Hi Istvan,
On 23/2/2011, at 2:24 p.m., istvan.kolossvary.hu wrote:
> Yes, I mean a NAB program. A sample program is included in the
> AmberTools manual. Like in Sander, NAB has a wrapper routine with an
> infinite loop calling XMIN every time a new gradient has been
> computed. You can dump the coordinates at that level at every step,
> even during line minimization.
I never did find out how this is to be done, since the minimisation algorithm seems to be called only once, and returns the final coordinates on completion. At least, that's how it appears in the NAB script.
I did, however, print out the coordinates after the first step, by requesting a minimisation of precisely one step. The coordinates are very different from the coordinates after one step of minimisation in sander. The energies and gradients, though, are almost the same. Specifically, only EEL14 was different once rounding was considered: 0.972 (NAB) vs. 0.957 (sander). I don't know why this should be - I would expect the energies to be identical. But it does seem clear that the coordinates are getting broken somehow in sander.
Ben
--
For greater security, I support S/MIME encryption.
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
- application/pkcs7-signature attachment: smime.p7s
Received on Mon Feb 28 2011 - 13:30:04 PST
