nn_option_pricing.config

Configuration objects for reproducible experiments.

The project uses frozen dataclasses instead of global constants so that experiments can be configured explicitly from scripts while keeping defaults in one place.

Functions

config_to_dict(config)	Convert experiment configuration objects into JSON-serializable values.
make_path_config([data_dir, output_dir])	Create a consistent set of experiment paths from base directories.

Classes

DatasetConfig([n_samples, seed, s0_min, ...])	Ranges and random seed used to generate synthetic option data.
ExperimentConfig([dataset, training, ...])	Top-level configuration grouping all experiment settings.
MonteCarloConfig([n_paths, ...])	Parameters controlling the Monte Carlo benchmark.
PathConfig([data_dir, output_dir, ...])	Filesystem locations for generated data, outputs, and artifacts.
TrainingConfig([seed, feature_set, ...])	Hyperparameters for neural network training and data splitting.