1Blurb::Defines the technique used to generate the MCMC proposal covariance.
2
3Description::
4The proposal covariance is used to define a multivariate normal (MVN)
5jumping distribution used to create new points within a Markov chain.
6That is, a new point in the chain is determined by sampling within a
7MVN probability density with prescribed covariance that is centered at
8the current chain point.  The accuracy of the proposal covariance has
9a significant effect on rejection rates and the efficiency of chain mixing.
10
11<b> Default Behavior </b>
12
13The default proposal covariance is \c prior when no emulator is
14present; \c derivatives when an emulator is present.
15
16<b> Expected Output </b>
17
18The effect of the proposal covariance is reflected in the MCMC chain
19values and the rejection rates, which can be seen in the diagnostic
20outputs from the QUESO solver within the \c QuesoDiagnostics
21directory.
22
23<b> Usage Tips </b>
24
25When derivative information is available inexpensively (e.g., from an
26emulator model), the derived-based proposal covariance forms a more
27accurate proposal distribution, resulting in lower rejection rates and
28faster chain mixing.
29
30Topics::	bayesian_calibration
31
32Examples::
33
34Theory::
35Faq::
36See_Also::
37