fitbenchmarking.utils.checkpoint module

This file implements a checkpointclass to save results as they are generated. This is also used to read in checkpointed data.

class fitbenchmarking.utils.checkpoint.Checkpoint(options: Options)

Bases: object

A class to build a checkpoint file. Each run can be added to the file as they finish. This class must be finalised to create the checkpoint.

add_result(result: FittingResult)

Add the result to the checkpoint file.

Parameters:

result (FittingResult) – The result to add.

finalise()

Add the trailing bracket to validate the json.

finalise_group(label='benchmark', failed_problems=None, unselected_minimizers=None)

Combine the problem and results file into the main checkpoint file.

load()

Load fitting results from a checkpoint file along with failed problems and unselected minimizers.

Returns:

Instantiated fitting results, unselected minimisers, failed problems

Return type:

Tuple[ dict[str, list[FittingResult]], dict, dict[str, list[str]]]