Re: [bug #22211] The 'LM63 3-site' relaxation dispersion model optimisation is not perfect. -- June 17, 2014 - 15:31

You might also be interested to know that it affects some of the
problems where pA is stuck at 0.5.  I can see this in the
Relax_disp.test_tp02_data_to_mp05 system test which currently fails.
Just run:

$ relax -sd Relax_disp.test_tp02_data_to_mp05

with and without the new minfx and look at the optimisation for the
second spin.  It looks like the tuned log barrier algorithm can handle
these tough edge cases much better now.

Regards,

Edward



On 17 June 2014 15:00, Edward d'Auvergne <edward@xxxxxxxxxxxxx> wrote:
> It'll soon be version 1.0.7.  You'll probably be interested to note
> that, for some problems, this also significantly decreases the number
> of iterations required to find the minimum.  The log barrier algorithm
> now converges much more quickly.  Hence this speeds up the dispersion
> analyses as well.
>
> Regards,
>
> Edward
>
>
>
> On 17 June 2014 14:55, Troels Emtekær Linnet <tlinnet@xxxxxxxxxxxxx> wrote:
> > Is that version 1.0.6 ?
> >
> > Or is it newer?
> >
> >
> > 2014-06-17 14:48 GMT+02:00 Edward d'Auvergne <edward@xxxxxxxxxxxxx>:
> > > Hi Troels,
> > >
> > > This is a very important change for the dispersion analysis (and to
> > > the frame order analysis as well).  I have found that the log barrier
> > > constraint algorithm was not tuned optimally in the minfx library
> > > (https://gna.org/projects/minfx/), and have fixed it upstream
> > > (http://article.gmane.org/gmane.science.mathematics.minfx.scm/25).
> > > This causes a number of tests to now fail in the trunk, simply due to
> > > the better optimisation, which I'll now look at and fix.  You should
> > > note though that this significantly improves the optimisation problem
> > > for many constrained dispersion models!  You should have a look at the
> > > newest minfx code.
> > >
> > > Regards,
> > >
> > > Edward
> > >
> > >
> > > On 17 June 2014 14:41, Edward d  Auvergne
> > > <NO-REPLY.INVALID-ADDRESS@xxxxxxx> wrote:
> > > > Update of bug #22211 (project relax):
> > > >
> > > >                   Status:                    None => Invalid
> > > >              Open/Closed:                    Open => Closed
> > > >
> > > >     _______________________________________________________
> > > >
> > > > Follow-up Comment #1:
> > > >
> > > > This is a non-optimal minfx setting for the log barrier constraint
> > > > algorithm
> > > > which has been fixed upstream
> > > > (http://article.gmane.org/gmane.science.mathematics.minfx.scm/25).  To
> > > > use the
> > > > newest minfx version, either wait for a new relax release, or do the
> > > > following:
> > > >
> > > > 1)  Delete the minfx directory in the base relax directory (if there is
> > > > one).
> > > >
> > > > 2)  In the relax base directory, obtain the newest minfx code using the
> > > > subversion command:
> > > >
> > > > $ svn co http://svn.gna.org/svn/minfx/trunk/minfx
> > > >
> > > >
> > > > Alternatively, if you know how, just install the trunk version of minfx
> > > > as a
> > > > Python package.
> > > >
> > > >     _______________________________________________________
> > > >
> > > > Reply to this item at:
> > > >
> > > >   <http://gna.org/bugs/?22211>
> > > >
> > > > _______________________________________________
> > > >   Message sent via/by Gna!
> > > >   http://gna.org/
> > >
> > > _______________________________________________
> > > relax (http://www.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