fastoad.cmd.calc_runner module

Tools for running multiple computations

class fastoad.cmd.calc_runner.CalcRunner(configuration_file_path: str | PathLike, input_file_path: str | PathLike | None = None, optimize: bool = False)[source]

Bases: object

Class for running FAST-OAD computations for a specific configuration.

It is specifically designed to run several computations concurrently with run_cases(). For each computation, data can be isolated in a specific folder.

configuration_file_path: str | PathLike

Configuration file, common to all computations

input_file_path: str | PathLike | None = None
optimize: bool = False

For activating MDO instead MDA

run(input_values: VariableList | None = None, calculation_folder: str | PathLike | None = None) → DataFile[source]

Run the computation.

This method is useful to set input values on-the-fly, and/or isolate the computation data in a dedicated folder.

Parameters:

  • input_values – if provided, these values will supersede the content of input file (specified in configuration file)

  • calculation_folder – if specified, all data, including configuration file, will be stored in that folder. The input file in this folder will contain data from input_values

Returns:

the written output data

run_cases(input_list: list[VariableList], destination_folder: str | PathLike, *, max_workers: int | None = None, use_MPI_if_available: bool = True, overwrite_subfolders: bool = False) → list[DataFile | None][source]

Run computations concurrently.

The data of each computation will be isolated in a dedicated subfolder of destination folder.

Parameters:

  • input_list – a computation will be run for each item of this list

  • destination_folder – The data of each computation will be isolated in a dedicated subfolder of this folder.

  • max_workers – if not specified, all available processors will be used. Set to -1 to use all available processors except 1 (useful when running locally while keeping some CPU for working meanwhile).

  • use_MPI_if_available – If False, or if no MPI implementation is available, computations will be run concurrently using the multiprocessing library.

  • overwrite_subfolders – if False, calculations that match existing subfolders won’t be run (allows batch continuation)

Returns:

a list of output data files, with one entry per input case in input_list. Entries are DataFile instances for successful computations, or None for failed computations. Failed computations are caught and logged as warnings rather than raising exceptions, allowing the batch to continue.