I used fix_langevin command with fix_nve to perform a BD as mentioned in LAMMPS documents, I put Tstart=Tstop=298.15 K (25 C) and damp = 482 fmsec. Also, I used a time step = 300.0 fmsec, and my system is a 24 bead-spring chain.
My simulation run well, but I noticed that the temperature fluctuates below the value that I put in the command, and when I calculate the ave/time/temp I also got around 150-160 K. what is/are the reason/s ?
I can increase the values in the command until I get a fluctuations around 298.15 K and also an ave/time/temp around this value, but now Can I say that the results of mu simulation are in ( ave/time/temp that I get ) or in what I used in the command ?
If you set a thermostat, you should definitely communicate the important details about that thermostat, including the target temperature.
As for the temperature “that you actually get”, it is difficult to converge to the thermodynamic ensemble for a small number of degrees of freedom, and large deviations are expected.
I used fix_langevin command with fix_nve to perform a BD as mentioned in
LAMMPS documents, I put Tstart=Tstop=298.15 K (25 C) and damp = 482 fmsec.
Also, I used a time step = 300.0 fmsec, and my system is a 24 bead-spring
chain.
My simulation run well, but I noticed that the temperature fluctuates
below the value that I put in the command, and when I calculate the
ave/time/temp I also got around 150-160 K. what is/are the reason/s ?
there is not enough information to answer this question. most likely, you
are making a mistake somewhere.
I can increase the values in the command until I get a fluctuations around
298.15 K and also an ave/time/temp around this value, but now Can I say
that the results of mu simulation are in ( ave/time/temp that I get ) or in
what I used in the command ?
what values in what commands? what exactly are you asking here?
good answers require well posed questions with sufficient detail.
In the second paragraph, I mean when I put Tstart=Tstop=298.15 K in fix_langevin command I will get a fluctuations around 150-160 K and an ave/time/temp around 150 K, but to get a fluctuations and ave/time/temp around 298.15 K I can increase Tstart and Tstop = 2298.15 K. Now Can I say that the results of my simulation are in ( ave/time/temp that I get which is around 298.15 K ) or in what I used in fix_langevin command which is 2298.15 K ?
In the second paragraph, I mean when I put Tstart=Tstop=298.15 K in
fix_langevin command I will get a fluctuations around 150-160 K and an
ave/time/temp around 150 K, but to get a fluctuations and ave/time/temp
around 298.15 K I can increase Tstart and Tstop = 2*298.15 K. Now Can I
say that the results of my simulation are in ( ave/time/temp that I get
which is around 298.15 K ) or in what I used in fix_langevin command which
is 2*298.15 K ?
i think you can safely say, that your simulation results are crap.
try running without thermostat and check whether you can run a stable
simulation that conserves (total) energy.
if not, reduce your time step until you can. then turn the thermostat back
on and see what happens.
I did what you said and I got a constant total energy then I turned the thermostat back again but the total energy fluctuates by small values and I can decrees the fluctuations by decreasing the time step more and more but it will be too small. when I use dt=1.0 fmsec the fluctuations of the total energy are small, Can I accept the results now ?
Also, when I used small dt and Tstart=Tstop=298.15 K in fix_langevin, the temperature fluctuates around 298.15 K which I need.
I did what you said and I got a constant total energy then I turned the
thermostat back again but the total energy fluctuates by small values and I
can decrees the fluctuations by decreasing the time step more and more but
it will be too small. when I use dt=1.0 fmsec the fluctuations of the total
energy are small, Can I accept the results now ?
a time step of 1fs seems very small for particles as heavy as the ones
you have. 1fs is typical with particles of about 15-20 a.u. mass.
since typically the maximal time step scales O(sqrt(M)), you should be able
to run with at least 15fs timestep, probably even 25fs.
when you turn on the langevin thermostat, you'll have some additional
damping of fast motions, which might help to get a stable and accurate time
integration at a larger time step.
the big unknown here is, that your pair coefficients and force constants
are different than those for typical atomic systems, so that may also have
an impact.
but then again, you have such a tiny system, it doesn't really matter that
much to use a small timestep.
I will use 25 fs for time step, but there will be a fluctuations in the total energy, Can I neglect it ?
I have put my pair coefficients and force constants by myself, and I validate it by calculating the persistance length of it ( because this bead-spring chain represents a dsDNA chain and each bead represents a 6 basepairs of the dsDNA chain), so I change these coefficients and constants until I get the best persistance length at specific salt concentration.
I used 25 fs for time step, but there are fluctuations in the total energy, Can I neglect it ?
I want to dump the total energy of the system with time, I checked the compute command but I do not find a total energy compute command, Is there a way to do that ?
I need the structure of dumped file to be
1. I used 25 fs for time step, but there are fluctuations in the total
energy, Can I neglect it ?
this is not a question for this mailing list, but for your adviser (and/or
a text book on MD).
keep in mind, that you have a *tiny* system, so your ensemble average
barely deserves that name.
2. I want to dump the total energy of the system with time, I checked the
compute command but I do not find a total energy compute command, Is there
a way to do that ?
I need the structure of dumped file to be
time total energy
you can access the total energy through its thermo keyword. how to make
use of it (and how to output those), is explained in the manual. please
help yourself to another read.
I have another question about msd compute command:
you said in LAMMPS documents “A vector of four quantities is calculated by this compute. The first 3 elements of the vector are the squared dx,dy,dz displacements, summed and averaged over atoms in the group. The 4th element is the total squared displacement, i.e. (dxdx + dydy + dz*dz), summed and averaged over atoms in the group.”
Is this the same as calculating the msd for the center of mass of the group of atoms- that used in this command- with time ? I mean this command do the following
<(Rcom(t0+t)-Rcom(t0))^2>
and for the time t, Can it be adjusted by a dump command ?
I have another question about msd compute command:
you said in LAMMPS documents "A vector of four quantities is calculated by
this compute. The first 3 elements of the vector are the squared dx,dy,dz
displacements, summed and averaged over atoms in the group. The 4th element
is the total squared displacement, i.e. (dx*dx + dy*dy + dz*dz), summed and
averaged over atoms in the group."
Is this the same as calculating the msd for the center of mass of the
group of atoms- that used in this command- with time ? I mean this command
do the following
<(Rcom(t0+t)-Rcom(t0))^2>
this is something that you can easily answer yourself by doing the math
and see if those two properties are indeed identical or not.
and for the time t, Can it be adjusted by a dump command ?
what has computing the msd to do with the dump command??
please understand, that this mailing list is no substitute for you
thinking and solving problems yourself. after all, it is *your* research
and *your* work. we're happy to help with technical issues about LAMMPS,
but when it comes to basic scientific tasks, you should do your own work.
and your questions indicate, that you are not thinking enough.