Create a new bayesim_simulation_result object — new_simulation

Constructs a result object from a complete simulation run.

Usage

new_simulation_result(
  config_fingerprint = character(1),
  task_results = list(),
  task_grid = NULL,
  summary = NULL,
  timing = list(total = 0),
  errors = NULL,
  checkpoint_path = NULL
)

Arguments

config_fingerprint: Character hash uniquely identifying the configuration
task_results: List of bayesim_task_result objects
task_grid: Tibble with task grid information (task_id, data_idx, fit_idx, rep_idx, status)
summary: Tibble with one row per task, columns for metrics and diagnostics
timing: List containing total, by_phase, and other timing breakdowns
errors: Tibble of failed tasks with error details
checkpoint_path: Path to the final checkpoint file, or NULL

Value

A validated bayesim_simulation_result object

Details

The bayesim_simulation_result class encapsulates the complete results of a simulation study, including all individual task results, an aggregated summary, timing breakdowns, error information, and checkpoint location.

Validation rules:

config_fingerprint must be a scalar character
task_results must be a list where all elements are bayesim_task_result
summary and errors must be data frames
timing$total must be non-negative
checkpoint_path must be NULL or a scalar character

Examples

# Create a simulation result
task1 <- new_task_result(
  task_id = "task_001",
  status = "success",
  metrics = list(rmse = 0.05),
  timing = list(total = 5.0)
)

result <- new_simulation_result(
  config_fingerprint = "abc123",
  task_results = list(task1),
  task_grid = tibble::tibble(
    task_id = "task_001",
    data_idx = 1L,
    fit_idx = 1L,
    rep_idx = 1L,
    status = "success"
  ),
  summary = tibble::tibble(task_id = "task_001", rmse = 0.05),
  timing = list(total = 10.0, by_phase = list(setup = 2.0, fit = 8.0)),
  errors = tibble::tibble(task_id = character(), error_message = character()),
  checkpoint_path = "/path/to/checkpoint.rds"
)