Re: r22822 - in /trunk/test_suite/system_tests: relax_disp.py scripts/relax_disp/hansen_data.py -- April 22, 2014 - 21:22

Wow, the kex grid search bounds change really made such a difference.
For a test I turned off the Relax_disp.opt_func_tol and
Relax_disp.opt_max_iterations value changing in the script.  The
correct values were found with the original 6 increments, but the test
took longer.  So increasing the increments to 7 is the best and
fastest option.

Cheers,

Edward


On 22 April 2014 21:09, Troels Emtekær Linnet <tlinnet@xxxxxxxxxxxxx> wrote:
> Hi Edward.
>
> If I did not increase the GRID with +1, the found grid R2 values would be
> horrible,
> and the minimised values would be very wrong.
>
> Try to change it back, and see what happens.
>
> I think it is quite unstable, with so low grid values.
>
>
>
>
>
> 2014-04-22 20:53 GMT+02:00 Edward d'Auvergne <edward@xxxxxxxxxxxxx>:
> > Hi Troels,
> >
> > Did you see a change in the system test time?  The lower grid search
> > sizes are simply to allow the tests to complete faster.
> >
> > Regards,
> >
> > Edward
> >
> >
> >
> > On 22 April 2014 20:48,  <tlinnet@xxxxxxxxxxxxx> wrote:
> > > Author: tlinnet
> > > Date: Tue Apr 22 20:48:18 2014
> > > New Revision: 22822
> > >
> > > URL: http://svn.gna.org/viewcvs/relax?rev=22822&view=rev
> > > Log:
> > > Changes to systemtest Relax_disp.test_hansen_cpmg_data_auto_analysis.
> > >
> > > Regarding bug #21953, (https://gna.org/bugs/index.php?21953) - Weird
> > > performance of grid search
> > >
> > > The grid search needed to be increased by +1 and lowering of some
> > > results by one digit.
> > >
> > > Modified:
> > >     trunk/test_suite/system_tests/relax_disp.py
> > >     trunk/test_suite/system_tests/scripts/relax_disp/hansen_data.py
> > >
> > > Modified: trunk/test_suite/system_tests/relax_disp.py
> > > URL:
> > > http://svn.gna.org/viewcvs/relax/trunk/test_suite/system_tests/relax_disp.py?rev=22822&r1=22821&r2=22822&view=diff
> > >
> > > ==============================================================================
> > > --- trunk/test_suite/system_tests/relax_disp.py (original)
> > > +++ trunk/test_suite/system_tests/relax_disp.py Tue Apr 22 20:48:18 2014
> > > @@ -624,7 +624,7 @@
> > >          print("%-20s %20.15g %20.15g" % ("phi_ex", spin70.phi_ex,
> > > spin71.phi_ex))
> > >          print("%-20s %20.15g %20.15g" % ("kex", spin70.kex,
> > > spin71.kex))
> > >          print("%-20s %20.15g %20.15g\n" % ("chi2", spin70.chi2,
> > > spin71.chi2))
> > > -        self.assertAlmostEqual(spin70.r2[r20_key1], 6.74326615264889,
> > > 3)
> > > +        self.assertAlmostEqual(spin70.r2[r20_key1], 6.74326615264889,
> > > 2)
> > >          self.assertAlmostEqual(spin70.r2[r20_key2], 6.57331164382438,
> > > 2)
> > >          self.assertAlmostEqual(spin70.phi_ex, 0.312767653822936, 3)
> > >          self.assertAlmostEqual(spin70.kex/10000,
> > > 4723.44390412119/10000, 3)
> > > @@ -648,7 +648,7 @@
> > >          print("%-20s %20.15g %20.15g" % ("kex", spin70.kex,
> > > spin71.kex))
> > >          print("%-20s %20.15g %20.15g\n" % ("chi2", spin70.chi2,
> > > spin71.chi2))
> > >          self.assertAlmostEqual(spin70.r2[r20_key1], 6.97233943292193,
> > > 3)
> > > -        self.assertAlmostEqual(spin70.r2[r20_key2], 9.409506394526, 3)
> > > +        self.assertAlmostEqual(spin70.r2[r20_key2], 9.409506394526, 2)
> > >          self.assertAlmostEqual(spin70.pA, 0.989856804525044, 3)
> > >          self.assertAlmostEqual(spin70.dw, 5.60889078920945, 3)
> > >          self.assertAlmostEqual(spin70.kex/10000,
> > > 1753.01607073019/10000, 3)
> > > @@ -656,7 +656,7 @@
> > >          self.assertAlmostEqual(spin71.r2[r20_key1], 5.003171547206, 3)
> > >          self.assertAlmostEqual(spin71.r2[r20_key2], 6.90210797727492,
> > > 3)
> > >          self.assertAlmostEqual(spin71.pA, 0.985922406455826, 3)
> > > -        self.assertAlmostEqual(spin71.dw, 2.00500965892672, 3)
> > > +        self.assertAlmostEqual(spin71.dw, 2.00500965892672, 2)
> > >          self.assertAlmostEqual(spin71.kex/10000,
> > > 2481.10839579617/10000, 3)
> > >          self.assertAlmostEqual(spin71.chi2, 15.6595374286822, 3)
> > >
> > >
> > > Modified:
> > > trunk/test_suite/system_tests/scripts/relax_disp/hansen_data.py
> > > URL:
> > > http://svn.gna.org/viewcvs/relax/trunk/test_suite/system_tests/scripts/relax_disp/hansen_data.py?rev=22822&r1=22821&r2=22822&view=diff
> > >
> > > ==============================================================================
> > > --- trunk/test_suite/system_tests/scripts/relax_disp/hansen_data.py
> > > (original)
> > > +++ trunk/test_suite/system_tests/scripts/relax_disp/hansen_data.py
> > > Tue Apr 22 20:48:18 2014
> > > @@ -23,7 +23,7 @@
> > >      ds.numeric_only = False
> > >
> > >  # The grid search size (the number of increments per dimension).
> > > -GRID_INC = 6
> > > +GRID_INC = 7
> > >
> > >  # The number of Monte Carlo simulations to be used for error analysis
> > > at the end of the analysis.
> > >  MC_NUM = 3
> > >
> > >
> > > _______________________________________________
> > > relax (http://www.nmr-relax.com)
> > >
> > > This is the relax-commits mailing list
> > > relax-commits@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-commits
> >
> > _______________________________________________
> > 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