IUBio GIL .. BIOSCI/Bionet News .. Biosequences .. Software .. FTP

restart files

abonvin at igc.phys.chem.ethz.ch abonvin at igc.phys.chem.ethz.ch
Wed May 28 02:41:34 EST 1997


>The xplor restart files are formatted so maybe round-off error is
>the problem.

This might also add to the problem. To increase precision, another way
to restart would be, not to use the restart file, but to write out
to binary files the positions and velocities, separately, and then 
read then in again before restarting. In that way you precision will be
higher, BUT you will loose the time informationa as well as the energies
statistics over the entire simulation.

An example:

		{* save coordinates and velocities as binary trajectory*}
		{* files to avoid instabilities when restarting        *}

  write traj reset end
  write traj ascii=false output=$savexou end
  vector do (xcomp = x) (all)
  vector do (ycomp = y) (all)
  vector do (zcomp = z) (all)
  vector do (x = vx) (all)
  vector do (y = vy) (all)
  vector do (z = vz) (all)
  close $savexou end
  write traj reset end
  write traj ascii=false output=$savevou end
  close $savevou end
  vector do (x = xcomp) (all)
  vector do (y = ycomp) (all)
  vector do (z = zcomp) (all)


And the restarting part would look like:

        if ($restart = true) then
		{* restore coordinates and velocities *}
		{* first velocities                   *}
          read traj
            asci=false
            inpu=$savevin
            begi=1
            skip=1
            stop=1
          end
          close $savevin end
          vector do (vx = x) (all)
          vector do (vy = y) (all)
          vector do (vz = z) (all)
		{* then coordinates                   *}
          read traj
            asci=false
            inpu=$savexin
            begi=1
            skip=1
            stop=1
          end
          close $savexin end

and then restart you dynamic with iasvel=current
 
Alexandre

==========================================================================
| Dr. Alexandre Bonvin           | Phone: Int+41-1-632-5504              |
| Physical Chemistry, ETHZ       | Fax:   Int+41-1-632-1039              |
| Universitatstrasse 6           | Email: abonvin at igc.phys.chem.ethz.ch  |
| 8092 Zurich, Switzerland       | http://igc.chem.ethz.ch/abonvin       | 
==========================================================================




More information about the X-plor mailing list

Send comments to us at archive@iubioarchive.bio.net