CRAN stochvol 3.2.5

• Bugfix in the fast sampler. In the centered parameterization,
  the proposal distribution for parameters phi and mu did not
  have the documented variance. Thanks to Luis Gruber.
• Bugfix 'print.svdraws' for the case when re-sampling was
  done. Underflow may have occurred.

CRAN stochvol 3.2.4

Full Changelog: v3.2.3...v3.2.4

CRAN stochvol 3.2.3

News since v3.2.1:

• Minor changes and fixes in the documentation
• Bugfix in the indexing of svsample_roll() for expanding window estimation. Thanks to Weixian Nie.
• Bugfix in the validation of the inverted gamma prior for sigma^2. Thanks to A. Slavik.

CRAN stochvol 3.2.1

stochvol 3.2.1

• Minor adaptions to NEWS.md to comply with NEWS extractor
  'tools:::.build_news_db_from_package_NEWS_md'
• Export further C++ functions 'cholesky_tridiagonal',
  'forward_algorithm', 'backward_algorithm',
  'inverse_transform_sampling', 'find_mixture_indicator_cdf'
• Bugfix in 'print.svsim'

CRAN stochvol 3.2.0 JSS

stochvol 3.2.0

• Appeared in the Journal of Statistical Software
• Include reference to the JSS paper
• Bugfix: initial value for 'mu' was incorrect when
  'mu' was set to constant a priori
• Typo corrections in the documentation

CRAN stochvol 3.1.0

stochvol 3.1.0

• NEW FUNCTION 'svlm', which has a formula interface;
  it is a wrapper around 'svsample'; many thanks
  to Peter Knaus for his help
• Turn on printing on Windows
• Implement Geweke test in C++; it is feasible to
  execute it as a CRAN test now
• Small change in the behavior of 'predict.svdraws':
  when 'newdata' is given then 'steps' is ignored;
  A warning is shown if this is relevant
• Updated examples for 'svsample' and 'predict' so
  that they use the extractors as intended
• Simplified vignette: cache some results to
  reduce dependencies
• Bugfix in adaptation that does not affect
  correctness

CRAN stochvol 3.0.6

stochvol 3.0.6

• Correct wrongly submitted vignettes

stochvol 3.0.5

• Re-added first vignette with the necessary updates
• Bugfix in 'residuals.svdraws' and 'plot.svresid'; thanks
  to David Zoltan Szabo

CRAN stochvol 3.0.4

Changes in Version 3.0.4:

• Updated vignette
• Prevent some compilation warnings
• Updated CITATION file

CRAN stochvol 3.0.3

Changes in Version 3.0.3:

• Bugfix in 'svsample' when the input data contains 0s

Changes in Version 3.0.2:

• Bugfix in the fast access functions 'svsample_fast_cpp' and
  'svsample_general_cpp': dimensions were incorrect
• Bring out the adaptation object for 'svsample_general_cpp' to
  the R level; used to be accessible from C++ only
• Update to the adaptation object in C++

CRAN stochvol 3.0.1

From the NEWS file:
Changes in Version 3.0.1:

• Bugfix for calls to 'svsample' with draws = 1
• Fix some #include directives
• Avoid compilation problems on Solaris

Changes in Version 3.0.0:

• New model: heavy-tailed SV with leverage and its sampling
  function 'svtlsample'
• Change in the heavy-tailed models: now they operate with a
  normalized Student's t-distribution, i.e.\ its standard
  deviation is 1 for all degrees of freedom
• New, optional API for specifying prior distributions
• It is possible to set any or all of 'mu', 'phi', 'sigma',
  'nu', 'rho' to a constant value, i.e.\ set a Dirac prior
• Replaced the prior distribution for the degrees of freedom
  parameter 'nu' with an exponential distribution; it used to be
  a uniform distribution
• New set of unit tests including a Geweke test
• Entirely refactored C++ backend
• New, rethought set of exported C++ functions: 'update_fast_sv',
  'update_general_sv', 'update_regression', and 'update_t_error';
  all of them are documented
• Vastly improved the computational efficiency of the former
  'svlsample' code
• New feature: run independent MCMC chains on the same data set
• Integration of the 'parallel' package: option to run independent
  MCMC chains on a 'SNOW' cluster or using the 'multicore' strategy
• Modified backend for 'svdraws' and 'svpredict' objects to
  to incorporate independent chains: they contain 'mcmc.list'
  objects instead of plain 'mcmc' objects.
  WARNING! This may break code that exploited the backed!
• Additional extractor functions for 'svdraws' objects: 'vola',
  'sv_beta', 'sv_tau', 'sampled_parameters', index chains via [],
  'as.array'
• New extractors for 'svpredict' objects: 'predy', 'predlatent',
  'predvola', index chains via []
• CITATION file updated
• New vignette
• SV with leverage also includes 'latent0' from now on
• New feature: rolling window estimation via the functions
  'svsample_roll', 'svtsample_roll', 'svlsample_roll', and
  'svtlsample_roll'
• Unified plotting between models: removed the 'scaling' plot
  from heavy-tailed model outputs
• Many new, small examples
• Small bugfixes
• New "fast-access" functions to circumvent input validation:
  'svsample_fast_cpp' and 'svsample_general_cpp'; their arguments
  are slightly different from the 'svsample' family of functions,
  they come with documentation
• 'svsample2' is set to deprecated