[gmx-users] umbrella sampling, deshuffling and version 3.3

[Enrico Piccinini]

Hi all!
I am using the recently released version 3.3 for my umbrella sampling runs.
I have noticed something strange in my .pdo files, which seems 
conflicting with trajectory files.

This is my umbrella sampling configuration ppa file:
verbose            = yes
runtype            = umbrella
group_1            = K2
reference_group    = REFaC
reftype            = com
pulldim            = N N Y
k1                 = 8368.0 ;kJ mol^-1 nm^-2
pos1               = 0.006  -0.12792  2.06367

and pos1 values have been calculated trough g_traj

My .pdo files reads as follows:
# UMBRELLA      3.0
# Component selection: 0 0 1
# nSkip 1
# Ref. Group 'REFaC'
# Nr. of pull groups 1
# Group 1 'K2'  Umb. Pos. 2.063670  Umb. Cons. 8368.000000
#####
0.000000        -0.065040
0.001000        -0.064709
0.002000        -0.064338

First value is known not to be zero because this is the second step 
after the equilibration run. However value -0.065 means that the pulled 
group is 1.998 nm above my reference group. If I check the trajectory 
for the same time with g_traj i get 3.08 between pulled and reference 
groups. All  the other times show similar behaviour.
My results after g_wham do not seen reasonable.
I am wondering if something may got wrong due to the shuffling procedure 
or something else is involved (here including my faults, eventually).
Thanks
Enrico