Efficient Computer Game Search in Desktop Grids based on Worker Group Scheduling

• Main Authors: Zih-Yang-Huang, 黃子洋
• Other Authors: CHEN,LONG-BIN
• Format: Others
• Language: zh-TW
• Published: 2014
• Online Access: http://ndltd.ncl.edu.tw/handle/55347540777779048609

碩士 === 東海大學 === 資訊工程學系 === 102 === AI has long been based on computer chess game full of information as the main study , perfect information game is easy to grasp the disk status , trial bureau low complexity and exclude factors that influence the probability of nature. Dark Chess computers belong to the game of incomplete information , and which contains the probability of factors , the higher complexity of the trial bureau makes the game not the best result obtained search method , but through sound turned chess , move, eat chess and the weight distribution of choices can result tends to idealize and greatly improve the odds . Because in real life there are many applications are non- perfect information , this type of problem increasing attention . The game tree search depth has always been a number of computer chess games headaches , to get a more accurate search results will inevitably have to improve the depth of the tree , so spend a lot of computing time, and BOINC volunteer computing platform is the world's one of the best grid computing platform , many scientific computing are using this platform for computing and get excellent results, this paper combines computer Dark Chess with BOINC volunteer computing technology, proposed job scheduling and anonymous official will be divided into two categories Worker and Worker concept clustering , and rewrite BOINC internal mechanisms to improve the speed of the game searching in order to achieve faster and more accurate results.