Summarises acceptance and admissibility diagnostics for bsvarSIGNs

Summarises the stored effective sample size and admissibility weights recomputed from posterior draws and identification restrictions. These diagnostics do not reconstruct the full proposal/rejection history of the sampler. They report quantities obtained from the stored posterior distribution and its identification restrictions. The upstream stored ESS is an importance-weight ESS, computed before weighted resampling, not an autocorrelation-based MCMC convergence diagnostic. Recomputed kernel weights describe the stored draws and need not reproduce the original weight vector.

Usage

acceptance_diagnostics(object, ...)

# S3 method for class 'PosteriorBSVARSIGN'
acceptance_diagnostics(
  object,
  kernel_tol = 1e-12,
  ess_threshold = 20,
  sparse_threshold = 0.1,
  model = "model1",
  ...
)

# Default S3 method
acceptance_diagnostics(object, ...)

Arguments

object

A PosteriorBSVARSIGN object.

...

Reserved for future extensions.

kernel_tol

Numerical tolerance used to classify near-zero admissibility weights.

ess_threshold

Effective-sample-size threshold below which a warning flag is raised.

sparse_threshold

Share of near-zero admissibility weights above which a sparse-support warning flag is raised.

model

Label identifying the model specification.

Value

A bsvar_post_tbl with columns model, object_type, metric, value, flag, and message. Each row reports one diagnostic metric.

See also

posterior_provenance

Examples

# \donttest{
data(optimism, package = "bsvarSIGNs")
sign_irf <- matrix(c(1, rep(NA, 3)), 2, 2)
spec_s <- suppressMessages(
  bsvarSIGNs::specify_bsvarSIGN$new(optimism[, 1:2], p = 1,
                                     sign_irf = sign_irf)
)
post_s <- bsvars::estimate(spec_s, S = 5, show_progress = FALSE)

diag <- acceptance_diagnostics(post_s)
print(diag)
#> # A tibble: 13 × 16
#>    model  object_type            metric        value flag  message analysis_type
#>    <chr>  <chr>                  <chr>         <dbl> <lgl> <chr>   <chr>        
#>  1 model1 acceptance_diagnostics posterior_dr…     5 FALSE ""      acceptance_d…
#>  2 model1 acceptance_diagnostics effective_sa…     5 TRUE  "ESS b… acceptance_d…
#>  3 model1 acceptance_diagnostics max_tries       Inf FALSE ""      acceptance_d…
#>  4 model1 acceptance_diagnostics irf_sign_res…     1 FALSE ""      acceptance_d…
#>  5 model1 acceptance_diagnostics zero_restric…     0 FALSE ""      acceptance_d…
#>  6 model1 acceptance_diagnostics structural_s…     0 FALSE ""      acceptance_d…
#>  7 model1 acceptance_diagnostics narrative_re…     0 FALSE ""      acceptance_d…
#>  8 model1 acceptance_diagnostics kernel_mean       1 FALSE ""      acceptance_d…
#>  9 model1 acceptance_diagnostics kernel_median     1 FALSE ""      acceptance_d…
#> 10 model1 acceptance_diagnostics kernel_min        1 FALSE ""      acceptance_d…
#> 11 model1 acceptance_diagnostics kernel_max        1 FALSE ""      acceptance_d…
#> 12 model1 acceptance_diagnostics kernel_zero_…     0 FALSE ""      acceptance_d…
#> 13 model1 acceptance_diagnostics kernel_cv         0 FALSE ""      acceptance_d…
#> # ℹ 9 more variables: interval_probability <dbl>, interval_method <chr>,
#> #   scale_by <chr>, scale_var <chr>, scale_factor <dbl>, units <chr>,
#> #   source_package <chr>, source_version <chr>, n_draws <int>
# }