Generate a sample from a probability distribution with the covariance-matching slice sampling method.
cov.match.sample(target.dist, x0, sample.size, tuning=1,
theta=1, limit=length(x0)*100)A list with elements X, evals, grads, and
adapt.rate. adapt.rate indicates the fraction of
crumb draws that resulted in adaptation. This sampler follows
the calling convention of compare.samplers.
Target distribution; see make.dist.
Initial coordinates.
Sample size to draw.
A tuning parameter; corresponds to \(\sigma_c\) in sec. 4 of Thompson and Neal (2010).
A factor to scale the crumb standard deviation in
every direction after a proposal is rejected. So, after \(k\)
proposals, crumbs have standard deviation theta^k*tuning
in directions orthogonal to all the proposal gradients.
A limit on the number of log-density evaluations per observation before sampling is aborted.
This function implements the covariance-matching method of slice
sampling, as described by Thompson and Neal (2010). It can be passed
to compare.samplers in the samplers list
argument.
Thompson, M. B. and Neal, R. M. (2010). Covariance-adaptive slice sampling. Technical Report TR-1002, Dept. of Statistics, University of Toronto.
compare.samplers,
shrinking.rank.sample