How to perform a CI-NEB calculation with some atoms fixed in position?

[Lianheng Tong]

Hi all,

I am posting this question on behalf of a colleague.

He is performing a CI-NEB calculation on copper substrates. The substrate 
is quite thin within the simulation cell, and has to be fixed in position 
during the calculation, otherwise the forces on the substrate atoms will be 
too large.

The CI-NEB calculation has 7 replicas, in which some atoms are fixed using 
CONSTRAINTS subsection in MOTION section. No initial velocities were set. 
 My colleague noticed the fixed atoms moved in replicas 2 to 7 at the first 
step, while the fixed atoms in replica 1 did not move (as intended). In the 
next steps, the atoms were fixed in all replicas.

It seems that this behaviour did not happen with CP2K 2.3 but is observed 
in versions 2.4 and 2.5.

Furthermore, running the calculation on a single processor gave the same 
behaviour as running on multiple processors, so this may not be a 
parallelisation issue.

I had a look at the source code, and it seems to me that the constraint 
information isn't even used for the default POT_OPT == FULL setting, and 
only COLVARs are used for POT_OPT == FE or ME.

So, the question is:

Can we perform a standard CI-NEB calculation with some atoms fixed in each 
replica?  And if so, what is the correct way of doing such calculations 
with CP2K?

We want to make sure we are doing things correctly before reporting this as 
a possible bug.

I have attached the restart file, which contains the CP2K input.

Many thanks and best regards,

Lianheng

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20140404/16c6fa7b/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: WalkeronCu.restart
Type: application/octet-stream
Size: 291486 bytes
Desc: not available
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20140404/16c6fa7b/attachment.obj>