norm.proposal function

Manage proposal functions tune variance for metropolis sampler

Generate new proposals for the x from the current. Generates all x at once.

norm.proposal(m, n, sigma)

mvnorm.proposal(m, n, Sigma)

bmvnorm.proposal(m, n, Sigma)

Arguments

• m: number of records
• n: number of parameters
• sigma: variance
• Sigma: variance

Details

norm.proposal - Independent Normal proposal - every component is independent, with variances of individual components determined by sigma. The recycling rule applies to sigma, so sigma may be a scalar, an m vector or a m by n matrix.

mvnorm.proposal - Multivariate Normal proposal - all components of all points are correlated. In this case Sigma is the joint covariance of the m*n components of the proposal points.

bmvnorm.proposal - Block Multivariate Normal proposal - components of points are correlated, but points are independent. Here Sigma is an array of m covariance matrices that determine the covariance of the m proposal points.

Returns

An list object with get, set and tune functions to manage the state of the proposals. - proposal: propose new set of parameters from last

• get: get variance values

• set: set variance values

• tune: tune the variance for proposal functions

Author(s)

Simon Wotherspoon