[gmx-users] Tabulated soft-core potential, Ewald and free energy
feenstra at chem.vu.nl
Thu Apr 10 09:00:07 CEST 2003
Filip Ryjácek wrote:
> Hi all,
> I am using soft-core approach for free energy calculation in GORMACS
> (works wery well, comparing to AMBER/gibbs for example!), and I have
> encountered two problems:
> a) As show in output and in do_fnbf() and do_14(), GROMACS has only
> tabulated routines for free-energy coulomb and LJ interactions (inl3301
> and inl3302). In manual v3.1.1 on page number 122 is written: "Note that
> table lookup is significantly slower than computation of the most simple
> LJ and C interaction". So I guess GROMACS is slower for free energy
> calc. than for mol. dynamics! Is there any technical/theoretical reason
> for that?
> b) I don't understand exactly the 'FEP_TABLE_LENGTH' variable. Why is it
> set to 5nm? Is it something like cutoff?
This is Davids & Eriks hometurf, but since they have apparently not yet
answered, I'll give it a shot.
The 'most simple LJ and C' interaction, are implemented in assembly
(on Pentiums) or come from optimized C or Fortran libraries (on SGI,
IBM etc.). Since implementing the free energy code in assembly is not
feasible (too much work is an understatement). Reason is this soft-
core potential, which is much more complicated than 1/r^6 or so.
So then, the table lookup is faster than the calculating this soft-core
on the inner loops. No way around, as long as you want soft-core.
Don't know about point 'b)'.
| | |
| _ _ ___,| K. Anton Feenstra |
| / \ / \'| | | Dept. of Pharmacochem. - Vrije Universiteit Amsterdam |
|( | )| | | De Boelelaan 1083 - 1081 HV Amsterdam - Netherlands |
| \_/ \_/ | | | Tel: +31 20 44 47608 - Fax: +31 20 44 47610 |
| | Feenstra at chem.vu.nl - www.chem.vu.nl/~feenstra/ |
| | "If You See Me Getting High, Knock Me Down" |
| | (Red Hot Chili Peppers) |
More information about the gromacs.org_gmx-users