NEWS.md

In shrinkTVP: Efficient Bayesian Inference for Time-Varying Parameter Models with Shrinkage

Changes in Version 2.1.0

• Added a more robust solve method for armadillo matrices
• Fixed issues with exported Bhattacharya and Rue algorithm
• Fixed issue with the usenames option for xcolor in the vignette

Changes in Version 2.0.6

• Fixed a bug in LPDS()
• Fixed issues on Fedora and Debian
• Changed compiler flag -DARMA_DONT_PRINT_ERRORS to -DARMA_WARN_LEVEL=0

Changes in Version 2.0.5

• Modified shrinkTVP.ltx to work with new jss.cls

Changes in Version 2.0.4

• Added CITATION file
• Modified DESCRIPTION as well as some documentation files

Changes in Version 2.0.3

• Fixed a bug in the sampling step for c_xi (Thanks to Tony Chernis for pointing it out)
• Fixed a bug in the sampling step for a_tau (Thanks to Wenjie Zhao for pointing it out)
• Added DG_MH_step to exported functions
• Fixed some typos in the documentation
• Added a newline to the print() function

Changes in Version 2.0.2

• Moved to stochvol 3.0
• Fixed an issue with the exported version sample_TG_TVP function
• Fixed an issue with the exported Bhattacharya algorithm
• define STRICT_R_HEADERS, include float.h, adjust some constants

Changes in Version 2.0.1

• Fixed an issue on Solaris

Changes in Version 2.0

• Added option to use triple gamma prior
• Added one-step update function updateTVP for modular use in other samplers
• Added convenience methods (residuals.shrinkTVP, fitted.shrinkTVP, etc...)
• Re-factored C++ code, improving on speed in some places
• Added forecast_shrinkTVP (and associated plot.forecast_shrinkTVP method), which allows for forecasting
• All C++ functions are now exported for use in other packages
• simTVP now allows user to adjust parameters governing SV process
• Re-factored plot.shrinkTVP to use layout() instead of par(mfrow = ...)
• Bug fixes
• Updated the vignette

Changes in Version 1.1.1

• Updated the vignette.

Changes in Version 1.1.0

• Overhauled plotting functions. The default is now shaded areas instead of lines and more customisation options are available.
• plot.mcmc.tvp now also adds time series information to the x-axis labels, thanks to zoo::index.
• Changed the way LPDS are calculated, moving them from the shrinkTVP function to a dedicated LPDS function.
• Added the new eval_pred_dens function, which allows the user to evaluate the one-step ahead predctive density.
• Objects returned from the shrinkTVP function are now of class shrinkTVP and not shrinkTVP_res.

Changes in Version 1.0.2:

• Added a vignette
• Bug fixes

Changes in Version 1.0.1:

• plot.mcmc.tvp now adds a horizontal line at 0
• Added a nicer print method for shrinkTVP_res objects
• Fixed some typos and errors in the documentation

shrinkTVP documentation built on Nov. 23, 2023, 1:12 a.m.