[gmx-users] GROMACS problem

[Erik Lindahl]

David van der Spoel wrote:

>On Thu, 25 Jul 2002, alexander yakovenko wrote:
>
>>Dear all,
>>I work with nonproteion molecules using gromacs, and have some problems.
>>Does anybody know is the gromacs MD simulation crash if the number of
>>charge groups in rtp file more than 32 ? I have about ten examples of
>>such crashing, but it may be the problem of my molecules type. This crush
>>took place in MD simulation only, but not in energy minimization.
>>
>Charge groups can not be larger than 32 atoms (we use an integer as
>bitmask somewhere). If you use PME you can certainly make your charge
>groups smaller, it wil lnot affect the coulomb interactions.
>
Just to clarify; there is a very good reason why you shouldn't ever 
consider charge groups of size>32
(otherwise we would of course fix that limitation). Charge groups are 
used in the neighborsearching too
speed things up, so if you have two huge chargegroups some of the atoms 
may be very close even if the
centers-of-mass of the charge groups are not. If the centers are further 
apart than the cut-off, you will
still not get any interaction at all between the atoms close to each 
other!.

Since a cut-off is typically in the order of 1nm, even the best case of 
perfectly spherical charge groups
could lead to severe errors if you have as much as 32 atoms in them.

Cheers,

Erik

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20020725/d5907fc9/attachment.html>