Skip to content

Interfaces

This page describes the interfaces provided to customize the LNS search. All interfaces can be found in the fastlane.interfaces module. Example implementations of these interfaces can be found in the lib submodule.

Adaptive Strategy Interface

This interface gives guidelines on how to implement adaptive strategies for adaptive LNS.

AdaptiveStrategy

Bases: ABC

Adaptive strategy interface to select declarative LNS configuration.

get_initial_config(config_catalog: ConfigCatalog, initial_model: Model) -> ActiveConfig abstractmethod

Abstract method to get initial LNS configuration.

Parameters:

Name Type Description Default
config_catalog ConfigCatalog

Full declarative LNS catalog.

 required
initial_model Model

Initial model.

 required

Returns:

Type Description
ActiveConfig

LNS configuration

update_config(active_config: ActiveConfig, config_catalog: ConfigCatalog, stats: list[dict[str, Any]], lns_object: LNS) -> ActiveConfig abstractmethod

Abstract method to update LNS configuration.

Parameters:

Name Type Description Default
active_config ActiveConfig

Active LNS configuration.

 required
config_catalog ConfigCatalog

Full LNS configuration catalog.

 required
stats list[dict[str, Any]]

Statistics.

 required
lns_object LNS

LNS object.

 required

Returns:

Type Description
ActiveConfig

New LNS configuration.

Auto Destruction Converter

All custom auto destruction converters (for determining destruction rates automatically) should respect the this interface.

AutoDestructionConverter

Bases: ABC

Converter interface for computing destruction percentages of auto-mode destroy operators.

compute_auto_destruction_percent(config_name: str, project_operators: list[ProjectOperator], destroy_operator_name: str) -> float abstractmethod

Compute destruction percentage of auto-mode destroy operator.

Parameters:

Name Type Description Default
config_name str

Config name.

 required
project_operators list[ProjectOperator]

Project operators.

 required
destroy_operator_name str

Destroy operator name.

 required

Returns:

Type Description
float

Destruction percentage of auto-mode destroy operator.

convert_auto_in_config(config: ActiveConfig, lns_object: Optional[LNS] = None) -> ActiveConfig

Convert automatic values in LNS configuration into concrete percentages.

Parameters:

Name Type Description Default
config ActiveConfig

Active LNS configuration containing automatic values.

 required
lns_object Optional[LNS]

LNS object.

 None

Returns:

Type Description
ActiveConfig

LNS configuration with all automatic values replaced by concrete percentages.

Solver Interface

Solver implementations should follow the Solver Interface and make use of the SolverConfig class.

Solver

Bases: ABC

Solver interface.

__init__() -> None

Initialization of the solver object.

Before solving, setup() method must be called to correctly initialize the solver.

add(name: str, parameters: list[str], program: str) -> None

Add a program to the solver.

Parameters:

Name Type Description Default
name str

Name of the program.

 required
parameters list[str]

Parameters for the program.

 required
program str

Program to be added.

 required

assign_external(external: Union[Symbol, int], truth: bool) -> None

Assign truth value to external atom.

Parameters:

Name Type Description Default
external Union[Symbol, int]

External atom.

 required
truth bool

Truth value.

 required

get_name() -> str abstractmethod classmethod

Get the name under which the solver will be listed in options.

Returns:

Type Description
str

Name of the solver.

get_stats() -> dict[str, Any]

Get statistics of the last solve call.

Returns:

Type Description
dict[str, Any]

Statistics dictionary.

ground(parts: list[tuple[str, list[Symbol]]] = [('base', [])], context: Any = None) -> None

Ground base encoding.

Parameters:

Name Type Description Default
parts list[tuple[str, list[Symbol]]]

Parts to ground.

 [('base', [])]
context Any

Context for grounding.

 None

release_external(external: Union[Symbol, int]) -> None

Release external atom.

Parameters:

Name Type Description Default
external Union[Symbol, int]

External atom.

 required

setup(lns_object: LNS, args: list[str] = [], files: Optional[list[str]] = None) -> None abstractmethod

Initialization of the solver.

Parameters:

Name Type Description Default
lns_object LNS

LNS object.

 required
args list[str]

clingo arguments.

 []
files Optional[list[str]]

ASP files to be loaded.

 None

solve(config: Optional[SolverConfig], assumptions: list[tuple[clingo.symbol.Symbol, bool]] = [], require_model: bool = False) -> Optional[Model] abstractmethod

Solve with fixed atoms.

Parameters:

Name Type Description Default
config Optional[SolverConfig]

Solver configuration.

 required
assumptions list[tuple[Symbol, bool]]

Assumptions for solving (fixed atoms).

 []
require_model bool

If True, ignore cutoff time until a model is found.

 False

Returns:

Type Description
Optional[Model]

Last obtained model.