Including a dust-like foreground component with pol=True results in unphysically large foreground bandpowers at all scales, and many negative eigenvalues in the Fisher matrix calculation, while the CMB bandpowers are all sensible. Likely a bug, but it's not clear where.
Including a dust-like foreground component with
pol=Trueresults in unphysically large foreground bandpowers at all scales, and many negative eigenvalues in the Fisher matrix calculation, while the CMB bandpowers are all sensible. Likely a bug, but it's not clear where.