[gmx-users] plateau in msd (glass transition); ref_t

[Anja Kuhnhold]

Hi everyone,

I'm simulating a bead-spring polymer model (1600 chains and 10 beads per chain in a 26.6^3 box with pbc) with LJ and FENE potentials.
I calculate the mean-square-displacement for different temperatures. For T=0.46 (in LJ units) I expected to get a plateau in the msd curve (glass transition)- but there is none. The curve for T=0.46 is similar to the one for T=1.0 (above glass transition) with only a small shift to lower values- but no plateau.

My .mdp-files look as follows:

integrator         = md-vv
dt                 = 0.0035
nsteps         = 100
nstxout         = 1
nstvout         = 1
nstfout         = 1
nstlog         = 1
ns_type         = grid
pbc                 = xyz
periodic_molecules = yes
rvdw                 = 1.12
rlist                 = 1.3
tcoupl         = nose-hoover
tc-grps         = System
tau_t                 = 20
ref_t                 = 55.32
vdwtype         = Shift
rcoulomb         = 1.12
coulombtype = Reaction-Field-zero
epsilon_rf         = 0

I have 6 of these files, where only nsteps and nst*out are changed by multiplying them by 10 (so I get overlapping intervals for the msd).


Is the ref_t correct? The results in the md.log file say T is about 89.
Why do I not get a plateau? What did I not consider?

Thanks in advance.
Anja