[gmx-users] Unable to minimize energy

[Paul Whitford]

This is likely an embarrassingly trivial problem, but let's see.

I am following the energy minimization scheme presented in the groningen
tutorial

http://md.chem.rug.nl/education/mdcourse/MDpract.html

I believe I follow the directions exactly, yet I can not minimize below
Fmax~500

I download 1AKI, as in the tutorial,
I issue

pdb2gmx -f 1AKI.pdb
grompp -f minim.mdp
mdrun -v

and I use the mdp file

"

title		= Minimization of Hen Egg White Lysozyme (1AKI.pdb)	; Title of run

; The following lines tell the program the standard locations where to
find certain files
cpp		= /lib/cpp	; Preprocessor
include		= -I../top	; Directories to include in the topology format

; Parameters describing what to do, when to stop and what to save
integrator	= steep		; Algorithm (steep = steepest descent minimization)
emtol		= 1.0		; Stop minimization when the maximum force < 1.0 kJ/mol
nsteps		= 200		; Maximum number of (minimization) steps to perform
nstenergy	= 10		; Write energies to disk every nstenergy steps
nstxtcout	= 10		; Write coordinates to disk every nstxtcout steps
xtc_grps	= Protein	; Which coordinate group(s) to write to disk
energygrps	= Protein	; Which energy group(s) to write to disk

; Parameters describing how to find the neighbors of each atom and how
to calculate the interactions
nstlist		= 5		; Frequency to update the neighbor list and long range forces
ns_type		= simple	; Method to determine neighbor list (simple, grid)
rlist		= 1.0		; Cut-off for making neighbor list (short range forces)
coulombtype	= cut-off	; Treatment of long range electrostatic interactions
rcoulomb	= 1.0		; long range electrostatic cut-off
rvdw		= 1.0		; long range Van der Waals cut-off
constraints	= none		; Bond types to replace by constraints
pbc		= no		; Periodic Boundary Conditions (yes/no)"


I tried multiple choices of potentials and I tried gromacs 3.3.2, 3.3.3 and
cvs, with single and double precision.  They all finish with the same
output(except single finishes in fewer steps)

"Step=   57, Dmax= 3.2e-11 nm, Epot= -9.59522e+03 Fmax= 8.07834e+02, atom=
478
Step=   60, Dmax= 9.6e-12 nm, Epot= -9.59522e+03 Fmax= 8.07834e+02, atom=
478
Step=   62, Dmax= 5.8e-12 nm, Epot= -9.59522e+03 Fmax= 8.07834e+02, atom=
478
Step=   65, Dmax= 1.7e-12 nm, Epot= -9.59522e+03 Fmax= 8.07834e+02, atom=
478
Step=   67, Dmax= 1.0e-12 nm, Epot= -9.45085e+03 Fmax= 8.07834e+02, atom=
478
Stepsize too small, or no change in energy.
Converged to machine precision,
but not to the requested precision Fmax < 1

writing lowest energy coordinates.

Back Off! I just backed up confout.gro to ./#confout.gro.5#

Steepest Descents converged to machine precision in 68 steps,
but did not reach the requested Fmax < 1.
Potential Energy  = -9.59521978368287e+03
Maximum force     =  8.07834306442108e+02 on atom 478
Norm of force     =  1.49469417798063e+02
"

This seems unreasonably high.  Are there some obvious mistakes people make
that result in this kind of behavior?  I have tried multiple pdb structure
and none of them can minimize below fmax~100.  I have also tried l-bfgs and
cg.  None of them can converge to the request precision.

thanks

-Paul
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20080522/6983a1e0/attachment.html>