-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Hello Josh and Geoff-- Goeff wrote: > > Another issue may be if you have very large couplings you may also > > create very large forces on the atoms, and you will need to > > rebalance (probably lower) the RDC force constants. You may need to > > make the time step shorter, and refine longer, similar to Choy et > > al. Journal of Biomolecular NMR, 21: 31-40, 2001. > > If you look at figure 4 of that paper, the runs would complete, but > the RMS or E<SANI> would hit a point in the minimization where it > would get stuck. Plotting that from the log file may give you some > insight. Slowing the dynamics allowed it to smoothly reach a > solution. > Do note that that work used the old torsion angle dynamics facility within the XPLOR (or CNS) interface (dynamics torsion) which employs strictly fixed timestep sizes. The standard Python scripts use the IVM with variable timesteps such that those types of discontinuities will be much less likely. It is true, however, that one must always be mindful of balancing force constants. Pretty much the only way to detect a force constant imbalance is getting unexpectedly bad fits for one or more terms. best regards-- Charles -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.10 (GNU/Linux) Comment: Processed by Mailcrypt 3.5.9 <http://mailcrypt.sourceforge.net/> iEYEARECAAYFAkzi4KMACgkQPK2zrJwS/lZv4gCffGcaPL8ZQLl2mbY/8YB1R7zB tFwAnjaoPirw2q5OPAlhRozWEyA/0j88 =3T/s -----END PGP SIGNATURE-----
