,

Big pressure fluctuations are natural to npt dynamics of small systems… Although it sounds paradoxical,
the pressure fluctuations with a barostat may be higher than for constant volume simulations (Brown and
Clarke, Molecular Physics 51, 1243-1252, 1984)

In your input below the dumping constant for the thermostat and barostat are the same: that may lead to
excessive fluctuations due to coupling of the two. To avoid that, change your pressure dumping constant
to at least 5 times higher than the thermostat one. I do not know if that will solve the issue of the big
fluctuations, that are natural in small simulation cells.

vale

MD simulations are full of surprises. Sometimes the results appear to be wrong, but the simulation is really correct. Sometimes the results appear to be correct, but the simulation is really wrong. Before starting something new, it is always a good idea to first try to reproduce a well-defined result that is related to your target calculation. In your case, a good result to reproduce would be NPT dynamics for a 3D periodic cell containing water molecules. There are lots of good papers on this. See:

http://lammps.sandia.gov/doc/Section_howto.html

for information on TIP3P and SPC water models. The peptide example can serve as a starting geometry for H2O.

Aidan