Differences between revisions 2 and 12 (spanning 10 versions)
⇤ ← Revision 2 as of 2015-10-30 23:23:29 →
Size: 5855
Editor: XmlRpcBot
Comment:
← Revision 12 as of 2022-02-10 13:59:05 → ⇥
Size: 4904
Editor: XmlRpcBot
Comment:
Deletions are marked like this. Additions are marked like this.
Line 3: Line 3:
== Delete relaxation constraints ==

Operator-counting constraints based on the delete relaxation. By default the constraints encode an easy-to-compute relaxation of h^+. If both use_time_vars and use_integer_vars are set to true and this is the only constriant in an operator-counting heursitic with integer variables for the operator counts, the resulting heuristic is the optimal delete-relaxation heuristic h^+. For details, see

 * Tatsuya Imai and Alex Fukunaga.<<BR>>
 [[https://www.jair.org/index.php/jair/article/download/10972/26119/|On a practical, integer-linear programming model for delete-freetasks and its use as a heuristic for cost-optimal planning]].<<BR>>
 ''Journal of Artificial Intelligence Research'' 54:631-677. 2015.

{{{
delete_relaxation_constraints(use_time_vars=false, use_integer_vars=false)
}}}

 * ''use_time_vars'' (bool): use variables for time steps. With these additional variables the constraints enforce an order between the selected operators. Leaving this off (default) corresponds to the time relaxation by Imai and Fukunaga. Switching it on, can increase the heuristic value but will increase the size of the constraints which has a strong impact on runtime. Constraints involving time variables use a big-M encoding, so they are more useful if used with integer variables.
 * ''use_integer_vars'' (bool): restrict auxiliary variables to integer values. These variables encode whether operators are used, facts are reached, which operator first achieves which fact, and in which order the operators are used. Restricting them to integers generally improves the heursitic value at the cost of increased runtime.
Line 4: Line 18:
Computes a set of landmarks in each state using the LM-cut method. For each landmark L the constraint sum_{o in L} Count_o >= 1 is added to the operator counting LP temporarily. After the heuristic value for the state is computed, all temporary constraints are removed again. For details, see
Line 5: Line 20:
Computes a set of landmarks in each state using the LM-cut method. For each landmark L the constraint sum_{o in L} Count_o >= 1 is added to the operator counting LP temporarily. After the heuristic value for the state is computed, all temporary constraints are removed again. For details, see
Line 8: Line 22:
 In ''Proceedings of the Twenty-Fourth International Conference on Automated Planning and Scheduling (ICAPS 2014)'', pp. 226-234. AAAI Press 2014.  In ''Proceedings of the Twenty-Fourth International Conference on Automated Planning and Scheduling (ICAPS 2014)'', pp. 226-234. AAAI Press, 2014.
Line 11: Line 26:
 In ''Proceedings of the Twenty-Third International Joint Conference on Artificial Intelligence (IJCAI 2013)'', pp. 2268–2274. 2013.  In ''Proceedings of the Twenty-Third International Joint Conference on Artificial Intelligence (IJCAI 2013)'', pp. 2268-2274. AAAI Press, 2013.
Line 17: Line 32:
== Posthoc optimization constraints for iPDB patterns == == Posthoc optimization constraints ==
Line 19: Line 34:
A pattern collection is discovered, using iPDB hillclimbing (see [Doc/[[Doc/Heuristic#iPDB|iPDB]]]).The generator will compute a PDB for each pattern and add the constraint h(s) <= sum_{o in relevant(h)} Count_o. For details, see The generator will compute a PDB for each pattern and add the constraint h(s) <= sum_{o in relevant(h)} Count_o. For details, see
Line 22: Line 38:
 In ''Proceedings of the Twenty-Third International Joint Conference on Artificial Intelligence (IJCAI 2013)'', pp. 2357-2364. 2013.  In ''Proceedings of the Twenty-Third International Joint Conference on Artificial Intelligence (IJCAI 2013)'', pp. 2357-2364. AAAI Press, 2013.
Line 25: Line 41:
pho_constraints_ipdb(pdb_max_size=2000000, collection_max_size=20000000, num_samples=1000, min_improvement=10, max_time=infinity) pho_constraints(patterns=systematic(2))
Line 28: Line 44:
 * ''pdb_max_size'' (int [1, infinity]): maximal number of states per pattern database
 * ''collection_max_size'' (int [1, infinity]): maximal number of states in the pattern collection
 * ''num_samples'' (int [1, infinity]): number of samples (random states) on which to evaluate each candidate pattern collection
 * ''min_improvement'' (int [1, infinity]): minimum number of samples on which a candidate pattern collection must improve on the current one to be considered as the next pattern collection
 * ''max_time'' (double [0.0, infinity]): maximum time in seconds for improving the initial pattern collection via hill climbing. If set to 0, no hill climbing is performed at all.
== Posthoc optimization constraints for manually specified patterns ==
The generator will compute a PDB for each pattern and add the constraint h(s) <= sum_{o in relevant(h)} Count_o. For details, see
 * Florian Pommerening, Gabriele Roeger and Malte Helmert.<<BR>>
 [[http://ijcai.org/papers13/Papers/IJCAI13-347.pdf|Getting the Most Out of Pattern Databases for Classical Planning]].<<BR>>
 In ''Proceedings of the Twenty-Third International Joint Conference on Artificial Intelligence (IJCAI 2013)'', pp. 2357-2364. 2013.

{{{
pho_constraints_manual(patterns=<none>, combo=false, max_states=1000000)
}}}

 * ''patterns'' (list of list of int): list of patterns (which are lists of variable numbers of the planning task). Default: each goal variable is used as a single-variable pattern in the collection.
 * ''combo'' (bool): use the combo strategy
 * ''max_states'' (int [1, infinity]): maximum abstraction size for combo strategy
== Posthoc optimization constraints for systematically generated patterns ==
All (interesting) patterns with up to pattern_max_size variables are generated. The generator will compute a PDB for each pattern and add the constraint h(s) <= sum_{o in relevant(h)} Count_o. For details, see
 * Florian Pommerening, Gabriele Roeger and Malte Helmert.<<BR>>
 [[http://ijcai.org/papers13/Papers/IJCAI13-347.pdf|Getting the Most Out of Pattern Databases for Classical Planning]].<<BR>>
 In ''Proceedings of the Twenty-Third International Joint Conference on Artificial Intelligence (IJCAI 2013)'', pp. 2357-2364. 2013.

{{{
pho_constraints_systematic(pattern_max_size=1, only_interesting_patterns=true)
}}}

 * ''pattern_max_size'' (int): max number of variables per pattern
 * ''only_interesting_patterns'' (bool): Only consider the union of two disjoint patterns if the union has more information than the individual patterns.		  * ''patterns'' ([[Doc/PatternCollectionGenerator|PatternCollectionGenerator]]): pattern generation method
Line 60: Line 47:
Line 62: Line 50:
 In ''Proceedings of the Thirteenth International Conference on Principles and Practice of Constraint Programming (CP 2007)'', pp. 651–665. 2007.  In ''Proceedings of the Thirteenth International Conference on Principles and Practice of Constraint Programming (CP 2007)'', pp. 651-665. Springer-Verlag, 2007.
Line 65: Line 54:
 In ''Proceedings of the Twenty-Third International Joint Conference on Artificial Intelligence (IJCAI 2013)'', pp. 2268–2274. 2013.  In ''Proceedings of the Twenty-Third International Joint Conference on Artificial Intelligence (IJCAI 2013)'', pp. 2268-2274. AAAI Press, 2013.
Line 68: Line 58:
 In ''Proceedings of the Twenty-Fourth International Conference on Automated Planning and Scheduling (ICAPS 2014)'', pp. 226-234. AAAI Press 2014.  In ''Proceedings of the Twenty-Fourth International Conference on Automated Planning and Scheduling (ICAPS 2014)'', pp. 226-234. AAAI Press, 2014.
Line 74: Line 64:
/* moin code generated by txt2tags 2.6b (http://txt2tags.sf.net) */ /* moin code generated by txt2tags 2.6 (http://txt2tags.org) */

Contents

  1. Delete relaxation constraints
  2. LM-cut landmark constraints
  3. Posthoc optimization constraints
  4. State equation constraints

Delete relaxation constraints

Operator-counting constraints based on the delete relaxation. By default the constraints encode an easy-to-compute relaxation of h+. If both use_time_vars and use_integer_vars are set to true and this is the only constriant in an operator-counting heursitic with integer variables for the operator counts, the resulting heuristic is the optimal delete-relaxation heuristic h+. For details, see 

delete_relaxation_constraints(use_time_vars=false, use_integer_vars=false)

  • use_time_vars (bool): use variables for time steps. With these additional variables the constraints enforce an order between the selected operators. Leaving this off (default) corresponds to the time relaxation by Imai and Fukunaga. Switching it on, can increase the heuristic value but will increase the size of the constraints which has a strong impact on runtime. Constraints involving time variables use a big-M encoding, so they are more useful if used with integer variables.

  • use_integer_vars (bool): restrict auxiliary variables to integer values. These variables encode whether operators are used, facts are reached, which operator first achieves which fact, and in which order the operators are used. Restricting them to integers generally improves the heursitic value at the cost of increased runtime.

LM-cut landmark constraints

Computes a set of landmarks in each state using the LM-cut method. For each landmark L the constraint sum_{o in L} Count_o >= 1 is added to the operator counting LP temporarily. After the heuristic value for the state is computed, all temporary constraints are removed again. For details, see 

lmcut_constraints()

Posthoc optimization constraints

The generator will compute a PDB for each pattern and add the constraint h(s) <= sum_{o in relevant(h)} Count_o. For details, see 

pho_constraints(patterns=systematic(2))

State equation constraints

For each fact, a permanent constraint is added that considers the net change of the fact, i.e., the total number of times the fact is added minus the total number of times is removed. The bounds of each constraint depend on the current state and the goal state and are updated in each state. For details, see 

state_equation_constraints()

FastDownward: Doc/ConstraintGenerator (last edited 2024-10-07 12:37:27 by XmlRpcBot)