[gmx-users] Results from different OS

[K.A. Feenstra]

Oliver Beckstein wrote:
> 
> I'm not claiming any in-depth knowledge about how gromacs runs on
> different machines but by just looking at the numbers that you appended
> I'd say it's consistent: The difference in the averages is well
> accounted for by the fluctuations (the section RMS-FLUCTUATIONS); at least
> this seems to be true for the energies (everything else from then on
> arrived as binary gibberish).
> 
> Perhaps an interesting question would be how large are these fluctuations
> supposed to be?

Generally speakinjg, fluctuations in a 300K MD simulation are large.
The fluctuations from the logfile seem reasonable enough, and indeed
the differences in averages are minimal compared to the fluctuations.

When comparing across hardware and/or OS, keep in mind that and MD
simulations is a numerical integration. This means everything is
basically the result of an incredible number of additions (*), and any
rounding that will occurr, will have some influence in the result.
Add to that the innate chaotic behaviour of a newtonian multi-body
system (this is already true for 3 (yes, three!) bodies) behaves
chaotically. That means that any (small) difference in rounding can 
have large effects on the results. Nevertheless, averages should
converge as they seem to do...

(*) to get an idea: running 10000 particles for 10 ps at 2 fs dt,
one would have to add rougly 10M interactions (10k*1k) per step,
for 20k steps (10ps/2fs), makes about 200,000,000,000 (200G) steps!

-- 
Groetjes,

Anton
 ________ ___________________________________________________________
|        | Anton Feenstra                                            |
| .      | Dept. of Pharmacochemistry - Vrije Universiteit Amsterdam |
| |----  | De Boelelaan 1083 - 1081 HV Amsterdam - The Netherlands   |
| |----  | Tel: +31 20 44 47608 - Fax: +31 20 44 47610               |
| ' __   | Feenstra at chem.vu.nl - http://www.chem.vu.nl/afdelingen/FAR|
|  /  \  |-----------------------------------------------------------|
| (    ) | Dept. of Biophysical Chemistry - University of Groningen  |
|  \__/  | Nijenborgh 4 - 9747 AG Groningen - The Netherlands        |
|   __   | Tel +31 50 363 4327 - Fax +31 50 363 4800                 |
|  /  \  | K.A.Feenstra at chem.rug.nl - http://md.chem.rug.nl/~anton   |
| (    ) |-----------------------------------------------------------|
|  \__/  | "If You See Me Getting High, Knock Me Down"               |
|        | (Red Hot Chili Peppers)                                   |
|________|___________________________________________________________|