Where Ignoring Delete Lists Works, Part II: Causal Graphs J¨ org Hoffmann INRIA Nancy, France June 14, 2011 J¨ org Hoffmann Where Ignoring Delete Lists Works, Part II: Causal Graphs 1/23
Outline ◮ What happened? ◮ On causal graphs and h + ◮ Guaranteed global analysis ◮ Approximate local analysis ◮ Diagnosis ◮ Conclusion J¨ org Hoffmann Where Ignoring Delete Lists Works, Part II: Causal Graphs 2/23
Outline ◮ What happened? ◮ On causal graphs and h + ◮ Guaranteed global analysis ◮ Approximate local analysis ◮ Diagnosis ◮ Conclusion J¨ org Hoffmann Where Ignoring Delete Lists Works, Part II: Causal Graphs 3/23
FF J¨ org Hoffmann Where Ignoring Delete Lists Works, Part II: Causal Graphs 4/23
Where Ignoring Delete Lists Works Mystery Blocksworld−Arm Pipesworld−Tank Mprime Depots Pipesworld−NoTank Rovers Freecell Driverlog PSR Airport Hanoi [0] Blocksworld−NoArm [0] local minima ed <= c Transport [0] Grid [0] Tyreworld [0,6] Elevators [0,1] Satellite [4,4] flat ed <= c Logistics [0,1] Zenotravel [2,2] Ferry [0,1] Miconic−STRIPS [0,1] Gripper [0,1] Movie [0,1] Simple−Tsp [0,0] undirected harmless recognized unrecognized red: no local minima at all under h + J¨ org Hoffmann Where Ignoring Delete Lists Works, Part II: Causal Graphs 5/23
Can we recognize this automatically? E at E == 1 EUR mv D E D += 1 EUR B C at D 1 EUR A mv B D mv C D mv D C at B at C mv A B at A J¨ org Hoffmann Where Ignoring Delete Lists Works, Part II: Causal Graphs 6/23
Can we recognize this automatically? E at E == 1 EUR mv D E D += 1 EUR B C at D 1 EUR A mv B D mv C D mv D C at B at C mv A B at A Works only in trivialities; explodes quickly J¨ org Hoffmann Where Ignoring Delete Lists Works, Part II: Causal Graphs 6/23
Time passes . . . ← me in 2002 J¨ org Hoffmann Where Ignoring Delete Lists Works, Part II: Causal Graphs 7/23
Time passes . . . ← me in 2003 J¨ org Hoffmann Where Ignoring Delete Lists Works, Part II: Causal Graphs 7/23
Time passes . . . ← me in 2004 J¨ org Hoffmann Where Ignoring Delete Lists Works, Part II: Causal Graphs 7/23
Time passes . . . ← me in 2005 J¨ org Hoffmann Where Ignoring Delete Lists Works, Part II: Causal Graphs 7/23
Time passes . . . ← me in 2006 J¨ org Hoffmann Where Ignoring Delete Lists Works, Part II: Causal Graphs 7/23
Time passes . . . ← me in 2007 J¨ org Hoffmann Where Ignoring Delete Lists Works, Part II: Causal Graphs 7/23
Time passes . . . ← me in 2008 J¨ org Hoffmann Where Ignoring Delete Lists Works, Part II: Causal Graphs 7/23
Time passes . . . ← me in 2009 J¨ org Hoffmann Where Ignoring Delete Lists Works, Part II: Causal Graphs 7/23
2009 J¨ org Hoffmann Where Ignoring Delete Lists Works, Part II: Causal Graphs 8/23
2009 J¨ org Hoffmann Where Ignoring Delete Lists Works, Part II: Causal Graphs 8/23
2009 J¨ org Hoffmann Where Ignoring Delete Lists Works, Part II: Causal Graphs 8/23
Carlos & Luciana Shortly after the presentation. Carlos, Luciana, and J¨ org sit around a table. The conversation goes like this: J¨ org Hoffmann Where Ignoring Delete Lists Works, Part II: Causal Graphs 9/23
Carlos & Luciana Shortly after the presentation. Carlos, Luciana, and J¨ org sit around a table. The conversation goes like this: Carlos/Luciana: “When we made PDDL models, it was very hard to know how to design them so that planners would perform better. Couldn’t one build a tool based on recognizing h + toplogy?” J¨ org Hoffmann Where Ignoring Delete Lists Works, Part II: Causal Graphs 9/23
Carlos & Luciana Shortly after the presentation. Carlos, Luciana, and J¨ org sit around a table. The conversation goes like this: Carlos/Luciana: “When we made PDDL models, it was very hard to know how to design them so that planners would perform better. Couldn’t one build a tool based on recognizing h + toplogy?” J¨ org: “Oh yeah, I already tried that during my PhD, but it didn’t work.” J¨ org Hoffmann Where Ignoring Delete Lists Works, Part II: Causal Graphs 9/23
Carlos & Luciana Shortly after the presentation. Carlos, Luciana, and J¨ org sit around a table. The conversation goes like this: Carlos/Luciana: “When we made PDDL models, it was very hard to know how to design them so that planners would perform better. Couldn’t one build a tool based on recognizing h + toplogy?” J¨ org: “Oh yeah, I already tried that during my PhD, but it didn’t work.” Carlos/Luciana: “But couldn’t we do something like XYZ?” J¨ org Hoffmann Where Ignoring Delete Lists Works, Part II: Causal Graphs 9/23
Carlos & Luciana Shortly after the presentation. Carlos, Luciana, and J¨ org sit around a table. The conversation goes like this: Carlos/Luciana: “When we made PDDL models, it was very hard to know how to design them so that planners would perform better. Couldn’t one build a tool based on recognizing h + toplogy?” J¨ org: “Oh yeah, I already tried that during my PhD, but it didn’t work.” Carlos/Luciana: “But couldn’t we do something like XYZ?” J¨ org: “Hm I don’t think so.” J¨ org Hoffmann Where Ignoring Delete Lists Works, Part II: Causal Graphs 9/23
Carlos & Luciana Shortly after the presentation. Carlos, Luciana, and J¨ org sit around a table. The conversation goes like this: Carlos/Luciana: “When we made PDDL models, it was very hard to know how to design them so that planners would perform better. Couldn’t one build a tool based on recognizing h + toplogy?” J¨ org: “Oh yeah, I already tried that during my PhD, but it didn’t work.” Carlos/Luciana: “But couldn’t we do something like XYZ?” J¨ org: “Hm I don’t think so.” Carlos/Luciana: “ αβγ maybe?” J¨ org Hoffmann Where Ignoring Delete Lists Works, Part II: Causal Graphs 9/23
Carlos & Luciana Shortly after the presentation. Carlos, Luciana, and J¨ org sit around a table. The conversation goes like this: Carlos/Luciana: “When we made PDDL models, it was very hard to know how to design them so that planners would perform better. Couldn’t one build a tool based on recognizing h + toplogy?” J¨ org: “Oh yeah, I already tried that during my PhD, but it didn’t work.” Carlos/Luciana: “But couldn’t we do something like XYZ?” J¨ org: “Hm I don’t think so.” Carlos/Luciana: “ αβγ maybe?” . . . [45 minutes later] . . . J¨ org Hoffmann Where Ignoring Delete Lists Works, Part II: Causal Graphs 9/23
Carlos & Luciana Shortly after the presentation. Carlos, Luciana, and J¨ org sit around a table. The conversation goes like this: Carlos/Luciana: “When we made PDDL models, it was very hard to know how to design them so that planners would perform better. Couldn’t one build a tool based on recognizing h + toplogy?” J¨ org: “Oh yeah, I already tried that during my PhD, but it didn’t work.” Carlos/Luciana: “But couldn’t we do something like XYZ?” J¨ org: “Hm I don’t think so.” Carlos/Luciana: “ αβγ maybe?” . . . [45 minutes later] . . . J¨ org: “Look, just consider Blocksworld and Logistics. One has local minima, the other doesn’t. Still both have deletes.” J¨ org Hoffmann Where Ignoring Delete Lists Works, Part II: Causal Graphs 9/23
Carlos & Luciana Shortly after the presentation. Carlos, Luciana, and J¨ org sit around a table. The conversation goes like this: Carlos/Luciana: “When we made PDDL models, it was very hard to know how to design them so that planners would perform better. Couldn’t one build a tool based on recognizing h + toplogy?” J¨ org: “Oh yeah, I already tried that during my PhD, but it didn’t work.” Carlos/Luciana: “But couldn’t we do something like XYZ?” J¨ org: “Hm I don’t think so.” Carlos/Luciana: “ αβγ maybe?” . . . [45 minutes later] . . . J¨ org: “Look, just consider Blocksworld and Logistics. One has local minima, the other doesn’t. Still both have deletes.” J¨ org: “And there is no other obvious difference in their structure . . . ” J¨ org Hoffmann Where Ignoring Delete Lists Works, Part II: Causal Graphs 9/23
Carlos & Luciana Shortly after the presentation. Carlos, Luciana, and J¨ org sit around a table. The conversation goes like this: Carlos/Luciana: “When we made PDDL models, it was very hard to know how to design them so that planners would perform better. Couldn’t one build a tool based on recognizing h + toplogy?” J¨ org: “Oh yeah, I already tried that during my PhD, but it didn’t work.” Carlos/Luciana: “But couldn’t we do something like XYZ?” J¨ org: “Hm I don’t think so.” Carlos/Luciana: “ αβγ maybe?” . . . [45 minutes later] . . . J¨ org: “Look, just consider Blocksworld and Logistics. One has local minima, the other doesn’t. Still both have deletes.” J¨ org: “And there is no other obvious difference in their structure . . . ” J¨ org: “. . . Causal graphs!!!” J¨ org Hoffmann Where Ignoring Delete Lists Works, Part II: Causal Graphs 9/23
Recommend
More recommend
