Opened 14 years ago

Closed 14 years ago

#196 closed bug (fixed)

stability Igor interface

Reported by: porcar@… Owned by: srkline
Priority: major Milestone: Analysis Wish List
Component: Analysis Keywords:
Cc: Blocking:
Task:

Description

The new analysis interface is not as stable as the old curve fitting routine.
Many times, it lost its mind and I have to use the old curve fit option (a bit more painful for epsilon and constraint) to resume fitting.

Change History (1)

comment:1 Changed 14 years ago by srkline

  • Resolution set to fixed
  • Status changed from new to closed

Lionel was having issues with the coreshellcylinder model stopping immediately after starting a fit, giving a "singular matrix" error, returning chi-squared of zero. The problem seemed to be that he had the scale and the shell SLD free (and the shell thickness) for the fit. Apparently these are enough correlated parameters to cause this particular model to have conniptions and take a really bad step, quite quickly and consistently, even with constraints on dimensions and SLD (but not scale).

Advised Lionel to fix more of the correlated parameters, this converged for me.

Why the menu option for the interface works better for him, I don't know.

Note: See TracTickets for help on using tickets.