Skip to contents

Create a new bayesim_fit_result object

Source: R/results.R
new_fit_result.Rd

Constructs a result object from a Bayesian model fitting operation.

Usage

new_fit_result(
  success = TRUE,
  fit = NULL,
  draws = NULL,
  diagnostics = list(),
  timing = list(total = 0, warmup = 0, sample = 0),
  warnings = character(),
  error = NULL,
  data_bundle = NULL
)

Arguments

success

Logical scalar indicating if the fit succeeded

fit

The backend fit object (may be NULL if removed by retention policy)

draws

Matrix of posterior draws (S x P), with column names for parameters

diagnostics

Named list of diagnostic values (scalars or named numeric vectors)

timing

List containing total, warmup, and sample timing in seconds

warnings

Character vector of warning messages captured during fitting

error

A condition object if the fit failed, NULL otherwise

data_bundle

Optional list containing the training/test data used for fitting

Value

A validated bayesim_fit_result object

Details

The bayesim_fit_result class encapsulates all outputs from a Bayesian model fitting operation, including the posterior draws, diagnostics, timing information, and any warnings or errors encountered.

Validation rules:

  • If success is FALSE, error must be non-NULL

  • If success is TRUE, error must be NULL

  • timing$total must be non-negative

  • If draws is not NULL, it must be a matrix with column names

Examples

# Successful fit
draws <- matrix(rnorm(1000), ncol = 2, nrow = 500)
colnames(draws) <- c("alpha", "beta")
result <- new_fit_result(
  success = TRUE,
  draws = draws,
  diagnostics = list(rhat = c(alpha = 1.01, beta = 1.00)),
  timing = list(total = 10.5, warmup = 5.0, sample = 5.5)
)

# Failed fit
result <- new_fit_result(
  success = FALSE,
  error = simpleError("Convergence failed"),
  diagnostics = list(),
  timing = list(total = 2.0, warmup = 2.0, sample = 0)
)