scml.scml2020.components.prediction

Classes

TradePredictionStrategy	A prediction strategy for expected inputs and outputs at every step
ExecutionRatePredictionStrategy	A prediction strategy for expected inputs and outputs at every step
FixedTradePredictionStrategy	Predicts a fixed amount of trade both for the input and output products.
MarketAwareTradePredictionStrategy	Predicts an amount based on publicly available market information. Falls
FixedERPStrategy	Predicts that the there is a fixed execution rate that does not change for all partners
MeanERPStrategy	Predicts the mean execution fraction for each partner

Module Contents

class scml.scml2020.components.prediction.TradePredictionStrategy(*args, predicted_outputs: int | numpy.ndarray = None, predicted_inputs: int | numpy.ndarray = None, add_trade=False, **kwargs)

A prediction strategy for expected inputs and outputs at every step

Parameters:

• predicted_inputs (-) – None for default, a number of an n_steps numbers giving predicted inputs

• predicted_outputs (-) – None for default, a number of an n_steps numbers giving predicted outputs

Provides:

• expected_inputs : n_steps vector giving the predicted inputs at every time-step. It defaults to the number of lines.

• expected_outputs : n_steps vector giving the predicted outputs at every time-step. It defaults to the number of lines.

• input_cost : n_steps vector giving the predicted input cost at every time-step. It defaults to catalog price.

• output_price : n_steps vector giving the predicted output price at every time-step. It defaults to catalog price.

Hooks Into:

• init

• before_step

• step

Abstract:

• trade_prediction_init: Called during init() to initialize the trade prediction.

• trade_prediction_step: Called during step() to update the trade prediction.

Remarks:

• Attributes section describes the attributes that can be used to construct the component (passed to its __init__ method).

• Provides section describes the attributes (methods, properties, data-members) made available by this component directly. Note that everything provided by the bases of this components are also available to the agent (Check the Bases section above for all the bases of this component).

• Requires section describes any requirements from the agent using this component. It defines a set of methods or properties/data-members that must exist in the agent that uses this component. These requirement are usually implemented as abstract methods in the component

• Abstract Objects Layer section describes abstract methods that MUST be implemented by any descendant of this component.

• Hooks Into section describes the methods this component overrides calling super () which allows other components to hook into the same method (by overriding it). Usually callbacks starting with on_ are hooked into this way.

• Overrides section describes the methods this component overrides without calling super effectively disallowing any other components after it in the MRO to call this method. Usually methods that do some action (i.e. not starting with on_) are overridden this way.

expected_outputs = None

Expected output quantity every step

expected_inputs = None

Expected input quantity every step

input_cost: numpy.ndarray = None

Expected unit price of the input

output_price: numpy.ndarray = None

Expected unit price of the output

_add_trade = False

abstract trade_prediction_init() → None

Will be called to update expected_outputs, expected_inputs, input_cost, output_cost during init()

trade_prediction_before_step() → None

Will be called at the beginning of every step to update the prediction

trade_prediction_step() → None

Will be called at the end of every step to update the prediction

init()

before_step()

step()

class scml.scml2020.components.prediction.ExecutionRatePredictionStrategy

A prediction strategy for expected inputs and outputs at every step

Provides:

• predict_quantity : A method for predicting the quantity that will actually be executed from a contract

Abstract:

• predict_quantity : A method for predicting the quantity that will actually be executed from a contract

Hooks Into:

• internal_state

Remarks:

• Attributes section describes the attributes that can be used to construct the component (passed to its __init__ method).

• Provides section describes the attributes (methods, properties, data-members) made available by this component directly. Note that everything provided by the bases of this components are also available to the agent (Check the Bases section above for all the bases of this component).

• Requires section describes any requirements from the agent using this component. It defines a set of methods or properties/data-members that must exist in the agent that uses this component. These requirement are usually implemented as abstract methods in the component

• Abstract Objects Layer section describes abstract methods that MUST be implemented by any descendant of this component.

• Hooks Into section describes the methods this component overrides calling super () which allows other components to hook into the same method (by overriding it). Usually callbacks starting with on_ are hooked into this way.

• Overrides section describes the methods this component overrides without calling super effectively disallowing any other components after it in the MRO to call this method. Usually methods that do some action (i.e. not starting with on_) are overridden this way.

abstract predict_quantity(contract: negmas.Contract)

class scml.scml2020.components.prediction.FixedTradePredictionStrategy(*args, add_trade=True, **kwargs)

Bases: TradePredictionStrategy

Predicts a fixed amount of trade both for the input and output products.

Hooks Into:

• internal_state

• on_contracts_finalized

Remarks:

• Attributes section describes the attributes that can be used to construct the component (passed to its __init__ method).

• Provides section describes the attributes (methods, properties, data-members) made available by this component directly. Note that everything provided by the bases of this components are also available to the agent (Check the Bases section above for all the bases of this component).

• Requires section describes any requirements from the agent using this component. It defines a set of methods or properties/data-members that must exist in the agent that uses this component. These requirement are usually implemented as abstract methods in the component

• Abstract Objects Layer section describes abstract methods that MUST be implemented by any descendant of this component.

• Hooks Into section describes the methods this component overrides calling super () which allows other components to hook into the same method (by overriding it). Usually callbacks starting with on_ are hooked into this way.

• Overrides section describes the methods this component overrides without calling super effectively disallowing any other components after it in the MRO to call this method. Usually methods that do some action (i.e. not starting with on_) are overridden this way.

_add_trade = True

trade_prediction_init()

Will be called to update expected_outputs, expected_inputs, input_cost, output_cost during init()

property internal_state

on_contracts_finalized(signed: List[negmas.Contract], cancelled: List[negmas.Contract], rejectors: List[List[str]]) → None

class scml.scml2020.components.prediction.MarketAwareTradePredictionStrategy(*args, predicted_outputs: int | numpy.ndarray = None, predicted_inputs: int | numpy.ndarray = None, add_trade=False, **kwargs)

Bases: TradePredictionStrategy

Predicts an amount based on publicly available market information. Falls back to fixed prediction if no information is available

Hooks Into:

• internal_state

• on_contracts_finalized

Remarks:

• Attributes section describes the attributes that can be used to construct the component (passed to its __init__ method).

• Provides section describes the attributes (methods, properties, data-members) made available by this component directly. Note that everything provided by the bases of this components are also available to the agent (Check the Bases section above for all the bases of this component).

• Requires section describes any requirements from the agent using this component. It defines a set of methods or properties/data-members that must exist in the agent that uses this component. These requirement are usually implemented as abstract methods in the component

• Abstract Objects Layer section describes abstract methods that MUST be implemented by any descendant of this component.

• Hooks Into section describes the methods this component overrides calling super () which allows other components to hook into the same method (by overriding it). Usually callbacks starting with on_ are hooked into this way.

• Overrides section describes the methods this component overrides without calling super effectively disallowing any other components after it in the MRO to call this method. Usually methods that do some action (i.e. not starting with on_) are overridden this way.

init()

trade_prediction_init()

Will be called to update expected_outputs, expected_inputs, input_cost, output_cost during init()

__update()

trade_prediction_step()

Will be called at the end of every step to update the prediction

trade_prediction_before_step()

Will be called at the beginning of every step to update the prediction

property internal_state

class scml.scml2020.components.prediction.FixedERPStrategy(*args, execution_fraction=0.95, **kwargs)

Bases: ExecutionRatePredictionStrategy

Predicts that the there is a fixed execution rate that does not change for all partners

Parameters:

execution_fraction – The expected fraction of any contract’s quantity to be executed

Provides:

• predict_quantity : A method for predicting the quantity that will actually be executed from a contract

Hooks Into:

• internal_state

Remarks:

• Attributes section describes the attributes that can be used to construct the component (passed to its __init__ method).

• Provides section describes the attributes (methods, properties, data-members) made available by this component directly. Note that everything provided by the bases of this components are also available to the agent (Check the Bases section above for all the bases of this component).

• Requires section describes any requirements from the agent using this component. It defines a set of methods or properties/data-members that must exist in the agent that uses this component. These requirement are usually implemented as abstract methods in the component

• Abstract Objects Layer section describes abstract methods that MUST be implemented by any descendant of this component.

• Hooks Into section describes the methods this component overrides calling super () which allows other components to hook into the same method (by overriding it). Usually callbacks starting with on_ are hooked into this way.

• Overrides section describes the methods this component overrides without calling super effectively disallowing any other components after it in the MRO to call this method. Usually methods that do some action (i.e. not starting with on_) are overridden this way.

_execution_fraction = 0.95

predict_quantity(contract: negmas.Contract)

class scml.scml2020.components.prediction.MeanERPStrategy(*args, execution_fraction=0.95, **kwargs)

Bases: ExecutionRatePredictionStrategy

Predicts the mean execution fraction for each partner

Parameters:

execution_fraction – The expected fraction of any contract’s quantity to be executed

Provides:

• predict_quantity : A method for predicting the quantity that will actually be executed from a contract

Hooks Into:

• internal_state

• init

• on_contract_executed

• on_contract_breached

Remarks:

• Attributes section describes the attributes that can be used to construct the component (passed to its __init__ method).

• Provides section describes the attributes (methods, properties, data-members) made available by this component directly. Note that everything provided by the bases of this components are also available to the agent (Check the Bases section above for all the bases of this component).

• Requires section describes any requirements from the agent using this component. It defines a set of methods or properties/data-members that must exist in the agent that uses this component. These requirement are usually implemented as abstract methods in the component

• Abstract Objects Layer section describes abstract methods that MUST be implemented by any descendant of this component.

• Hooks Into section describes the methods this component overrides calling super () which allows other components to hook into the same method (by overriding it). Usually callbacks starting with on_ are hooked into this way.

• Overrides section describes the methods this component overrides without calling super effectively disallowing any other components after it in the MRO to call this method. Usually methods that do some action (i.e. not starting with on_) are overridden this way.

_execution_fraction = 0.95

_total_quantity = None

predict_quantity(contract: negmas.Contract)

init()

property internal_state

on_contract_executed(contract: negmas.Contract) → None

on_contract_breached(contract: negmas.Contract, breaches: List[negmas.Breach], resolution: negmas.Contract | None) → None