[gmx-users] problem with pulling (afm_rate is not equal to real pull-velocity)

[Thomas Schlesier]

Dear all,
i have with Gromacs 3.3.3 problem with the afm-pulling (the velocity are
not correct)
My system consists of ureas with short alkane linkers (see picture below)

+___+___+
|        |        |
x       x       x
x       x       x
|        |        |
____*____

x are the ureas (which form hbonds), | are the linkers and _ are
constraints and distance restraince (so that the molecules have a nearly
constant distance). * is the REF-group and + are the PULL-groups.
The initial position of the three spring was at the position of the
three PULL-groups. For the pull direction i took the vector between the
REF-group and the middle PULL-group (and normalized the vector). The
pull-direction for the other two PULL-groups was the same (so the
pulling of the three molecules should be parallel).
For all springs i took the same velocity and spring constant. From the
*.pdo i calculeted the distance which the springs moved, and from a plot
distance against time i can calculate the velocity.

And here comes the problem: the velocity should be 10E-4 nm/ps
For the middle spring it was: 9.995E-5 nm/ps (here i think is a small
rounding error, due to the normalisation)
But for one of the outer springs it was: 7.6939E-5 nm/ps ; additionally
the velocity wasn't realy constant, the fit with an polynom (2end order)
was much better: 6.9173E-5 (for the linear term) and 6.4715E-10 (for the
quadratic term).

In a simulation with 15 parallel molecules the decrease of the
pulling-velocity was even more extrem. The velocity of the outter most
springs was more then 5 times small then the velocity i have writen in
the *.ppa file.

Has anybody an idea from what this problem arises?
Thanks for answers.
Thomas