MC probability bug in PMTRAION flag

Cho S chot... at
Wed Dec 13 03:27:51 UTC 2017


I have encountered a bug (feature?) within CP2K MC (Monte Carlo) code 
version 5. The bug showed up when I tried to run GEMC_NVT (two boxes) in 
order to get a vapor-liquid coexistence curve.

To demonstrate the bug, I have prepared two identical boxes containing 1 
water molecule each. Any MC move performed on water molecules should 50% 
end up in box 1, and 50% in box2.

The bug is related to ‘PMTRAION’ and ‘PMTRAION_MOL’ flags (probability of 
attempting a conformational change)


I observed that when I performed 5000 PMTRAION_MOL moves, about 75% of the 
moves ended up in box1, and 25% ended up in box 2. This did not happen with 
other moves like PMSWAP (swapping between 2 boxes) or PMTRANS 
(translational move) which gave ~2500 moves for each box. I don’t think 
this is a problem with my compilation since the precompiled executable also 
shows the same problem. (see attached)

Could anyone clarify if this behavior is intended?


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: conformational.tar.gz
Type: application/octet-stream
Size: 2993078 bytes
Desc: not available
URL: <>

More information about the CP2K-user mailing list