bnelearn.experiment.multi_unit_experiment module¶
In this file multi-unit experiments MultiUnitExperiment are defined and
their analytical BNEs (if known) are assigned. Also, the SplitAwardExperiment
is implemented as well, as it shares most its properties.
- class bnelearn.experiment.multi_unit_experiment.MultiUnitExperiment(config: ExperimentConfig)[source]¶
Bases:
_MultiUnitSetupEvalMixin,ExperimentExperiment class for the standard multi-unit auctions.
- action_size: int¶
- b_opt: torch.Tensor¶
- bne_env: AuctionEnvironment¶
- bne_utilities: Tensor¶
- env: Environment¶
- epoch: int¶
- input_length: int¶
- learners: Iterable[learners.Learner]¶
- models: Iterable[torch.nn.Module]¶
- n_models: int¶
- observation_size: int¶
- plot_xmax: float¶
- plot_xmin: float¶
- plot_ymax: float¶
- plot_ymin: float¶
- positive_output_point: Tensor¶
- sampler: ValuationObservationSampler¶
- v_opt: torch.Tensor¶
- valuation_size: int¶
- class bnelearn.experiment.multi_unit_experiment.SplitAwardExperiment(config: ExperimentConfig)[source]¶
Bases:
_MultiUnitSetupEvalMixin,ExperimentExperiment class of the first-price sealed bid split-award auction.
- action_size: int¶
- b_opt: torch.Tensor¶
- bne_env: AuctionEnvironment¶
- bne_utilities: Tensor¶
- env: Environment¶
- epoch: int¶
- input_length: int¶
- learners: Iterable[learners.Learner]¶
- models: Iterable[torch.nn.Module]¶
- n_models: int¶
- observation_size: int¶
- plot_xmax: float¶
- plot_xmin: float¶
- plot_ymax: float¶
- plot_ymin: float¶
- positive_output_point: Tensor¶
- pretrain_transform(player_position)[source]¶
Some experiments need specific pretraining transformations. In most cases, pretraining to the truthful bid (i.e. the identity function) is sufficient.
- Args:
- player_position (:int:) the player for which the transformation is
requested.
- Returns
(:callable:) pretraining transformation
- sampler: ValuationObservationSampler¶
- v_opt: torch.Tensor¶
- valuation_size: int¶