dpmmDistance

For an MCMC draw from a DPMM D_x, let Z_x be the vector of Z-scores of the
observations based on that observation's current group, and let A_x be the
0,1 adjacency matrix where $$[A_x]_{ij} = 1$$ if observations i and j are in
the same group in draw D_x (so the diagonal is always 1s).
Then we define the DPMM distance between D_x and D_y as: $$
 d(D_x, D_y) = |Z_x - Z_y|^T(|A_x - A_y| + I)|Z_x - Z_y|.
$$

Trace plots and convergence diagnostics for Markov Chain Monte Carlo (MCMC) algorithms on highly multivariate or unordered spaces. Methods outlined in a forthcoming paper.

Luke Duttweiler

genMCMCDiag

Generalized Convergence Diagnostics for Difficult MCMC
Algorithms

dpmmDistance function

<dl><dt>x</dt>
<dd>List with elements 'Zscore' and 'Adj'</dd>
<dt>y</dt>
<dd>List with elements 'Zscore' and 'Adj', both of same dimensions as in x.</dd></dl>

Arguments

DPMM Distance — dpmmDistance

<dl>

<dt>x</dt>
<dd>List with elements 'Zscore' and 'Adj'</dd>


<dt>y</dt>
<dd>List with elements 'Zscore' and 'Adj', both of same dimensions as in x.</dd>

</dl>

dpmmDistance: DPMM Distance

Description

Usage

Value

Arguments