AMBER Archive (2007)

Subject: Re: AMBER: gb minimization using nab

From: Andreas Svrcek-Seiler (svrci_at_tbi.univie.ac.at)
Date: Tue Nov 13 2007 - 05:58:05 CST

Hi,
> I'm using the newton and nmode functions of NAB to perform a
> Newton-Raphson
> minimization followed by a vibrational frequencies calculation in implicit
> solvation. When I use the option gb=1, I've found correct GB energy compared
> with what I've obtained using the Sander module of Amber 8. But when I switch
> to gb=2, I've got a GB energy really much lower, which gives rise to
> incorrect frequencies (a lot of them are negative then).
...your output shows that the rms gradient is still ~20kcal/(mol A) when
entering nmode.
normal modes should be computed at (as close to as possible) a local
minimum of your energy function. So you should do conjgrad()
until frms is down to -say- 1.e-3 or even lower before
tryinng newton-raphson (down to -say- 1.e-10).
Only then is should be safe to call nmode().

Besides (I believe) the second derivatives are only available for
gb=1 (though I'm not sure - in doubt read the source)

good luck,
Andreas
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu