Differences between revisions 9 and 10
Revision 9 as of 2015-07-18 06:35:27
Size: 9068
Editor: XmlRpcBot
Comment:
Revision 10 as of 2015-10-22 10:13:13
Size: 9393
Editor: XmlRpcBot
Comment:
Deletions are marked like this. Additions are marked like this.
Line 12: Line 12:
lm_exhaust(cost_type=NORMAL, transform=<none>, reasonable_orders=false, only_causal_landmarks=false, disjunctive_landmarks=true, conjunctive_landmarks=true, no_orders=false, lm_cost_type=NORMAL) lm_exhaust(cost_type=NORMAL, transform=<none>, cache_h=true, reasonable_orders=false, only_causal_landmarks=false, disjunctive_landmarks=true, conjunctive_landmarks=true, no_orders=false, lm_cost_type=NORMAL)
Line 20: Line 20:
 * ''cache_h'' (bool): chache heuristic estimates
Line 33: Line 34:
lm_hm(m=2, cost_type=NORMAL, transform=<none>, reasonable_orders=false, only_causal_landmarks=false, disjunctive_landmarks=true, conjunctive_landmarks=true, no_orders=false, lm_cost_type=NORMAL) lm_hm(m=2, cost_type=NORMAL, transform=<none>, cache_h=true, reasonable_orders=false, only_causal_landmarks=false, disjunctive_landmarks=true, conjunctive_landmarks=true, no_orders=false, lm_cost_type=NORMAL)
Line 43: Line 44:
 * ''cache_h'' (bool): chache heuristic estimates
Line 54: Line 56:
lm_merged(lm_graphs, cost_type=NORMAL, transform=<none>, reasonable_orders=false, only_causal_landmarks=false, disjunctive_landmarks=true, conjunctive_landmarks=true, no_orders=false, lm_cost_type=NORMAL) lm_merged(lm_graphs, cost_type=NORMAL, transform=<none>, cache_h=true, reasonable_orders=false, only_causal_landmarks=false, disjunctive_landmarks=true, conjunctive_landmarks=true, no_orders=false, lm_cost_type=NORMAL)
Line 64: Line 66:
 * ''cache_h'' (bool): chache heuristic estimates
Line 81: Line 84:
lm_rhw(cost_type=NORMAL, transform=<none>, reasonable_orders=false, only_causal_landmarks=false, disjunctive_landmarks=true, conjunctive_landmarks=true, no_orders=false, lm_cost_type=NORMAL) lm_rhw(cost_type=NORMAL, transform=<none>, cache_h=true, reasonable_orders=false, only_causal_landmarks=false, disjunctive_landmarks=true, conjunctive_landmarks=true, no_orders=false, lm_cost_type=NORMAL)
Line 90: Line 93:
 * ''cache_h'' (bool): chache heuristic estimates
Line 103: Line 107:
lm_zg(cost_type=NORMAL, transform=<none>, reasonable_orders=false, only_causal_landmarks=false, disjunctive_landmarks=true, conjunctive_landmarks=true, no_orders=false, lm_cost_type=NORMAL) lm_zg(cost_type=NORMAL, transform=<none>, cache_h=true, reasonable_orders=false, only_causal_landmarks=false, disjunctive_landmarks=true, conjunctive_landmarks=true, no_orders=false, lm_cost_type=NORMAL)
Line 112: Line 116:
 * ''cache_h'' (bool): chache heuristic estimates

A landmark graph specification is either a newly created instance or a landmark graph that has been defined previously. This page describes how one can specify a new landmark graph instance. For re-using landmark graphs, see Landmark Predefinitions.

Warning: See OptionCaveats for using cost types with Landmarks

Exhaustive Landmarks

Exhaustively checks for each fact if it is a landmark.This check is done using relaxed planning.

lm_exhaust(cost_type=NORMAL, transform=<none>, cache_h=true, reasonable_orders=false, only_causal_landmarks=false, disjunctive_landmarks=true, conjunctive_landmarks=true, no_orders=false, lm_cost_type=NORMAL)
  • cost_type ({NORMAL, ONE, PLUSONE}): Operator cost adjustment type. No matter what this setting is, axioms will always be considered as actions of cost 0 by the heuristics that treat axioms as actions.

    • NORMAL: all actions are accounted for with their real cost

    • ONE: all actions are accounted for as unit cost

    • PLUSONE: all actions are accounted for as their real cost + 1 (except if all actions have original cost 1, in which case cost 1 is used). This is the behaviour known for the heuristics of the LAMA planner. This is intended to be used by the heuristics, not search engines, but is supported for both.

  • transform (AbstractTask): Optional task transformation for the heuristic. Currently only adapt_costs is available.

  • cache_h (bool): chache heuristic estimates

  • reasonable_orders (bool): generate reasonable orders

  • only_causal_landmarks (bool): keep only causal landmarks

  • disjunctive_landmarks (bool): keep disjunctive landmarks

  • conjunctive_landmarks (bool): keep conjunctive landmarks

  • no_orders (bool): discard all orderings

  • lm_cost_type ({NORMAL, ONE, PLUSONE}): landmark action cost adjustment

Relevant options: reasonable_orders, only_causal_landmarks

Language features supported:

  • conditional_effects: ignored, i.e. not supported

h^m Landmarks

The landmark generation method introduced by Keyder, Richter & Helmert (ECAI 2010).

lm_hm(m=2, cost_type=NORMAL, transform=<none>, cache_h=true, reasonable_orders=false, only_causal_landmarks=false, disjunctive_landmarks=true, conjunctive_landmarks=true, no_orders=false, lm_cost_type=NORMAL)
  • m (int): subset size (if unsure, use the default of 2)

  • cost_type ({NORMAL, ONE, PLUSONE}): Operator cost adjustment type. No matter what this setting is, axioms will always be considered as actions of cost 0 by the heuristics that treat axioms as actions.

    • NORMAL: all actions are accounted for with their real cost

    • ONE: all actions are accounted for as unit cost

    • PLUSONE: all actions are accounted for as their real cost + 1 (except if all actions have original cost 1, in which case cost 1 is used). This is the behaviour known for the heuristics of the LAMA planner. This is intended to be used by the heuristics, not search engines, but is supported for both.

  • transform (AbstractTask): Optional task transformation for the heuristic. Currently only adapt_costs is available.

  • cache_h (bool): chache heuristic estimates

  • reasonable_orders (bool): generate reasonable orders

  • only_causal_landmarks (bool): keep only causal landmarks

  • disjunctive_landmarks (bool): keep disjunctive landmarks

  • conjunctive_landmarks (bool): keep conjunctive landmarks

  • no_orders (bool): discard all orderings

  • lm_cost_type ({NORMAL, ONE, PLUSONE}): landmark action cost adjustment

Relevant options: m, reasonable_orders, conjunctive_landmarks, no_orders

Merged Landmarks

Merges the landmarks and orderings from the parameter landmarks

lm_merged(lm_graphs, cost_type=NORMAL, transform=<none>, cache_h=true, reasonable_orders=false, only_causal_landmarks=false, disjunctive_landmarks=true, conjunctive_landmarks=true, no_orders=false, lm_cost_type=NORMAL)
  • lm_graphs (list of LandmarkGraph):

  • cost_type ({NORMAL, ONE, PLUSONE}): Operator cost adjustment type. No matter what this setting is, axioms will always be considered as actions of cost 0 by the heuristics that treat axioms as actions.

    • NORMAL: all actions are accounted for with their real cost

    • ONE: all actions are accounted for as unit cost

    • PLUSONE: all actions are accounted for as their real cost + 1 (except if all actions have original cost 1, in which case cost 1 is used). This is the behaviour known for the heuristics of the LAMA planner. This is intended to be used by the heuristics, not search engines, but is supported for both.

  • transform (AbstractTask): Optional task transformation for the heuristic. Currently only adapt_costs is available.

  • cache_h (bool): chache heuristic estimates

  • reasonable_orders (bool): generate reasonable orders

  • only_causal_landmarks (bool): keep only causal landmarks

  • disjunctive_landmarks (bool): keep disjunctive landmarks

  • conjunctive_landmarks (bool): keep conjunctive landmarks

  • no_orders (bool): discard all orderings

  • lm_cost_type ({NORMAL, ONE, PLUSONE}): landmark action cost adjustment

Precedence: Fact landmarks take precedence over disjunctive landmarks, orderings take precedence in the usual manner (gn > nat > reas > o_reas).

Relevant options: Depends on landmarks

Note: Does not currently support conjunctive landmarks

Language features supported:

  • conditional_effects: supported if all components support them

RHW Landmarks

The landmark generation method introduced by Richter, Helmert and Westphal (AAAI 2008).

lm_rhw(cost_type=NORMAL, transform=<none>, cache_h=true, reasonable_orders=false, only_causal_landmarks=false, disjunctive_landmarks=true, conjunctive_landmarks=true, no_orders=false, lm_cost_type=NORMAL)
  • cost_type ({NORMAL, ONE, PLUSONE}): Operator cost adjustment type. No matter what this setting is, axioms will always be considered as actions of cost 0 by the heuristics that treat axioms as actions.

    • NORMAL: all actions are accounted for with their real cost

    • ONE: all actions are accounted for as unit cost

    • PLUSONE: all actions are accounted for as their real cost + 1 (except if all actions have original cost 1, in which case cost 1 is used). This is the behaviour known for the heuristics of the LAMA planner. This is intended to be used by the heuristics, not search engines, but is supported for both.

  • transform (AbstractTask): Optional task transformation for the heuristic. Currently only adapt_costs is available.

  • cache_h (bool): chache heuristic estimates

  • reasonable_orders (bool): generate reasonable orders

  • only_causal_landmarks (bool): keep only causal landmarks

  • disjunctive_landmarks (bool): keep disjunctive landmarks

  • conjunctive_landmarks (bool): keep conjunctive landmarks

  • no_orders (bool): discard all orderings

  • lm_cost_type ({NORMAL, ONE, PLUSONE}): landmark action cost adjustment

Relevant Options: reasonable_orders, only_causal_landmarks, disjunctive_landmarks, no_orders

Language features supported:

  • conditional_effects: supported

Zhu/Givan Landmarks

The landmark generation method introduced by Zhu & Givan (ICAPS 2003 Doctoral Consortium).

lm_zg(cost_type=NORMAL, transform=<none>, cache_h=true, reasonable_orders=false, only_causal_landmarks=false, disjunctive_landmarks=true, conjunctive_landmarks=true, no_orders=false, lm_cost_type=NORMAL)
  • cost_type ({NORMAL, ONE, PLUSONE}): Operator cost adjustment type. No matter what this setting is, axioms will always be considered as actions of cost 0 by the heuristics that treat axioms as actions.

    • NORMAL: all actions are accounted for with their real cost

    • ONE: all actions are accounted for as unit cost

    • PLUSONE: all actions are accounted for as their real cost + 1 (except if all actions have original cost 1, in which case cost 1 is used). This is the behaviour known for the heuristics of the LAMA planner. This is intended to be used by the heuristics, not search engines, but is supported for both.

  • transform (AbstractTask): Optional task transformation for the heuristic. Currently only adapt_costs is available.

  • cache_h (bool): chache heuristic estimates

  • reasonable_orders (bool): generate reasonable orders

  • only_causal_landmarks (bool): keep only causal landmarks

  • disjunctive_landmarks (bool): keep disjunctive landmarks

  • conjunctive_landmarks (bool): keep conjunctive landmarks

  • no_orders (bool): discard all orderings

  • lm_cost_type ({NORMAL, ONE, PLUSONE}): landmark action cost adjustment

Relevant options: reasonable_orders, no_orders

Language features supported:

  • conditional_effects: We think they are supported, but this is not 100% sure.

FastDownward: Doc/LandmarkFactory (last edited 2024-01-11 22:26:37 by XmlRpcBot)