[gmx-users] Reg: Simulating large number of molecules

[Mark Abraham]

On 24/12/2010 9:06 PM, vinothkumar mohanakrishnan wrote:
 > Hi all
 >
 > I want to simulate more than 1000 molecules of say octane r any
 > alkanes in gromacs. after doing a little math i find my box dimensions
 > for the molecules. but when i use genbox to insert multiple molecules
 > in my local system after running some time it says cannot allocate
 > memory but when the same command i gave it in a Linux cluster
 > interactively (HPCE) it goes well and fills the box with required
 > number of molecules and i get the correct density.

OK, so I think you've demonstrated that your Linux cluster nodes have
more memory available.

 > but even then there are limitations such as i can't go beyond 1500
 > molecules with required box dimensions.

Read genbox -h. Molecules get placed randomly at interstices. Close
packings are intrinsically non-random, so you won't get them from random
packing.

 > So how one can about simulating more than 1000 molecules of alkanes r
 > any molecules for that matter? I know that this problem doesn.t occur
 > for water. any help is highly appreciated.

If you want more molecules, use a bigger box. If you want more molecules
at high density, use a bigger box and very careful, very gentle pressure
coupling to compress it.

Mark