State for the Metropolis algorithm
Use particle marginal metropolis algorithm
Run the metropolis algorithm for a DLM, using the kalman filter to calculate the likelihood
Metropolis kernel without re-evaluating the likelihood from the previous time step and keeping track of the acceptance ratio
Update the diagonal values of a covariance matrix by adding a Gaussian perturbation and ensuring the resulting diagonal is symmetric
Update the diagonal values of a covariance matrix by adding a Gaussian perturbation and ensuring the resulting diagonal is symmetric
the standard deviation of the innovation distribution
a diagonal DenseMatrix[Double], representing a covariance matrix
a distribution over the diagonal matrices
Add a Random innovation to a numeric value using the Gaussian distribution
Add a Random innovation to a numeric value using the Gaussian distribution
the standard deviation of the innovation distribution
the starting value of the Double
a Rand[Double] representing a perturbation of the double a which can be drawn from
Add a Random innovation to a DenseVector[Double] using the Gaussian distribution
Add a Random innovation to a DenseVector[Double] using the Gaussian distribution
the standard deviation of the innovation distribution
the starting value of the parameter
a Rand[DenseVector[Double]] representing a perturbation of the double a which can be drawn from
Simulate from a multivariate normal distribution given the cholesky decomposition of the covariance matrix
A Single Step without acceptance ratio this requires re-evaluating the likelihood at each step
Propose a new value of the parameters on the log scale