[gmx-users] Time step problem on coarse-grained LJ chain

[Tomy van Batis]

Dear all

I am trying to simulate a Lennard-Jones coarse-grained polymer melt. Until
know I simulated only one chain, in order to check if the parameters I used
are OK.

I simulated one chain of 50 LJ particles with *ε=1KJ/mol, m=1gr/mol, σ=1nm*.
This means that my characteristic time of the system *τ=σ(m/ε)1/2* is *τ=1ps.
*

I am using RIGID bonds between the particles also with length *σ=1nm*.

I have seen in many papers which are using similar systems that the time
step they use is Dt=0.01τ.

When I use *Dt=0.01τ* I don't get any error messages in my *.log* file or
in the resulting *.gro* file.

The problem that I have is when I am trying to convert my *.xtc* file to *
.pdb* in order to visualize it in VMD.

When I use:

*trjconv -f  ***.xtc   -s  ***.tpr      -o  ***.pdb     -pbc whole*

the trajectory looks fine.

BUT, when I use

*trjconv    -f  ***.xtc    -s ***.tpr   -o ***.pdb     -pbc nojump*

, it seems the chain is blowing after few steps (the bonds break and the
particles spread in every direction).

I would think that for a single chain, these two options of trjconv should
not make such a difference.

I used a relatively big box for the simulation (10 X 10 X 60). The
*strange*think is that I don't have this problem when I increase the
box size (to
50X50X60) , or of course when I decrease the time step.

Is it possible that I am not able to use such a large time-step in GROMACS,
or is it something with the *trjconv* that I don't understand ?

Thanks in advance

Best Regards, Chrysostomos
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20111124/2f753790/attachment.html>