[gmx-users] Some poblems~

Erik Lindahl lindahl at csb.stanford.edu
Thu Feb 19 17:58:01 CET 2004


> "Reading file topol.tpr, VERSION 3.1.4 (single precision)"
>  Reading file topol.tpr, VERSION 3.1.4 (single precision)
> It seems double precision not work

This just means the file topol.tpr is in single precision. Gromacs 
files are completely portable both between different architectures and 
floating-point precision, but obviously you won't get more accuracy 
that what is in the file.

The pdb file or whatever starting structure you had probably didn't 
contain double precision coordinates though, so just consider it a 
notice :-)

> 2.use "./configure --disable-float" and "./configure --enable-mpi" 
> individually and found the following messges
> * You are compiling a double precision version of the package -
>    program names will be suffixed with _d to avoid overwriting single
>    precision files. You can override it with --program-suffix
> * Seems you are compiling with MPI support. You can install the MPI-
>   enabled programs with suffixed names to have both MPI and non-MPI
>   versions. This is useful e.g. on supercomputers where you usually
>   cannot run MPI-linked programs on the login node.
>   Set a suffix with e.g. --program-suffix=_mpi (or _mpi_d for double).
>   You only need MPI for mdrun, so if you already have non-MPI stuff
>   installed you can issue make mdrun; make install-mdrun.
> Does it means I 've only compiled MPI,not double precison?

No, these messages are just tips to avoid overwriting any single 
precision or non-mpi version with the new files.

The precision is echo:ed explicitly in the program name that scrolls by 
when you run most gromacs programs - look for a line like

        :-)  /usr/local/bin/grompp  (-:


        :-)  /usr/local/bin/grompp  (double precision) (-:

> 3.When I run a minmization job with "single precision"(maybe),I found 
> the following messages
> Negative w:  -6.111514486374e+13
> z=  -3.430048800000e+07
> gpa=   2.322916848568e+07, gpb=   5.327950593513e+07
> a=   0.000000000000e+00, b=   7.583112298282e-07
> EpotA=  -5.510806152344e+03, EpotB=  -5.482796875000e+03
> Negative number for sqrt encountered (-61115144863744.000000)
> Terminating minimization
> writing lowest energy coordinates.
> Back Off! I just backed up traj.trr to ./#traj.trr.1#
> Back Off! I just backed up confout.gro to ./#confout.gro.1#
> Conjugate Gradients did not converge in 403 steps
> Potential Energy  = -5.51081e+03
> Maximum force:  1.73107e+03
> I searched the forum and found maybe it's the result when using single 
> precision.Is my supposiiton correct?

It could happen in double too with the old CG minimizer in 3.1.4. 
Upgrade to 3.2 - you'll get a much better minimizer that avoids this 
problem entirely!

> 4.Finally,everytime I used grompp.I always found the waring "protein 
> has non-zero total charge"
> Is it a risk for my job?or how could I improve it?

Depends on what the charge is. If you have 100,000 atoms and the net 
charge is 0.001 it is probably just a rounding error from the 
summation. If you just have a small protein with a couple of unit 
charges you might want to add your own counterions.



More information about the gromacs.org_gmx-users mailing list