Next: The Details of the
Up: Further Discussion and Conclusions
Previous: Framework of the Models
  Contents
Conclusions
This section describes the conclusions, obtained from the Bayesian
inference process, about the parameters in the two models, and about
which model is correct.
- The electrostatic potential in the sample, as a function of
cobalt thickness, in the null model, is as shown in figure
5.8.
- The electrostatic potential in the sample, as a function of
cobalt thickness, in the main model, is as shown in figure
5.9.
- The magnetic flux density in the sample, as a function of
cobalt thickness, in the main model, is as shown in figure
5.10.
- The posterior probability of the null model, in which cobalt
ferro-magnetism is not detected, is estimated at
, and that of the main model, in which cobalt ferro-magnetism is detected, at
.
At first glance, the precision of the inferred electrostatic
potentials and magnetic flux densities appears a rather unimpressive
return on the experimental effort involved in obtaining the
measurements. In addition, a comparison (tables 5.4,
5.5) of the beliefs held about all of the parameters,
including those concerned with instrumental calibration and with
transient instrumental characteristics, before and after the data in
chapter 5, along with the published calibration data
(figures 5.6, 5.7,) have been taken into
account, shows a broadening, rather than the narrowing that one might
expect, of the probability distribution over the parameter set, due to
the data.
However, the very decisive choice of model with successful detection
of the Weiss field as the better model supplies a greater reward.
Table 5.4:
The Effect of the Experimental Evidence on Beliefs about
Parameters in the Null (Non-Magnetic Cobalt) Model. The Parameters
Are Described in Tables 5.6 and 5.7
| Parameter |
Prior Value |
Posterior Value |
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
|
| Parameter |
Prior Value |
Posterior Value |
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
 |
 |
|
Table 5.5:
The Effect of the Experimental Evidence on Beliefs about
Parameters in the Main (Magnetic Cobalt) Model. The Parameters Are
Described in Tables 5.6 and 5.7
|
|
Next: The Details of the
Up: Further Discussion and Conclusions
Previous: Framework of the Models
  Contents
Daniel Christopher Hatton
2004-11-30