ad.propSd: Adaptation of proposal standard deviation
Description
Adaptive MCMC: if acceptance rate of the chain is smaller than lower
or larger than upper
,
the proposal standard deviation propSd
=exp(l) is adapted with respect to function delta.n
,
that means, the new proposal standard deviation
is equal to exp(l-delta.n(batch)
), respectively exp(l+delta.n(batch)
).
Usage
ad.propSd(chain, propSd, batch, lower = 0.3, upper = 0.6, delta.n = function(n) min(0.05, 1/sqrt(n)))
Arguments
propSd
current proposal standard deviation
batch
number of batch (of chain)
delta.n
function of batch number
Value
adapted proposal standard deviation
References
Rosenthal, J. S. (2011). Optimal Proposal Distributions and Adaptive MCMC. In: Handbook of Markov Chain Monte Carlo, pp. 93-112.