Fourier Spectroscopy: A Bayesian Way

Table 3

Comparison of standard model and Bayesian model for spectral increment = 3.66 GHz. (second column) is well approximated by when the settings of the third column are chosen. However, an aliasing feature is present in the spectral domain 3000–3500 GHz (see Figure 14(a)). This feature disappears when the spatial settings are changed (fourth column) to the values at the maximum of the settings posterior of (fifth column). The maximum of the settings posterior reveals that the Nyquist assumptions ( = 0 GHz and = 3747.4 GHz) are not plausible, because an overfitting of the data is made.

Model
(GHz)	0	0	0	33.614
(GHz)	3747.4	3747.4	3747.4	911.912
( V^2)	N/A	19.8182	19.8182	1.0304
( V^2)	N/A	129.8670	129.8670	5.7641
(mm)
(m)	0	0
(nV)	N/A	61.735	61.735	62.146
(V/m)	N/A	5.822	5.822	6.007
(V/m^2)	N/A	0.000087	0.000087	10.102
	N/A	1051396.55	1051400.41	1052037.88
Odds	N/A			1
Aliasing feature	Yes	Yes	No	No
Overfitting data	Yes	Yes	Yes	No