gemseo.algos.problem_function module#

A function to be attached to a problem.

class ProblemFunction(function, output_evaluation_sequence, jacobian_evaluation_sequence, with_normalized_inputs, database, counter, stop_if_nan, design_space, store_jacobian=True, differentiation_method=None, **differentiation_method_options)[source]#

Bases: MDOFunction, Serializable

A function to be attached to a problem.

Parameters:

  • function (MDOFunction) -- The original function.

  • output_evaluation_sequence (Iterable[Callable[[NumberArray], NumberArray]]) -- The execution sequence to compute an output value from an input value.

  • jacobian_evaluation_sequence (Iterable[Callable[[NumberArray], NumberArray]]) -- The execution sequence to compute a Jacobian from an input value.

  • with_normalized_inputs (bool) -- Whether the function expects normalized inputs.

  • database (Database | None) -- The database to store and retrieve the evaluations; if None, do not use database.

  • counter (EvaluationCounter) -- The counter of evaluations.

  • stop_if_nan (bool) -- Whether the evaluation stops when a function returns NaN.

  • design_space (DesignSpace) -- The design space on which to evaluate the function.

  • store_jacobian (bool) --

    Whether to store the Jacobian matrices in the database.

    By default it is set to True.

  • differentiation_method (ApproximationMode | None) -- The differentiation method to compute the Jacobian. If None, use the original derivatives.

  • **differentiation_method_options (Any) -- The options of the differentiation method.

static check_function_output_includes_nan(value, stop_if_nan=True, function_name='', xu_vect=None)[source]#

Check if an array contains a NaN value.

Parameters:

  • value (ndarray) -- The array to be checked.

  • stop_if_nan (bool) --

    Whether to stop if value contains a NaN.

    By default it is set to True.

  • function_name (str) --

    The name of the function. If empty, the arguments function_name and xu_vect are ignored.

    By default it is set to "".

  • xu_vect (ndarray | None) -- The point at which the function is evaluated. None if and only if function_name is empty.

Raises:

  • DesvarIsNan -- If the value is a function input containing a NaN.

  • FunctionIsNan -- If the value is a function output containing a NaN.

Return type:

None

evaluate(x_vect)[source]#

Evaluate the function and store the output value in last_eval.

When the output dimension dim is not defined, it is inferred on the first evaluation.

Parameters:

x_vect (NumberArray) -- The input value of the function.

Returns:

Either the raw output value or its real part when force_real is True.

Return type:

OutputType

enable_statistics: ClassVar[bool] = True#

Whether to count the number of function evaluations.

property func: Callable[[ndarray[Any, dtype[number[Any]]]], ndarray[Any, dtype[number[Any]]] | Complex]#

The wrapped function.

property n_calls: int#

The number of times the function has been evaluated.

This count is both multiprocess- and multithread-safe, thanks to the locking process used by MDOFunction.evaluate().

stop_if_nan: bool#

Whether to stop the evaluation when a value is NaN.