Re: AMBER: Query on restart file

From: Robert Duke <>
Date: Thu, 30 Oct 2008 22:14:40 -0400

Check the restart for overflow of the file format (search for a short string of '****' where you would expect a number. You are probably running without iwrap = 1, right? Well, after a while, the primary images will go into space with a coordinate greater than the file format can handle (which is 9999. to -999. if memory serves - it is the negative overflow that kills you). When this happens, you must go back to your last good restrt and run at least 1 step with iwrap=1 set in &cntrl. This will wrap the atom primary images back into the primary unit cell (coordinate values closest to 0.). This is not done routinely because it interferes with things like diffusion measurements, but most folks don't want to do that anyway. There is probably lots more info on this problem in the reflector archives. Some folks just leave iwrap on all the time. I typically just check every 20 nsec or so and wrap if things are getting close to overflowing (overflowing in 17 nsec seems fast, but your system may have some hotter molecules than the ones I routinely deal with).
Regards - Bob Duke
  ----- Original Message -----
  From: Neha Gandhi
  Sent: Thursday, October 30, 2008 9:49 PM
  Subject: AMBER: Query on restart file

  Hi Amber community,

  I ran 17 ns of production runs (each trajectory file consist of 500 ps with 50 frames). I tried to run another 500 ps after 17ns but it just says cannot read coordinates from the previous .rst file. When I checked , there is no spelling error and I checked the previous rst files. To verify I again run the 16th ns md again and no issues with it. Only thing I cannot further run MD after 17th ns? What could be the problem and the possible solution? I m following the mm-pbsa tutorial in amber.

  Many thanks,

  Neha Gandhi,
  School of Biomedical Sciences,
  Curtin University of Technology,
  GPO Box U1987 Perth,
  Western Australia 6845

The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" (in the *body* of the email)
Received on Fri Oct 31 2008 - 05:12:30 PDT
Custom Search