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

Hello Jakob--

> I wish to write my own noe potential using the python interface. I  
> have noticed in a previous post in this list that you expect  
> python-implemented potentials to be significantly slower compared to  
> compiled code. Is this still true?

if the code is pure python it will be much slower than the compiled C++
and Fortran energy terms.

> Which version of python does xplor-nih use? If the numpy module  
> (numerical python) is installed I wondered if one could use the  
> numpy.array data structure to speed up the calculation.

Using numpy is probably a good idea. The current release of Xplor-NIH
has Python 2.4 embedded.

> More specifically, I wish to implement a new type of averaging of  
> ambiguous restraints, would it be better/possible to create a new  
> noePot class derived from the original class with a new averaging  
> method?
> 

You could derive from the NOEPot class, but I thin the only help that
would be is in reading XPLOR restraint tables. You would have to
override the calcEnergy and and calcEnergyAndDerivs methods, which is
just about all of the numerical guts. So probably better to start from
scratch. 

However, honestly, the easiest way to add a new restraint averaging type
is to install swig-1.3.22 and modify nmr/noePot.cc, as it would probably
only require changing a few lines of code.

best regards--
Charles

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Processed by Mailcrypt 3.5.8+ <http://mailcrypt.sourceforge.net/>

iD8DBQFGyaFKPK2zrJwS/lYRAnAWAJsEmugzCiJsXkrl5EvgHBvsOvVVuQCdEPhF
LieDl7tlx35+053N9hx9fWI=
=5qFw
-----END PGP SIGNATURE-----

Reply via email to