using PLUMED virtual atoms with MPI [PLUMED-2.1.2 + CP2K]

[Jaap Kroes]

Dear CP2K developers,


This is crossposted from the PLUMED mailing list since they think the issue 
may be in the interface rather than PLUMED itself. 
I'm trying to do a steered MD simulation with PLUMED(2.1.2) + CP2K (latest 
version, attached are some minimal inputs).

However the simulation crashes for virtual atoms (COM) constructed from a 
group of atoms of a certain size. 
The number of atoms should not have any significant effect on the resulting 
CV (at least not initially). 
It is difficult to identify which combations work and which don't.

(in a system of 63 atoms in total..)

surface: COM ATOMS=1                                         [works]
surface: COM ATOMS=1,2,3,4,5,6,7,8,9,10,11,12,13,14          [crashes]
surface: COM ATOMS=1,2,3,4,5,6,7,8,9,10,11,12,13,14,15       [works]
surface: COM ATOMS=1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16    [crashes]
surface: COM ATOMS=1-60                                      [crashes]

d_gr-pr: DISTANCE ATOMS=surface,61 COMPONENTS
restraint: ... 
  MOVINGRESTRAINT 
   ARG=d_gr-pr.z KAPPA0=0.1
    AT0=+3  STEP0=0000
    AT1=-3  STEP1=2000  
...


This only happens for parallel runs (pure MPI, pure openmp, or a mix of the 
two).
So I guess it is related to some MPI communication problem.

Did you ever see something similar or have any idea what could go wrong?

Thanks for any suggestions.


Best,
Jaap
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20150617/87e5cbcc/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: md.inp
Type: chemical/x-gamess-input
Size: 4365 bytes
Desc: not available
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20150617/87e5cbcc/attachment.inp>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: plumed.dat
Type: application/octet-stream
Size: 798 bytes
Desc: not available
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20150617/87e5cbcc/attachment.obj>