| Spatially-varying SAR models and Bayesian inference for high-resolution lattice data
Chiranjit Mukherjee, Prasad S. Kasibhatla & Mike West
- Publication: Annals of the Institute of Statistical Mathematics,
- Original version of paper.
- Two short videos that show example sequences of MCMC sampled φ-fields from the analysis of the first synthetic data set discussed in Section 5 of the paper.
Viewing MCMC samples this way can provide
useful subjective insights into the nature of the inferred spatial field, and aspects of the
uncertainty associated with the posterior for the field, as the MCMC progresses. Note that these videos show
MCMC sweeps across columns of the field only, rather than from the full and more
effective interlaced row-column MCMC algorithm used in the examples in the paper.
- The first video is over the first 30 iterations of the MCMC beginning at a constant field, giving an indication of
the rapid pick-up of relevant spatial patterns.
- The second video sequences through another 30 MCMC samples
beginning at MCMC iterate 3000, well after the chain has stabilized and appears to be sampling φ-fields
approximately from the posterior distribution.
- Data and Matlab code for the MCMC analysis of SVSAR models, as reported in the examples in the paper, together with a readme file. This allows interested readers to explore the models and reproduce all examples in the paper, as well as to develop the model and MCMC methods for other applications.
This software is made freely available to any interested user. The authors can provide
no support nor assistance with implementations beyond the example in the code, nor
extensions of the code for other purposes. The download has been tested to confirm all
details are operational as described.
It is understood by the user that neither the authors nor Duke University bear any responsibility
nor assume any liability for any end-use of this software. It is expected that appropriate
credit/acknowledgment be given should the software be included as an element in other software development
or in publications.