HoloPy

Alternate Minimizer

Registered by Tom Dimiduk

We have most of an openopt wrapper in holopy. A little work is needed to get fit result saving working correctly with openopt.

More importantly, we need to actually run fits with openopt and start to build up a knowledge base of which of its minimizers work well, what their convregence characteristics are, when they are better than nmpfit, etc.

VNM: I suggest using OpenOpt (http://openopt.org/). It has a large number of minimizers (http://openopt.org/Problems), including nonlinear (http://openopt.org/NLP) and global (http://openopt.org/GLP) minimizers, and it has a general framework for handling constraints. Also it's packaged for Ubuntu and Enthought. It does not have a constrained Levenberg-Marquardt minimizer, but it does have several minimizers that appear more advanced than LM.

Blueprint information

Status:
Complete
Approver:
None
Priority:
High
Drafter:
None
Direction:
Approved
Assignee:
Tom Dimiduk
Definition:
Approved
Series goal:
None
Implementation:
Implemented
Milestone target:
milestone icon 2.0
Started by
Vinothan N. Manoharan
Completed by
Tom Dimiduk

Related branches

Sprints

Whiteboard

Need to remember to add a note to the docs that user needs both OpenOpt and DerApproximator installed for OpenOpt to work

(?)

Work Items

Dependency tree

* Blueprints in grey have been implemented.

This blueprint contains Public information 
Everyone can see this information.
Edit subscription

Subscribers

No subscribers.