Sample_KF_post: Sample the posterior distribution of the process using the...

In FastGaSP: Fast and Exact Computation of Gaussian Stochastic Process

Sample the posterior distribution of the process using the backward smoothing algorithm

Description

This function samples the posterior distribution of the process using the backward smoothing algorithm.

Usage

Sample_KF_post(index_obs, C_R_K_Q,W0,GG,W,VV,output,kernel_type,sample_type)

Arguments

index_obs	a vector where the entries with 1 have observations and entries with 0 have no observation.
C_R_K_Q	a list of matrices to compute the inverse covariance matrix in the dynamic linear model.
GG	a list of matrices defined in the dynamic linear model.
W	a list of coefficient matrices defined in the dynamic linear model.
VV	a numerical value of the variance of the nugget parameter.
output	a vector of the output.
kernel_type	a character to specify the type of kernel to use. The current version supports kernel_type to be "matern_5_2" or "exp", meaning that the matern kernel with roughness parameter being 2.5 or 0.5 (exponent kernel), respectively.
sample_type	a integer to specify the type of sample we need. 0 means the states. 1 means the first value of each state vector. 2 means the noisy observations.

Value

A matrix of the posterior samples.

Author(s)

Mengyang Gu [aut, cre]

Maintainer: Mengyang Gu <[e-mail]>

References

Hartikainen, J. and Sarkka, S. (2010). Kalman filtering and smoothing solutions to temporal gaussian process regression models. Machine Learning for Signal Processing (MLSP), 2010 IEEE International Workshop, 379-384.

M. Gu, Y. Xu (2019), fast nonseparable gaussian stochastic process with application to methylation level interpolation. Journal of Computational and Graphical Statistics, In Press, arXiv:1711.11501.

Campagnoli P, Petris G, Petrone S. (2009), Dynamic linear model with R. Springer-Verlag New York.

See Also

Sample_KF_post for more details about sampling from the posterior distribution.

FastGaSP documentation built on May 29, 2024, 1:30 a.m.