Prior Distributions in stochvol

The functions below can be supplied to specify_priors to overwrite the default set of prior distributions in svsample. The functions have mean, range, density, and print methods.

sv_constant(value)

sv_normal(mean = 0, sd = 1)

sv_multinormal(mean = 0, precision = NULL, sd = 1, dim = NA)

sv_gamma(shape, rate)

sv_inverse_gamma(shape, scale)

sv_beta(shape1, shape2)

sv_exponential(rate)

sv_infinity()

Arguments

value

The constant value for the degenerate constant distribution

mean

Expected value for the univariate normal distribution or mean vector of the multivariate normal distribution

sd

Standard deviation for the univariate normal distribution or constant scale of the multivariate normal distribution

precision

Precision matrix for the multivariate normal distribution

dim

(optional) Dimension of the multivariate distribution

shape

Shape parameter for the distribution

rate

Rate parameter for the distribution

scale

Scale parameter for the distribution

shape1

First shape parameter for the distribution

shape2

Second shape parameter for the distribution

Multivariate Normal

Multivariate normal objects can be specified several ways. The most general way is by calling sv_multinormal(mean, precision), which allows for arbitrary mean and (valid) precision arguments. Constant mean vectors and constant diagonal precision matrices of dimension D can be created two ways: either sv_multinormal(mean, sd, dim = D) or rep(sv_normal(mean, sd), length.out = D).

See also

Other priors: specify_priors()