blackboxopt.optimization_loops.utils

evaluation_function_wrapper(evaluation_function, evaluation_specification, objectives, logger)

Wrapper for evaluation functions. The evaluation result returned by the evaluation function is checked to contain all relevant objectives. An empty evaluation with a stacktrace is reported to the optiizer in case an unhandled Exception occurrs during the evaluation function call.

Source code in blackboxopt/optimization_loops/utils.py

def evaluation_function_wrapper(
    evaluation_function: Callable[[EvaluationSpecification], Evaluation],
    evaluation_specification: EvaluationSpecification,
    objectives: List[Objective],
    logger: logging.Logger,
) -> Evaluation:
    """Wrapper for evaluation functions. The evaluation result returned by the
    evaluation function is checked to contain all relevant objectives. An empty
    evaluation with a stacktrace is reported to the optiizer in case an unhandled
    Exception occurrs during the evaluation function call.
    """
    try:
        evaluation = evaluation_function(evaluation_specification)
    except Exception:
        stacktrace = traceback.format_exc()

        logger.warning("Report FAILURE due to unhandled error during evaluation")
        logger.debug(stacktrace)

        evaluation = evaluation_specification.get_evaluation(
            stacktrace=stacktrace, objectives={o.name: None for o in objectives}
        )

    raise_on_unknown_or_incomplete_objectives(
        known_objectives=objectives, reported_objectives=evaluation.objectives
    )

    return evaluation

init_max_evaluations_with_limit_logging(timeout_s, logger, max_evaluations=None)

[summary]

Parameters:

Name	Type	Description	Default
timeout_s	float	[description]	required
logger	Logger	[description]	required
max_evaluations	int	[description]	None

Returns:

Type	Description
float	[description]

Source code in blackboxopt/optimization_loops/utils.py

def init_max_evaluations_with_limit_logging(
    timeout_s: float, logger: logging.Logger, max_evaluations: int = None
) -> float:
    """[summary]

    Args:
        timeout_s: [description]
        logger: [description]
        max_evaluations: [description]

    Returns:
        [description]
    """
    if max_evaluations:
        logger.info(
            f"Starting optimization run until complete or {max_evaluations} evaluations"
        )
        return float(max_evaluations)

    if timeout_s == float("inf"):
        logger.info("Starting optimization run until complete")
    else:
        timeout_pretty = datetime.timedelta(seconds=timeout_s)
        logger.info(
            f"Starting optimization run until complete or {timeout_pretty} passed"
        )

    return float("inf")