analysis: analysis::fitting_toolkit::fit::svd_damping

def svd_damping_recipe	(	self,
		old_chi2 = `None`,
		new_chi2 = `None`
	)

A lot like lm_lambda_recipe but for svd fits. It adjusts the damping depending on the behavior of chi^2

Parameters:

	old_chi2	The value of chi^2 from the previous fit step
	new_chi2	The value of chi^2 from the current fit step.

Definition at line 671 of file fitting_toolkit.py.

00671                                                               :
00672         "adjust the damping parameter based on the old and new chi^2 values"
00673         if old_chi2 is None:
00674             self.svd_damping=1.0
00675         elif old_chi2 < new_chi2:
00676             if new_chi2/old_chi2 < 1.1:
00677                 self.svd_damping *=0.8 #on a slightly bad fit (bobbling), don't penalize too badly
00678             else:
00679                 self.svd_damping  *= 0.25 
00680         else:
00681             self.svd_damping   = min(self.svd_damping *2.0, 1.0)
00682 
    ##