Mll
Related Doc:
package smfsb
A simple bootstrap particle filter with multinomial resampling.
A simple bootstrap particle filter with multinomial resampling. Note that it
supports time series of observations on irregular time grids. Note that this is
the function which does most of the work for pfMll.
A sample from the prior on the state at time t0. The size of the sample
determines the number of particles used in the particle filter.
The time corresponding to the prior sample
The observed data in the form of a time series
The log-likelihood of the observation model
The transition kernel of the underlying process
A triple: the first element is the log of an unbiased estimate of the marginal likelihood of the model, the second is just the final time point, and the third is a sample from the posterior distribution of the final time point.
A function for evaluating marginal log-likelihoods of parametrised models.
A function for evaluating marginal log-likelihoods of parametrised models. It basically just threads a parameter value through the arguments required for bpf.
A function which when provided with a parameter value will return a sample from the prior on the intial state of a Markov process model. The size of the sample will determine the number of particles used in the particle filter.
The intial time corresponding to the prior sample
A function, which when provided with a parameter value, will return the transition kernel of a Markov process.
A function, which when provided with a parameter value, will return the log-likelihood of the observation model
The observed data in the form of a time series
Functions associated with particle filtering of Markov process models against time series data and the computation of marginal model likelihoods.