concepts.dm.pdsketch.crow.crow_planner_v2.CROWRecursiveSearcherV2#
- class CROWRecursiveSearcherV2[source]#
Bases:
object
Methods
apply_regression_rule_effect
(state, rule, ...)apply_regression_rule_prefix
(node, ...)Apply the regression rule for a prefix of the subgoal.
dfs
(node)The main entrance of the CROW planner.
main
()postprocess_plans
(node, all_possible_plans)solve_csp
(csp, max_csp_trials[, actions])Attributes
- __init__(executor, state, goal_expr, *, enable_reordering=False, max_search_depth=10, max_beam_size=20, is_goal_serializable=True, is_goal_refinement_compressible=True, enable_csp=True, max_csp_trials=10, max_global_csp_trials=100, max_csp_branching_factor=5, use_generator_manager=False, store_generator_manager_history=False, enable_simulation=False, simulator=None, enable_dirty_derived_predicates=False, enable_greedy_execution=False, allow_empty_plan_for_optimistic_goal=False, verbose=True)[source]#
Initialize the CROW planner.
- Parameters:
executor (PDSketchExecutor) – the executor used to execute the expressions.
state (State) – the current state.
goal_expr (str | ValueOutputExpression) – the goal expression to achieve.
enable_reordering (bool) – whether to enable reordering in regression rule applications.
max_search_depth (int) – the maximum search depth.
max_beam_size (int) – the maximum beam size when trying different refinements of the same goal.
is_goal_serializable (bool) – whether the goal has been serialized.
is_goal_refinement_compressible (bool) – whether the serialized goals are refinement-compressible (i.e., for each goal item, do we need to track all the skeletons?)
enable_csp (bool) – whether to enable CSP solving.
max_csp_trials (int) – the maximum number of CSP trials.
max_global_csp_trials (int) – the maximum number of CSP trials for the global CSP (i.e., the CSP associated with the root node of the search tree.)
max_csp_branching_factor (int) – the maximum branching factor for CSP.
use_generator_manager (bool) – whether to use the generator manager.
store_generator_manager_history (bool) – whether to store the history of the generator calls in the generator manager.
enable_simulation (bool) – whether to enable simulation in CSP solving.
simulator (PDSketchSimulatorInterface | None) – the simulator used for simulation.
enable_dirty_derived_predicates (bool) – whether to enable dirty derived predicates.
enable_greedy_execution (bool) – whether to enable greedy execution.
allow_empty_plan_for_optimistic_goal (bool) – whether to allow empty plan for optimistic goal.
verbose (bool) – whether to print verbose information.
- __new__(**kwargs)#
- apply_regression_rule_effect(state, rule, bounded_variables)[source]#
- Parameters:
rule (RegressionRule)
bounded_variables (None | Sequence[Variable] | Dict[str | Variable, str | int | slice | bool | float | Tensor | TensorValue | ObjectConstant | StateObjectReference] | Dict[str, Dict[str, StateObjectReference | slice | TensorValue]])
- apply_regression_rule_prefix(node, grounded_subgoals, placeholder_csp, prefix_length, bounded_variables)[source]#
Apply the regression rule for a prefix of the subgoal.
- Parameters:
node (CROWSearchNode)
placeholder_csp (ConstraintSatisfactionProblem)
prefix_length (int)
bounded_variables (None | Sequence[Variable] | Dict[str | Variable, str | int | slice | bool | float | Tensor | TensorValue | ObjectConstant | StateObjectReference] | Dict[str, Dict[str, StateObjectReference | slice | TensorValue]])
- Return type:
List[Tuple[List[ValueOutputExpression], List[bool], ConstraintSatisfactionProblem | None, Dict[int, Any], Dict[str, Any]]]
- dfs(node)[source]#
The main entrance of the CROW planner.
- Parameters:
node (CROWSearchNode)
- Return type:
- property generator_manager: GeneratorManager | None#