[gmx-users] LINCS warning

[Daniela S. Müller]

hi raghu,

in version 3.3 some bugs have been fixed, which you can 
read on the gromacs website.
E.g. the temperature coupling algorithm has been modified, 
which means that this thermodynamical property is 
evaluated and assigned differently than in v3.2. This will 
change your simulations totally, you certainly won't be 
getting the same trajectories.
In any case it's difficult to compare different runs if at 
the same time you are comparing different softwares.

daniela


On Sat, 9 Sep 2006 17:15:55 +1000 (EST)
  "Mark Abraham" <mark.abraham at anu.edu.au> wrote:
>> Dear all,
>> I have a query,
>> 1. Are two trajectories comparable if each one of them
>> has been run in a different version of gromacs? Eg,
>> v3.2 and v3.3.
> 
> Define "comparable". If you mean are they sampling the 
>same thermodynamic
> ensemble, then the answer is yes if you used the same 
>algorithms. If you
> mean are they numerically identical, the answer is no. 
>Round-off errors
> from different order of summation, etc. will cause tiny 
>differences that
> magnify chaotically.
> 
> Mark

--

Daniela S. Müller
Molecular Dynamics (MD)
Groningen Biomolecular Sciences and Biotechnology 
Institute (GBB)

MD group website: http://www.rug.nl/gbb/md

Address:
Daniela S. Mueller
Molecular Dynamics group
Dept. of Biophysical Chemistry
University of Groningen
Nijenborgh 4
9747 AG Groningen
The Netherlands

email: D.S.Mueller at rug.nl