[gmx-users] pull code for speptide : floating exception

ysun at mie.utoronto.ca ysun at mie.utoronto.ca
Mon Aug 18 21:41:02 CEST 2003


Dear Frauke 

I tried to modify the file  src/mdlib/pull.c  by covering the two lines:  

 /* while (dr[m] >  0.5*box[m][m])  rvec_dec(dr,box[m]);*/
 /* while (dr[m] < -0.5*box[m][m]) rvec_inc(dr,box[m]);*/

it continued showing error "floating exception".
I am studying this part to find out any other potential bugs.

Thanks!

Sun 



 


Quoting Frauke Meyer <fmeyer at ix.urz.uni-heidelberg.de>:

> Hi Sun,
> >
> > As you advised. I tried to skip -pn -pi, it worked and got output
> results.
> > However, when added -pn -pi in the mdrun, it had error: "floating
> exception"
> > just occurred at the beginning (may be the first step)!
> >
> > YOU mentioned you met this problem, due to step=0 the force equal zero, one
> of
> > the function caused the error, how did you solve your problem at that
> time?
> >
> 
> change the do_afm routine in pull.c:
> the lines that were causing the error in my case were
>   for (i=0;i<pull->pull.n;i++) {
>     /* compute how far the springs are stretched */
>      for (m=DIM-1; m>=0; m--) {
>       while (dr[m] >  0.5*box[m][m])  rvec_dec(dr,box[m]);
>       while (dr[m] < -0.5*box[m][m]) rvec_inc(dr,box[m]);
> 
>       dr[m]=pull->dims[m]*(pull->pull.spring[i][m]-pull->pull.x_unc[i][m]);
>     }
> 
> This is a bug, rvec_dec and rvec_inc change dr, which is anyway
> recalculated in the following line (dr[m]=...)
> So I just skipped the two while lines and everythin worked out finely so
> far. I doubt the correctness of them: why should the spring be less or
> more extended depending ont he box length?
> 
> Good luck,
> Frauke
> --
> Frauke Meyer
> Drug Discovery and Design Center
> Shanghai Institute of Materia Medica
> 555 Zu Chong Zhi Road, Zhangjiang Hi-Tech Park
> Shanghai 201203
> Tel: +86 21 50806600*1201, Fax: +86 21 50806600*1205
> http://www.dddc.ac.cn, e-mail:frauke.meyer at mpi-bpc.mpg.de
> 
> 
> 





More information about the gromacs.org_gmx-users mailing list