Sampling Functions¶
Listed below are the sampling methods for which functions are provided to simulating draws from distributions that can be specified up to constants of proportionalities. Model-based Sampler constructors are available for use with the mcmc()
engine as well as stand-alone functions that can be used independently.
- Approximate Bayesian Computation (ABC)
- Adaptive Mixture Metropolis (AMM)
- Adaptive Metropolis within Gibbs (AMWG)
- Binary Hamiltonian Monte Carlo (BHMC)
- Binary MCMC Model Composition (BMC3)
- Binary Metropolised Gibbs (BMG)
- Discrete Gibbs Sampler (DGS)
- Hamiltonian Monte Carlo (HMC)
- Metropolis-Adjusted Langevin Algorithm (MALA)
- Missing Values Sampler (MISS)
- No-U-Turn Sampler (NUTS)
- Random Walk Metropolis (RWM)
- Shrinkage Slice (Slice)
- Slice Simplex (SliceSimplex)
The following table summarizes the (d-dimensional) sample spaces over which each method simulates draws, whether draws are generated univariately or multivariately, and whether transformations are applied to map parameters to the sample spaces.
Model-Based Constructors | Stand-Alone Functions | |||||
---|---|---|---|---|---|---|
Method | Sample Space | Univariate | Multivariate | Transformations | Univariate | Multivariate |
ABC | ![]() |
No | Yes | Yes | No | No |
AMM | ![]() |
No | Yes | Yes | No | Yes |
AMWG | ![]() |
Yes | No | Yes | Yes | No |
BHMC | ![]() |
No | Yes | No | No | Yes |
BMC3 | ![]() |
Yes | Yes | No | Yes | Yes |
BMG | ![]() |
Yes | Yes | No | Yes | Yes |
DGS | Finite ![]() |
Yes | No | No | No | Yes |
HMC | ![]() |
No | Yes | Yes | No | Yes |
MALA | ![]() |
No | Yes | Yes | No | Yes |
MISS | Parameter-defined | Yes | Yes | No | No | No |
NUTS | ![]() |
No | Yes | Yes | No | Yes |
RWM | ![]() |
No | Yes | Yes | No | Yes |
Slice | ![]() |
Yes | Yes | Optional | Yes | Yes |
SliceSimplex | d-simplex | No | Yes | No | No | Yes |