FIXED_BOND vs RIGID_BOND

[omarove]

Hi all,

While defining a molecule, what is the difference in setting it as RIGID or FIXED? I have tried both (for CF3CH2F) and the simulation showed zero absolute adsoprtion for the RIGID_BOND, while the FIXED_BOND yielded some adsorption. Also, when setting the bond as HARMONIC_BOND, it runs the initialization cycles (1000) and equilibration cycles (5000), but it stops at the ~3600th cycle (out of 5000). The error message shows "no energy conservation in internal MC scheme". The number of cycles is definitely low (it is just to experiment the different types of bonds). Can anyone explain?

Thanks,
Omar

[David Dubbeldam]

See the examples for molecules. RIGID_BOND is used for molecules that are fully rigid, for example water or CO2. The bonding information is used for the CBMC-biasing.
RIGID_BOND is a type of bond-potential that has only one bond-distance.

[omarove]

Thank you for your reply, Prof. Dubbeldam. Yes, the description for the RIGID_BOND is there and I can see how it is used in the examples. Yet, for the FIXED_BOND, which also has one input, it is not clear what is means and when it is used. I could only use it with flexible molecules. At least, I would want to know the concept of FIXED_BOND, does it imply that the bond length is fixed? The only description available in the manual is that it is used for bonds constraint using the ‘SHAKE’ and ‘RATTLE’-algorithm. 

[David Dubbeldam]

FIXED_BOND is for fixing a single bond, while the full molecule can still be flexible via bend and torsion potentials. An example is the TraPPE model for united-atom alkanes. In Monte Carlo it is easy to fix bonds because the bonds are _generated_. In Molecular Dynamics algorithms like shake needs to be used.
I would say that is better not to use fixed bonds, but rather use a harmonic bond-potential.