## Odpop: An algorithm for open/distributed constraint optimization (2006)

### Cached

### Download Links

- [lang.is.kyushu-u.ac.jp]
- [jmvidal.cse.sc.edu]
- [liawww.epfl.ch]
- [www.aaai.org]
- [www.aaai.org]
- DBLP

### Other Repositories/Bibliography

Venue: | In AAAI |

Citations: | 29 - 5 self |

### BibTeX

@INPROCEEDINGS{Petcu06odpop:an,

author = {Adrian Petcu and Boi Faltings},

title = {Odpop: An algorithm for open/distributed constraint optimization},

booktitle = {In AAAI},

year = {2006},

pages = {703--708}

}

### OpenURL

### Abstract

Abstract. We propose ODPOP, a new distributed algorithm for open multiagent combinatorial optimization [3]. The ODOP algorithm explores the same search space as the dynamic programming algorithm DPOP [10] or the AND/OR search algorithm AOBB [2], but does so in an incremental, best-first fashion suitable for open problems. ODPOP has several advantages over DPOP. First, it uses messages whose size only grows linearly with the treewidth of the problem. Second, by letting agents explore values in a non-increasing order of preference, it saves a significant amount of messages and computation over the basic DPOP algorithm. To show the merits of our approach, we report on experiments with practically sized distributed meeting scheduling problems in a multiagent system. 1

### Citations

601 |
Constraint processing
- Dechter
- 2003
(Show Context)
Citation Context ...uted algorithm for open multiagent combinatorial optimization [3]. The ODOP algorithm explores the same search space as the dynamic programming algorithm DPOP [10] or the AND/OR search algorithm AOBB =-=[2]-=-, but does so in an incremental, best-first fashion suitable for open problems. ODPOP has several advantages over DPOP. First, it uses messages whose size only grows linearly with the treewidth of the... |

290 | Valued constraint satisfaction problems: Hard and easy problems
- Schiex, Fargier, et al.
- 1995
(Show Context)
Citation Context ...fy the notation, we use xi to denote either the variable itself, or its (virtual) agent, depending on the context. This is a multiagent instance of the valued CSP framework as defined by Schiex et al =-=[13]-=-. The goal is to find a complete instantiation X ∗ for the variables xi that maximizes the aggregate utility, i.e. the sum of utilities of individual relations. General optimization problems can conta... |

282 | Distributed Constraint Satisfaction Problems
- Yokoo, Hirayama
- 2000
(Show Context)
Citation Context ...e problem, and it may even be impossible. Distributed Constraint Satisfaction (DisCSP) has been developed to address such problems, formalized in the early 90’s by Dechter [1], Meisels [15] and Yokoo =-=[16]-=-. Existing techniques for distributed constraint optimization fall in two main categories: complete methods (guarantee optimality, but may be expensive), and local search methods (no quality guarantee... |

248 | Adopt: Asynchronous distributed constraint optimization with quality guarantees
- Modi, Shen, et al.
(Show Context)
Citation Context ...zation that feature unbounded domains (Faltings & Macho-Gonzalez 2005). The ODPOP algorithm explores the same search space as the dynamic programming algorithm DPOP (Petcu & Faltings 2005b) or ADOPT (=-=Modi et al. 2005-=-), but does so in an incremental, best-first fashion suitable for open problems. ODPOP has several advantages over DPOP. First, it uses messages whose size only grows linearly with the treewidth of th... |

218 | Algorithms for distributed constraint satisfaction: A review. Autonomous Agents and Multi-Agent Systems
- Yokoo, Hirayama
(Show Context)
Citation Context ...ality, but may be expensive), and local search methods (no quality guarantees, tipically less expensive). Complete algorithms can be again divided in two main categories: backtracking algorithms (see =-=[17, 1, 16, 14, 8, 5]-=-), and dynamic programming algorithms (see [10, 6]). Dynamic programming algorithms (e.g. DPOP [10]) have the important advantage that they produce fewer messages, therefore less overhead. DPOP for ex... |

170 |
A Sufficient Condition for Backtrackbounded Search
- Freuder
- 1985
(Show Context)
Citation Context ...s G and the property that adjacent nodes from the original graph fall in the same branch of the tree (e.g. x0 and x11 in Figure 1). DFS trees have already been investigated as a means to boost search =-=[4, 2]-=-. Due to the relative independence of nodes lying in different branches of the DFS tree, it is possible to perform search in parallel on these independent branches. Fig. 1: A problem graph and a roote... |

141 | A scalable method for multiagent constraint optimization
- Petcu, Faltings
- 2005
(Show Context)
Citation Context ...Abstract. We propose ODPOP, a new distributed algorithm for open multiagent combinatorial optimization [3]. The ODOP algorithm explores the same search space as the dynamic programming algorithm DPOP =-=[10]-=- or the AND/OR search algorithm AOBB [2], but does so in an incremental, best-first fashion suitable for open problems. ODPOP has several advantages over DPOP. First, it uses messages whose size only ... |

108 | Taking DCOP to the real world: Efficient complete solutions for distributed multi- event scheduling
- Maheswaran, Tambe, et al.
- 2004
(Show Context)
Citation Context ...stributed meeting scheduling in an organization with a hierarchical structure (a tree with departments as nodes, and a set of agents working in each department). The CSP model is the PEAV model from (=-=Maheswaran et al. 2004-=-). Each agent has multiple variables: one for the start time of each meeting it participates in, with 10 timeslots as values. Mutual exclusion constraints are imposed on the variables of an agent, and... |

85 | Interleaved backtracking in distributed constraint networks
- Hamadi
(Show Context)
Citation Context ...ality, but may be expensive), and local search methods (no quality guarantees, tipically less expensive). Complete algorithms can be again divided in two main categories: backtracking algorithms (see =-=[17, 1, 16, 14, 8, 5]-=-), and dynamic programming algorithms (see [10, 6]). Dynamic programming algorithms (e.g. DPOP [10]) have the important advantage that they produce fewer messages, therefore less overhead. DPOP for ex... |

70 | On the Feasibility of Distributed Constraint Satisfaction
- CoUin, Dechter, et al.
- 1991
(Show Context)
Citation Context ...nows all the variables in the problem, and it may even be impossible. Distributed Constraint Satisfaction (DisCSP) has been developed to address such problems, formalized in the early 90’s by Dechter =-=[1]-=-, Meisels [15] and Yokoo [16]. Existing techniques for distributed constraint optimization fall in two main categories: complete methods (guarantee optimality, but may be expensive), and local search ... |

42 | M-dpop: Faithful distributed implementation of efficient social choice problems
- Petcu, Faltings, et al.
- 2006
(Show Context)
Citation Context ...is a ternary constraint, as it involves 3 variables: x1,x3,x8). We mainly assume here only unary and binary constraints/relations; however, DPOP and ODPOP extend easily to non-binary constraints (see =-=[12]-=-). Dealing with these constraints requires minor modifications in the DFS traversal phase of DPOP/ODPOP (see Section 3.1), and in the bottom-up propagation phase (see Section 3.2). 1 Hard constraints ... |

39 | Selfstabilizing distributed constraint satisfaction - COLLIN, DECHTER, et al. - 1999 |

37 |
Modeling and solving distributed constraint satisfaction problems (dcsps
- Solotorevsky, Gudes, et al.
- 1996
(Show Context)
Citation Context ...variables in the problem, and it may even be impossible. Distributed Constraint Satisfaction (DisCSP) has been developed to address such problems, formalized in the early 90’s by Dechter [1], Meisels =-=[15]-=- and Yokoo [16]. Existing techniques for distributed constraint optimization fall in two main categories: complete methods (guarantee optimality, but may be expensive), and local search methods (no qu... |

36 | Open constraint programming
- Faltings, Macho-Gonzalez
- 2005
(Show Context)
Citation Context ...Fédérale de Lausanne (EPFL), CH-1015 Lausanne (Switzerland) {adrian.petcu, boi.faltings}@epfl.ch Abstract. We propose ODPOP, a new distributed algorithm for open multiagent combinatorial optimization =-=[3]-=-. The ODOP algorithm explores the same search space as the dynamic programming algorithm DPOP [10] or the AND/OR search algorithm AOBB [2], but does so in an incremental, best-first fashion suitable f... |

25 |
Yokoo M. Adopt: asynchronous distributed constraint optimization with quality guarantees
- Modi, Shen, et al.
- 2005
(Show Context)
Citation Context ...ality, but may be expensive), and local search methods (no quality guarantees, tipically less expensive). Complete algorithms can be again divided in two main categories: backtracking algorithms (see =-=[17, 1, 16, 14, 8, 5]-=-), and dynamic programming algorithms (see [10, 6]). Dynamic programming algorithms (e.g. DPOP [10]) have the important advantage that they produce fewer messages, therefore less overhead. DPOP for ex... |

21 |
Varakantham P. Taking DCOP to the Real World: Efficient Complete Solutions for Distributed Multi-Event Scheduling
- Maheswaran, Tambe, et al.
- 2004
(Show Context)
Citation Context ...istributed meeting scheduling in an organization with a hierarchical structure (a tree with departments as nodes, and a set of agents working in each department). The CSP model is the PEAV model from =-=[7]-=-. Each agent has multiple variables: one for the start time of each meeting it participates in, with 10 timeslots as values. Mutual exclusion constraints are imposed on the variables of an agent, and ... |

20 | Unifying cluster-tree decompositions for automated reasoning
- Kask, Dechter, et al.
- 2005
(Show Context)
Citation Context ...ty guarantees, tipically less expensive). Complete algorithms can be again divided in two main categories: backtracking algorithms (see [17, 1, 16, 14, 8, 5]), and dynamic programming algorithms (see =-=[10, 6]-=-). Dynamic programming algorithms (e.g. DPOP [10]) have the important advantage that they produce fewer messages, therefore less overhead. DPOP for example requires a number of messages which is linea... |

17 | Asynchronous search with private constraints
- Silaghi, Sam-Haroud, et al.
- 2000
(Show Context)
Citation Context |

12 | 2005a. Approximations in distributed optimization - Petcu, Faltings |

7 | and Boi Faltings, “Approximations in distributed optimization - Petcu - 2005 |

5 | fault-containing multiagent combinatorial optimization - Superstabilizing - 2005 |

3 | fault-containing multiagent combinatorial optimization - Petcu, Faltings, et al. - 2005 |