Re: [AMBER] Could not read velocities from restart file

From: David Condon <>
Date: Tue, 10 Apr 2012 14:16:40 -0400

Hi Dr. Steinbrecher,

thanks much for your prompt reply! For some bizarre reason, even on short
re-runs the error is reproduced.

The restart runs are in sequence:

-rw-r--r-- 1 dave dave 252901 2012-04-10 08:33 New/1.rst
> -rw-r--r-- 1 dave dave 252901 2012-04-10 08:33 New/2.rst
> -rw-r--r-- 1 dave dave 252901 2012-04-10 08:33 New/3.rst
> -rw-r--r-- 1 dave dave 245309 2012-04-10 12:59 New/4.rst
It looks like I will just have to wait until the 4th MD run finishes.


On Tue, Apr 10, 2012 at 10:08 AM, <> wrote:

> Hi,
> > restarted the computer to install some updates (first by killing the
> pmemd
> > process, which may not have been the proper way).
> I would agree that this is not the proper way and that one should split
> long simulations into pieces and keep intermediate rst-files (check the
> ntwr<0 option) so that precisely this does not happen, but anyways:
> > Upon restarting the simulation, I get the error at the bottom of the new
> > output file:
> >
> > | ERROR: Could not read velocities from 4.rst
> >>
> Check if your rst file contains the correct 3*N number of coordinates and
> 3*N velocities. If you were unlucky enough to kill the process right when
> it was halfway done with writing a rst file, you may have truncated and
> thereby corrupted it.
> However all is not lost I guess:
> if your rst file contains enough coordinates (3*natom, so natom/2+3 lines)
> and only misses velocities, you could delete all velocity lines and just
> restart your system (ntx=1) at 300K. This introduces a little disturbance
> but that should rapidly equilibrate away (in few ps). You may have to rest
> the box in xleap, as box coordinates are at the end of the rst-file.
> If your rst file is unusable, you could extract the last snapshot from the
> mdcrd-file and convert it into a rst (via ptraj) and restart your
> simulation from that point. crd-files have a lower accuracy than
> rst-files, but again, this should equilibrate away very quickly, compared
> to a multinanosecond simulation.
> if ptraj cannot read your crd file either, you can (if you used ASCII
> mdcrd format) parse through the mdcrd by yourself and see if you can
> extract a useful snapshot as far advanced in your trajectory as possible
> and restart from there.
> Each of these solutions would introduce a small but in my opinion
> acceptable inaccuracy into your system. That may be difficult to explain
> in a paper, but otherwise you should be ok.
> For materials and methods: ;-)
> "At 1448 ns simulation time, velocities were discarded and replaced by a
> Boltzmann velocity distribution at room temerature. This procedure was
> conducted to ensure proper system equilibration and stability. (innocent
> whistling)"
> Regards,
> Thomas
> Dr. Thomas Steinbrecher
> formerly at the
> BioMaps Institute
> Rutgers University
> 610 Taylor Rd.
> Piscataway, NJ 08854
> _______________________________________________
> AMBER mailing list
AMBER mailing list
Received on Tue Apr 10 2012 - 11:30:04 PDT
Custom Search