Hi Troels,
You should note that the minimise.execute user function does not allow
the log-barrier or the method of multipliers algorithm to be chosen.
This is instead via the constraint flag. Therefore this logic will
not work. The choice of constraint algorithm occurs in the back-end
code, as it is uses dead simple logic:
No gradients implemented = log-barrier.
Gradients implemented = Method of Multipliers.
You cannot mix these as the Method of Multipliers absolutely requires
the gradient, and the log-barrier method currently does not implement
its gradient.
Regards,
Edward
On 1 September 2014 14:26, <tlinnet@xxxxxxxxxxxxx> wrote:
Author: tlinnet
Date: Mon Sep 1 14:26:08 2014
New Revision: 25498
URL: http://svn.gna.org/viewcvs/relax?rev=25498&view=rev
Log:
Allow R2eff model to reach constrained methods in minfx through relax.
This is in systemtest Relax_disp.verify_estimate_r2eff_err_compare_mc()
This though not seem to be supported.
Modified:
trunk/specific_analyses/relax_disp/api.py
Modified: trunk/specific_analyses/relax_disp/api.py
URL:
http://svn.gna.org/viewcvs/relax/trunk/specific_analyses/relax_disp/api.py?rev=25498&r1=25497&r2=25498&view=diff
==============================================================================
--- trunk/specific_analyses/relax_disp/api.py (original)
+++ trunk/specific_analyses/relax_disp/api.py Mon Sep 1 14:26:08 2014
@@ -595,6 +595,18 @@
elif match('^[Nn]ewton$', algor):
allow = True
+ # Newton minimisation.
+ elif match('^[Nn]ewton$', algor):
+ allow = True
+
+ # Constrained method, Method of Multipliers.
+ elif match('^[Mm][Oo][Mm]$', algor) or match('[Mm]ethod of
[Mm]ultipliers$', algor):
+ allow = True
+
+ # Constrained method, Logarithmic barrier function.
+ elif match('^[Ll]og [Bb]arrier$', algor):
+ allow = True
+
# If the Jacobian and Hessian matrix have not been specified
for fitting, 'simplex' should be used.
else:
if match('^[Gg]rid$', algor):
_______________________________________________
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
Re: r25498 - /trunk/specific_analyses/relax_disp/api.py