Sv: Re: [gmx-users] double and single precision

Edvin Erdtman Edvin.Erdtman at
Tue Feb 21 16:53:13 CET 2012

Hi GMX-users!

If someone else get the same problem. I will present how I could
recover the edr files:

1. Calculate the number of bytes for header consisted of in single and
double precision, respectively.
   run some short test runs to calculate this
2. Calculate the number of bytes for each frame. The first frame took
apparently less space than the others. I guess it is due to the average
and sum data, that one can see from gmxdump -e
3. Calculate the number of frames in the first simulation with gmxcheck
3. I used head, tail and cat to cut and paste in the files. 

#bytes of corrupted file# - #bytes of header DP# - "#bytes per frame# *
#frames in first run# 

head -c #bytes of header SP# singleprec.edr > headerSP

tail -c YYYY corruptedr.edr | cat headerSP - > newedr.edr

check the new file with gmxcheck -e

I got these byte numbers (NPT-simulation):  
Single precision: Header: 964b, 1st frame: 240b, then 576 b/frame
Double precision: Header+1st frame: 1376, then 1084 b/frame

Thanks to Berk Hess who helped me a lot!


Dr. Edvin Erdtman
Instutitionen Ingenjörshögskolan
501 90 BORÅS
033- 435 45 37
edvin.erdtman at 

>>> 2012-02-20 kl. 09:38, skrev Mark Abraham <Mark.Abraham at>
> On 20/02/2012 7:29 PM, Edvin Erdtman wrote:
> > Hi
> >
> > So the best I can do now is to recreate the first part, and rerun
> > the simulations?
> >
> If you have a full frame (positions+velocities+maybe energies, or a 
> checkpoint file) then I expect so. Save your crossbreed files in case

> someone has a better idea.
> Mark
> -- 
> gmx-users mailing list    gmx-users at 
> Please search the archive at 
> before posting!
> Please don't post (un)subscribe requests to the list. Use the 
> www interface or send it to gmx-users-request at
> Can't post? Read 

More information about the gromacs.org_gmx-users mailing list