Why does lagrangian metadynamics puts hills at the prior location in the *HILLS.metadynLog?

[John]

CP2K wizards,

I noticed that the hills listed in *HILLS.metadynLog appear to lag by one 
timestep behind the history in *COLVAR.metadynLog file when Lagrangian 
metadynamics is used.

Check for example the regression test "hcn_meta_chaincoord.inp" in the 
QS\regtest-gps-2 test folder.

The metadynamics placed a hill at t=5.0 femtoseconds when  ss0=5.6857498 
according to "hcn_chaincoord_c-COLVAR.metadynLog" but 
"hcn_chaincoord_c-HILLS.metadynLog" lists the hill at ss0=0.57182 which is 
the value of the collective variable for the previous step.  Especially 
when these values are the same in both *COLVAR.metadynLog and  
*HILLS.metadynLog when lagrangian dynamics are not used as in the  
H2O-meta_coord test in the same folder.

I do not believe this is correct.  The value ss0 should be the same for the 
hcn_chaincoord_c-HILLS.metadynLog and hcn_chaincoord_c-COLVAR.metadynLog.  
Is there something I don't understand?

John J. Low
Argonne National Laboratory


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20140303/8eeb3a1e/attachment.htm>