Re: Precision with ModelFree (palmer.py) -- January 15, 2008 - 18:49

Hi,

Those differences are much less than I expected.  I was worried that
Modelfree might either ignore the higher precision or maybe even fail
with the different mfpar file.  But your test shows that everything is
fine.  The change, as you suggest, can go straight into the 1.3 line.

Cheers,

Edward



On Jan 15, 2008 6:18 PM, Sébastien Morin <sebastien.morin.1@xxxxxxxxx> wrote:
> Hi Ed,
>
> I checked and ModelFree accepts the higher precision gyromagnetic value 
> (-2.7126
> instead of -2.713). Giving the higher precision value doesn't cause any 
> error or
> warning and the value is already with 4 decimals in the mfout file.
>
> I tried two runs with the palmer.py script a the three stages. The 
> optimization
> is indeed slightly different.
>
> I used an isotropic diffusion tensor with 5 residues, doing 100 Monte Carlo
> simulations. You'll see in the attached file that differences are small.
>
> The modification in the 'create_mfpar' method of the generic_fns.palmer 
> module
> was changing line 382 (in the repository version of the 1.2 line) from :
>
> file.write('%-8.3f' % (self.relax.data.gx / 1e7))
>
> to :
>
> file.write('%-8.4f' % (self.relax.data.gx / 1e7))
>
> I think that this change should go in the 1.3 line.
>
> Do you think we need more tests ?
>
> Cheers,
>
>
> Séb  :)
>
>
>
>
>
>
> Selon Edward d'Auvergne <edward.dauvergne@xxxxxxxxx>, 14.01.2008:
>
>
> > Hi,
> >
> > The reason for the rounding is because I was mimicking the example of
> > the 'mfpar' file in the Modelfree manual
> (http://www.cumc.columbia.edu/dept/gsas/biochem/labs/palmer/software/modelfree_manual.pdf).
> >  There the numbers are given to 3 decimal places.  This may make a bit
> > of a difference in the optimisation, but I don't know if Modelfree
> > will accept higher precision numbers.  To increase this number, it
> > will have to be thoroughly tested using Modelfree.  It will involve
> > modifying the 'create_mfpar' method in the generic_fns.palmer module
> > (I'll be changing this one day soon from a class method to a module
> > function).  This change, if it doesn't cause problems with Modelfree,
> > should go into the 1.3 line after approval.
> >
> > Cheers,
> >
> > Edward
> >
> >
> > On Jan 11, 2008 6:54 PM, Sebastien Morin <sebastien.morin.1@xxxxxxxxx> 
> > wrote:
> > > Hi,
> > >
> > > I used the palmer.py script to run ModelFree from within relax 1.2
> > > (repository).
> > >
> > > The value for the 15N gyromagnetic ratio in the file
> > > 'generic_fns/nuclei.py' is -2.7126e7.
> > >
> > > However, in the 'mfpar' file created by the script 'palmer.py', the
> > > value specified is rounded up (-2.713).
> > >
> > > Is it something intended or should this be changed ?
> > >
> > > Cheers
> > >
> > >
> > > Séb
> > >
> > > --
> > > Sebastien Morin
> > > Etudiant au PhD en biochimie
> > > Laboratoire de resonance magnetique nucleaire
> > > Dr Stephane Gagne
> > > CREFSIP (Universite Laval, Quebec, CANADA)
> > > 1-418-656-2131 #4530
> > >
> > >
> > >
> > > _______________________________________________
> > > relax (http://nmr-relax.com)
> > >
> > > This is the relax-devel mailing list
> > > relax-devel@xxxxxxx
> > >
> > > To unsubscribe from this list, get a password
> > > reminder, or change your subscription options,
> > > visit the list information page at
> > > https://mail.gna.org/listinfo/relax-devel
>
>
>
> ------------------------
> Sébastien Morin
> Étudiant M.Sc. Biochimie
> Laboratoire S. Gagné
> 3252 Pav. Marchand (Université Laval)
> Tél : (418) 656-2131 #4530
> Fax : (418) 656-7176
> e-mail : sebastien.morin.1@xxxxxxxxx