[gmx-users] g_velacc. Reading frame 165000 time 825.000 Killed

[David van der Spoel]

Vitaly Chaban wrote:
> Hi,
> 
> I calculate the diffusion constants via Green-Kubo relation for a
> system of 216 molecules (5 sites in each). The velocities are saved
> every 0.005 ps and the total trajectory is 1000 ps long. Then applying
> g_velacc -nonormalize -mol -n index.ndx -acflen 1001
> The process starts and after some time is killed:
> 
> trn version: GMX_trn_file (single precision)
> Reading frame  165000 time  825.000   Killed
> 
> 
> If one gives a less part of the trajectory:
> 'g_velacc -nonormalize -mol -n index.ndx -acflen 1001 -b 300 -e 600'
> the calculation is OK.
> 
> Low memory or what? It doesn't look so according to the task manager.
> Has anybody experienced the same problem? Any tricks to overcome it?
> Found on version 3.3.1.
> 
> Thanks.
> 
This is typically low memory. The program reads in principle the whole 
trajectory in memory. You could do the analysis on half of the molecules 
at  a time and average in xmgrace.

What is a task manager?  On unix systems you could check the "limit" 
command.

-- 
David van der Spoel, Ph.D., Professor of Biology
Molec. Biophys. group, Dept. of Cell & Molec. Biol., Uppsala University.
Box 596, 75124 Uppsala, Sweden. Phone:	+46184714205. Fax: +4618511755.
spoel at xray.bmc.uu.se	spoel at gromacs.org   http://folding.bmc.uu.se