-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi Haydyn--

> Hi guys, I've been using XPLOR-NIH for a while now but until recently 
> had not come across the following error:
> 
> --> ERROR: matrix is not a rotation matrix.


Is this by chance using the original XPLOR torsion angle dynamics code
(DYNAmics TORSion)?? You probably should switch over to using DYNAmics
INTErnal instead. See eginput/protG/anneal.inp for an example of this.

> 
> This seems to happen when I use a torsion angle dynamics protocol, but 
> change my noe potential  from:
> 
>    potential  all soft
> 
> to:
> 
>    potential  all square
> 
> 
> The protocol I'm using for simulated annealing works perfectly when I 
> use the "soft" noe potential function, and dies with the aforementioned 
> error if this is changed to "square".   I've tried to isolate the 
> problem, but all I can conclude at this stage is that "soft" lets things 
> work, and "square" doesn't.  Can anyone give me the low-down on what 
> silly mistake I have probably made?  I did find something about ARIA/CNS 
> giving this error when using a torsion angle dynamics protocol, and that 
> it can be a timestep issue.  I feel quite happy with a 0.015 ps 
> timestep  and find that reducing it to less desirable values fails to 
> fix the problem.

If its a timestep issue, it's not too surprising that a softer
potential is more successful. The above mentioned internal dynamics
routine has a variable timestep facility which should be able to dance
around bad steps. [In addition, the code executes about three times
faster.] 

Please let me know if you need any further assistance with this.

Charles
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.1 (GNU/Linux)
Comment: Processed by Mailcrypt 3.5.8 <http://mailcrypt.sourceforge.net/>

iD8DBQE/DW2HPK2zrJwS/lYRAueZAJ0YajWA/Q1Cdz66UJdzmXjcfa04PQCeNywU
uYwKDz6RkHS6Fbtm6Fhf8NY=
=EQj4
-----END PGP SIGNATURE-----

Reply via email to