QUBIC

The user will have the choice with:\nFMM or CMM\nnumber of reconstructed bands or components (sky model)\nbeware: depending on which components we are dealing with, the noise in CMM components might be very different (eg. if one includes something as low as synchrotron, the uncertainty will be large and will propagate to other components due to noise correlations). This probably means that we'll need to find a model for the components covariance as a function of their overall amplitude. SOme of this might already be in the analytical work done by Mathias.