Version 1.9 Build 1367

News	FAQ
Search	Home

Next: Discussion Up: Recommendations for the AIPS++ Telescope Model Previous: Splitting Calibration

Calibration as a service

I now turn to the second question: ``can we abstract the calibration of a Telescope into a class which can be called upon in selfcalibration?''.

For Imaging we need calibrated data. For this we require a service apply of Telescope Model which returns $\widehat{D}$ = C^-1(D, P). The instrumental parameters P are internal to a Telescope Model and are estimated using a service solve which solves D_cal = C( $\widehat{D}$ , P) where D_cal is the data set observed for calibration. Presumably gain numbers are stored in things like tables inside the Telescope Model, but that does not concern us here.

In summary, the Telescope Model appears to be a non-linear equation D = C( $\widehat{D}$ , P) in the same way that Imaging Model is a linear function D = AI. The services of Telescope Model are:

apply: returns the calibrated data $\widehat{D}$ = C^-1(D, P),
solve: changes the internal state to P found by solving (in some way) D_cal = C(D, P).

The inverse in the apply need not be a formal inverse.

Next: Discussion Up: Recommendations for the AIPS++ Telescope Model Previous: Splitting Calibration