CHARMM Development Project
Posted By: gianluca PME and cutoffs - 08/25/13 05:12 AM
I wonder what are recommended cutoff parameters for non-bonded interactions (in particular VDW) when using PME.

In nbonds.doc I read:

For no-cutoff periodic systems:

NBONDS ATOM EWALD PMEWALD KAPPA 0.32 -
FFTX 32 FFTY 32 FFTZ 32 ORDER 6 -
CUTNB 12.0 CTOFNB 10.0 VDW VSHIFT

So, here the recommendation is 9/10/12 for CTONNB/CTOFNB/CUTNB

However, in file c26test/pme_p21.inp I find:

dyna verlet ... -
cutnb 12.0 ctofnb 10. cdie eps 1. -
ctonnb 9.99 vshift cutim 14.0 imgfrq 25 wmin .5 -
ewald pmewald

So, here they have 9.99/10/12 which corresponds to an almost abrupt truncation of VDW.

I would appreciate any advice on this. Thanks!
Posted By: rmv Re: PME and cutoffs - 08/25/13 02:13 PM
Do not use the test cases as examples of good usage.

With VSHIFT, the value of CTONNB is irrelevant, as it is not used; only switching functions use this value.

For the VDW terms, we use and recommend

VATOM VFSWITCH CUTNB 14. CTOFNB 12. CTONNB 10.

which is the default in most "all36" parameter sets.
© CHARMM forums