29-Nov-2017
Software Changes 29-Nov-2017 to Support Calibrated Eventlist (CBE)
Normalization
CBE normalization was implemented incorrectly in March 2017. It is now
done correctly, and the correction factors are now available to the user.
- CBE normalization attempts to correct for changes in detector
sensitivity that have an effect on the non-visibility-based imaging
algorithms. (It is similar to the visibility normalization that is
performed in the visibility-based image algorithms when the control
parameter vis_normalize is set to 1). When CBE normalization
is enabled, correction factors are computed for each detector selected,
such that the predicted count rates for the true image are within
statistical limits of those observed . We find that over the
course of the mission that knowledge has been degraded for various
reasons, so we use these factors to try to compensate. This is a
purely heuristic correction and its effects and underlying basis cannot
be assured (the same caveat applies to the visibility normalization).
- Cbe_normalize is a control parameter of the hsi_image_strategy
abstract class. This class is inherited by the hsi_image_raw
class. If cbe_normalize is set, it changes a weighting factor,
called a corrfactor, from the default value of 1. The corrfactors
are applied in all of the non-visibility imaging algorithms (other than
back projection).
- The corrfactors (stored as the info parameter array cbe_corr_factors)
are used in two ways. In Clean, they are used as multiplicative
factors on the dirty map for each sub-collimator before summing to get
the dirty map used with Clean. In Pixon, EM, and ForwardFit, the factors
are divided into the time profiles that are predicted by the image found
at any iteration before comparing to the observed count rate.
The default for cbe_normalize is 0.
To enable CBE normalization: o->set, /cbe_normalize
To disable CBE normalization: o->set, cbe_normalize=0
To retrieve the correction factors: corr = o->get(/cbe_corr_factors)