A. Junghanns. Pushing the Limits: New Developments in Single-Agent Search. PhD thesis, University of Alberta, 1999.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....less states can be explored; while for Atomix 1,000,000 states per second can be generated, this number is for Sokoban around 10,000. Sokoban has been shown to be PSPACEcomplete by Culberson [Cul98] Junghanns has analyzed it thoroughly and written the sophisticated solver program Rolling Stone [Jun99]. 4 Complexity of Atomix 4.1 Complexity of Sliding Block Puzzles The time complexity of sliding block puzzles was the subject of intense research in the past. Though seemingly trivial, most variations are at least NPhard and, some, even PSPACE complete. Table 2 shows some results. The table was ....

Andreas Junghanns. Pushing the Limits: New Developments in Single-Agent Search. PhD thesis, University of Alberta, 1999.

....function that takes additional search information into account #Cormen, Leiserson, Rivest 1990#. Current heuristic search techniques such as iterative deepening #Korf 1985#, transposition tables #Reinefeld Marsland 1994#, #nite state machine pruning #Taylor Korf 1993#, relevance cuts #Junghanns 1999#, symbolic representation #Edelkamp Re#el 1998# and memorybased heuristics #Hern#adv#ogyi Holte 2000# e#ciently solve problems with 10 20 states and beyond, which compares with performance gains in symbolic model checking #Burch et al. 1992#. In our work we handle safety errors such as ....

Junghanns, A. 1999. Pushing the Limits: New Developments in Single-Agent Search. Ph.D. Dissertation, University of Alberta.

.... function that takes additional search information into account (Cormen, Leiserson, Rivest 1990) Current heuristic search techniques such as iterative deepening (Korf 1985) transposition tables (Reinefeld Marsland 1994) finite state machine pruning (Taylor Korf 1993) relevance cuts (Junghanns 1999), symbolic representation (Edelkamp Reffel 1998) and memorybased heuristics (Hern advogyi Holte 2000) efficiently solve problems with 10 20 states and beyond, whichcompares with performance gains in symbolic model checking (Burch et al. 1992) In our work we handle safety errors such as ....

Junghanns, A. 1999. Pushing the Limits: New Developments in Single-Agent Search. Ph.D. Dissertation, University of Alberta.

....evaluation is orders of magnitude slower and more complicated. Moreover, a good static full board evaluation depends on performing many auxiliary local tactical searches. In this respect Go can be compared to the single agent puzzle of Sokoban. Andreas Junghanns Sokoban solver Rolling Stone [26] derives much of its strength from auxiliary searches that solve or approximately solve subproblems, and thereby help to speed up the main search by many orders of magnitude. 2.4.2 Size and Structure of the Problem Space The search space for 19 19 Go is large compared to other popular board ....

A. Junghanns. Pushing the Limits: New Developments in Single-Agent Search. PhD thesis, University of Alberta, 1999.

....heuristics have not been applied yet. As a straight forward extension to the max pair heuristic it might be conjectured that for pairs of fluents and their corresponding relaxed solution length a weighted bipartite minimum matching algorithm (available in cubic time) as applied in Sokoban [31] might lead to a better lower bound approximation. The success of HSP has inspired the planners GRT by Refanidis and Vlahavas [43] and FF by Hoffmann [28] and influenced the development of the planners STAN and MIPS. GRT abbreviates a heuristic search planner based on greedy regression tables to ....

....forward and backward traversal. Therefore, the FF Heuristic is an elaboration to the HSP Heuristic, since the latter only considers the first phase. The efforts in computing a very accurate heuristic estimate correlates with data in solving single agent challenges like the 24 Puzzle [37] Sokoban [31], and Rubik s Cube [36] and suggests that even involved work for improving the heuristic pays off. With enforced hill climbing it further employs another search strategy and drastically reduces the explored portion of search space. It makes use of the fact that phenomena like big plateaus or ....

[Article contains additional citation context not shown here]

A. Junghanns, Pushing the Limits: New Developments in Single-Agent Search, Ph.D. dissertation, University of Alberta, 1999.

No context found.

A. Junghanns, 1999, "Pushing the Limits: New Developments in Single-Agent Search", PhD thesis, University of Alberta, Edmonton, Canada

....nodes dominate the cost of the search for some problems. Some additional heuristics for deciding when to execute pattern searches could result in further improvements in the overall search eciency. There are numerous parameters in the search, each of which can be tuned for maximal performance [7, 11]. Pattern searches have also been applied to sliding tile puzzles [7] The pro14 gram dynamically learns penalty patterns, such as linear con icts [6] The cost of the pattern searches is small compared to the large reductions in the IDA search tree. Deadlock tables (or pattern databases) are ....

....heuristics for deciding when to execute pattern searches could result in further improvements in the overall search eciency. There are numerous parameters in the search, each of which can be tuned for maximal performance [7, 11] Pattern searches have also been applied to sliding tile puzzles [7]. The pro14 gram dynamically learns penalty patterns, such as linear con icts [6] The cost of the pattern searches is small compared to the large reductions in the IDA search tree. Deadlock tables (or pattern databases) are another way to store pattern information. However, the patterns in such ....

A. Junghanns. Pushing the Limits: New Developments in Single-Agent Search. PhD thesis, University of Alberta, 1999.

No context found.

A. Junghanns. Pushing the Limits: New Developments in Single-Agent Search. PhD thesis, University of Alberta, 1999.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC