Hi,
just a few comments:
> ---------------------------nmode.nab-------------
> m = getpdb_prm( "1bfeMin.pdb","leaprc.ff96", "", 0 );
> mm_options( "cut=16.0, ntpr=50, gb=1, gbsa=1" );
...I'd advise against using a finite cutoff for "thorough" minimization.
The conjgrad() routine usually converges badly with a finite cutoff
(which might explain why there's only one cg-step done before newton()
starts below).
setting gbsa=1 seems also not advisable, since the second derivatives
for the gbsa part are not implemented (besides gbsa also hampers
convergence of cg()-minimization.
> ------------------
> Running: /usr/local/nab-5.1.2/bin/teLeap -s -f leap.in
> -I/usr/local/nab-5.1.2/leap/cmd -I/usr/local/nab-5.1.2/leap/parm
> -I/usr/local/nab-5.1.2/leap/prep -I/usr/local/nab-5.1.2/leap/lib > tleap.out
> Reading parm file (tprmtop)
> title:
>
> mm_options: cut=16.0
> mm_options: ntpr=50
> mm_options: gb=1
> mm_options: gbsa=1
> iter Total bad vdW elect. cons. genBorn frms
> ff: 0 -4272.34 627.38 -467.20 -3077.11 26.71 -1382.11 3.45e-01
> mm_options: ntpr=1
...the rms-gradient (frms) is much too large (by orders of magnitude) for
starting NR minimization. "Plain" NR wnly works if you are so close to the
next local minimium that all (non-zero) Eigenvalues of the hessian are
positive (not sure about that, but it seems a reasonable assumption).
Baseline: Try cut=999., gbsa=0 ans see what happens.
I hope that help,
good luck,
Andreas
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Sun Feb 17 2008 - 06:07:41 PST
