外文翻譯---關于中國象棋的一些簡要介紹

1、<p><b>　　附 錄</b></p><p>　　附錄A：A Introduction About Chinese Chess</p><p>　　Some Major Differences Between Chinese Chess And Western Chess Are As Follows. </p><p>

2、　　1. Western chess is played on an 8×8 board with six different kinds of pieces, while Chinese chess is played on a 9×10 board with seven different kinds of pieces. The Chinese chess board is more difficult to

3、 represent on a computer. </p><p>　　2. The board of Chinese chess is more like a “battlefield” than a chessboard is. The Chinese chessboard has two special territories. One is the Palace. The Kings and Advi

4、sors can move only within their Palace. The other is the River. Elephants cannot cross the River. These constraints seem to reduce the complexity of Chinese chess, since these pieces at first may appear not to be able to

5、 threaten pieces on the other side of the River, but this is a fallacy. These pieces can, in fact, directly t</p><p>　　3. Cannons are very special pieces. Cannons move like Rooks but have to jump over piece

6、 to capture it. Thus, any piece co-operating with Cannons will have the same power as a Rook. Western chess has no similar piece. </p><p>　　4. Although Kings cannot leave the Palace, Chinese chess has a spe

7、cial rule that allows them to threaten their counterparts: Kings may not see each other face to face. Based on this rule, Kings sometimes have a power similar to that of Rooks, which is very useful in the endgame stage.

8、</p><p>　　5. Blocking rules apply to Elephants and Horses. Western chess has no similar rule. </p><p>　　6. Pawns move one space at a time. When a Pawn reaches the opposite bottom line, it cann

9、ot be turned into another piece as it can be in Western chess. </p><p>　　7. Repetitive situations are relatively simple in Western chess. In Chinese chess, however, because of the larger board and the chara

10、cteristics of certain pieces, repeating situations in which neither side wants to make a concession, is more likely to arise. Special rules are required to govern these situations. In some extreme situations, a referee m

11、ust make a judgement. Therefore, resolving repeating situations is a challenge in computer Chinese chess.</p><p>　　Quiescent Search </p><p>　　A quiescence search explores all captures and checks

12、 from the current position, terminating at quiescent positions, defined as those in which no captures or checks are available.</p><p>　　Processing Perpetual Situations </p><p>　　As we described,

13、 perpetual situations have to be identified during the search process. The influence of a check effect happens frequently. If the program only assesses at the first level, it may leads to a wrong board status that repeat

14、edly tries to kill the King.</p><p>　　Computing And Storing The Results In Advance </p><p>　　When the opponent is thinking, the program could make use of this “free time” to predict the opponen

15、t’s move and store the results. If the prediction is right, the results are fetched without further computation.</p><p>　　Forward Pruning </p><p>　　This includes razoring, futility pruning, and

16、null move.</p><p>　　Move Ordering </p><p>　　This includes killer heuristic, history heuristic (Schaeffer, 1983), and internal iterative deepening. </p><p>　　Transposition Table <

17、/p><p>　　Store position scores in a hash table for use if the same position is reached again (Marsland, 1986).</p><p>　　Balance between search and knowledge </p><p>　　Chinese-chess pro

18、grams are like Western-chess programs in that they can be divided into two types: knowledge-based and brute-force programs. A knowledge-based program emphasises the evaluation of positions. A brute-force program uses the

19、 ability to compute quickly and to explore the game tree deeply, so as to discover better moves. Even in Western-chess, the better approach has not been identified yet. Chinese chess has more different pieces and a large

20、r board than chess, so knowledge-based prog</p><p>　　Other Speed-Up Methods </p><p>　　Iterative deepening and aspiration windows (Kaindl, Shams, and Horacek, 1991). Before PVS begins, the progra

21、m uses a threat-sequence search, in particular when the game tree is narrow and deep. It determines whether serial checking may be performed. In such a tree, only the checking of the King and avoiding having one’s own Ki

22、ng checked, are considered. Comparing to the endgame stage, the middle game has fewer legal moves that can be played. Thus, more plies can be searched in the middle game. </p><p><b>　　Endgame </b>

23、;</p><p>　　In the endgame, few Pawns remain, so other strong pieces have more space to move. The search depth in this stage is limited owing to the freedom of the pieces and the large size of the Chinese che

24、ssboard. Since, human players can employ expert knowledge to help search in this stage, most Chinese-chess programs are inferior to human players in this stage. Hence, improving the endgame is key to defeating human top

25、players. </p><p>　　The earliest retrograde algorithm is proposed by Ströhlein (1970). Subsequently Van den Herik and Herschberg (1985) have significantly contributed to the retrograde concept. Then Thom

26、pson (1986, 1996) has published his great contributions to Western-chess endgames. His endgame databases built on retrograde analysis succeeded to increase the playing strength of many chess-playing programs. There after

27、 we have seen the contributions by Nalimov et al. (2001). In recent years, many people have show</p><p>　　Rather than being built on retrograde analysis, an endgame database can also be built on massive exp

28、erts’ knowledge (Chen, 1998). This kind of endgame database is smaller, more practical and easier to incorporate into the existing Chinese-chess programs.</p><p>　　附錄B：關于中國象棋的一些簡要介紹</p><p>　　國際象

29、棋和中國象棋一些主要的差別如下：</p><p>　　1、國際象棋具有六種不同的棋子，并且它是在一個8 × 8的棋盤上對弈的；而中國象棋具有七種不同棋子，棋盤是9 × 10的。中國象棋棋盤更難在計算機上表示 The Chinese chessboard is more difficult to represent on a computer.中國象棋棋盤更難在計算機上。 </p>

30、<p>　　2.  The board of Chinese chess is more like a “battlefield” than a chessboard is.2、中國象棋棋盤要比國際象棋棋盤看起來更像一個戰(zhàn)場。中國象棋The Chinese chessboard has two special territori棋盤有兩個特殊的區(qū)域：One is the Palace. The Kings and

31、Advisors can move only within their Pala一個是九宮，將和士智能在九宮內走動；The other is the Rive其二是楚河，象不能過河。 These constraints seem to reduce the complexity of Chinese chess, since these pieces at first may appear not to be able to threa

32、ten pieces on the other side of the River, but this is a fallacy. 這些制約因素似乎減少了</p><p>　　3.  Cannons are very special pieces.3、炮是一個非常特殊的棋子炮可以像炮。炮可以像車一樣走動，但是必須在隔一個子的情況下才能吃掉另一個子。因此，任何個棋子和炮配合就會發(fā)揮出和車一樣的威力，而國際

33、象棋里沒有類似的棋子。</p><p>　　4.  Although Kings cannot leave the Palace, Chinese chess has a special rule that allows them to threaten their counterparts: Kings may not see each other face to face.4、雖然將帥不能離開九宮，

34、但是有一個特殊的規(guī)則使他們能夠威脅到他們的對手：將帥不能直接碰面。根據這一規(guī)則，將帥有時會有和車類似的作用，特別是在殘局階段這種作用表現的比較明顯。</p><p>　　5.  Blocking rules apply to Elephants and Horses.5、象有堵象眼，馬有絆馬腿的規(guī)則。國際象棋并沒有類似的規(guī)則。 </p><p>　　6.  Pawns

35、move one space at a time.6、兵卒每次智能移動一格，并且當兵卒越過河岸到達對方區(qū)域后，它不能像國際象棋一樣變成另一個棋子。</p><p>　　7.  Repetitive situations are relatively simple in Western chess.7、在國際象棋象棋里，重復走一步棋的情況相對較少。但是，在中國象棋里由于更大的棋盤和某些特定的棋子特征，所以

36、就更可能出現雙方都不能退讓的重復的情況。這是就要有要特殊的規(guī)則來解決這些情況。在一些極端情況下，還需要裁判作出裁決。因此，怎么解決重復問題是計算機中國象棋中的一大挑戰(zhàn)。</p><p><b>　　技術簡介：</b></p><p>　　1、Quiescent Search靜態(tài)搜索 </p><p>　　A quiescence search

37、explores all captures and checks from the current position, terminating at quiescent positions, defined as those in which no captures or checks are available. 靜態(tài)搜索探索所有的從當前狀態(tài)和終止狀態(tài)下檢測并獲得的信息，直到局面相對平靜，也就是說沒有任何可以利用的檢測信息為止。

38、</p><p>　　2、 Processing Perpetual Situations處理重復情況 </p><p>　　As we described, perpetual situations have to be identified during the search pro 正如我們所描述的，重復狀態(tài)必須在搜索的過程中不斷被確認。檢測效果所帶來的檢測效果所帶影響經常發(fā)生。

39、如果程序僅僅進行初級的評估，這就可能導致一個不斷試圖殺死將帥的錯誤的棋盤狀態(tài)。</p><p>　　3、Computing And Storing The Results In Advance提前計算并存儲結果 </p><p>　　When the opponent is thinking, the program could make use of this “free time” t

40、o predict the opponent's move and store the results. 當對手思考的時候，程序可以利用這個“空閑時間”來預測對手的下一步的走法并把結果存儲起來。如果預測是正確的，就可以直接提取結果而不需要進行更深層次的計算。 </p><p>　　4、Forward Pruning前裁剪 </p><p>　　This includes ra

41、zoring, futility pruning, and null move. 這包括徒勞修剪，和空移動。</p><p>　　5、啟發(fā)式搜索Move Ordering啟發(fā)式搜索</p><p>　　This includes killer heuristic, history heuristic (Schaeffer, 1983), and internal iterative

42、deepening. 這包括逆向啟發(fā)式，歷史啟發(fā)式（謝弗，1983），和內部迭代深化。</p><p>　　6、Transposition Table置換表 </p><p>　　Store position scores in a hash table for use if the same position is reached again (Marsland, 1986).

43、 存放在在哈希表中某個位置中，如果再次達到相同的位置則調用它（馬斯蘭，1986）。</p><p>　　7、Balance between search and knowledge搜索效率和知識復雜度之間的平衡 </p><p>　　中國象棋和國際象棋程序相似的地方是Chinese-chess programs are like Western-chess programs in tha

44、t they can be divided into two types: knowledge-based and brute-force programs.中國象棋程序和中中它們都可以分為兩種類型：知識型和暴力型。 A knowledge-based program emphasises the evaluation of positions. 知識型程序注重評估函數的價值，暴力型A brute-force program uses

45、the ability to compute quickly and to explore the game tree deeply, so as to discover better moves暴力型程序為了得到一個比較好的走法，往往使用快速的計算能力以盡可能搜索足夠深的層次。即使在國際象棋里，也沒有確定更好的解決之道。中國象棋有更多的不同的棋子和一個比國際象棋更大的棋盤，所以知識型程序被認為有更大</p><p&

46、gt;　　8、Other Speed-Up Methods其他加速方法 </p><p>　　Iterative deepening and aspiration windows (Kaindl, Shams, and Horacek, 1991). Before PVS begins, the program uses a threat-sequence search, in particular when t

47、he game tree is narrow and deep. 迭代深化和渴望窗口（凱因德爾，沙姆斯和霍拉切克，1991）。在原型驗證開始之前，該程序使用優(yōu)化序列搜索，尤其是在博弈樹窄而深得情況下。 It determines whether serial checking may be performed. 它確定是否可以執(zhí)行一系列的檢測。在這樣的樹中，所要考慮的僅僅是是否檢測某方的將帥。和殘局階段相比，中盤很少有可以被執(zhí)行的

48、符合搜索策略的走法。 Thus, more plies can be searched in the middle game. 因此，中盤要搜索更多的層數。</p><p><b>　　Endgame </b></p><p><b>　　9、殘局 </b></p><p>　　In the endgame, few Pa

49、wns remain, so other strong pieces have more space to move. 在殘局階段，僅僅有幾個小卒依然存在，所以其他一些比較強的棋子有了更多的行動空間。 The search depth in this stage is limited owing to the freedom of the pieces and the large size of the Chinese chess

50、board. 在這個階段由于棋子的靈活度和棋盤的廣度 ， 搜索的深度是有限的。因此，人類玩家可以請教專業(yè)棋手以幫助設計這個階段，在這個階段中大部分的中國象棋程序都不如人類玩家。因此，改進殘局階段對擊敗頂級棋手是很關鍵的。 </p><p>　　The earliest retrograde algorithm is proposed by Ströhlein (1970). 最早提出逆行算法的是S

51、tröhlein（1970）。Subsequently Van den Herik and Herschberg (1985) have significantly contributed to the retrograde concept.隨后Van den Herik和Herschberg（1985年），大大促進了逆行的概念。然后 ，湯普森（1986年，1996年）為國際象棋殘局的產生做出了巨大貢獻。他的殘局庫建立在逆行算

52、法使得很多象棋博弈程序棋力增強的基礎之上。此后我們看到納利莫夫的貢獻（2001）。 In recent years, many people have shown interest in developing Chinese-chess endgame databases (Fang, 1996, 2001,  2003; Hsu and Liu, 2002; Wu and Beal, 2001a,b,c</p>

53、<p>　　Rather than being built on retrograde analysis, an endgame database can also be built on massive experts' knowledge (Chen, 1998). 如果不是建立在逆行分析之上，殘局數據庫也可以建立在大量的專家知識上（陳，1998年）。 This kind of endgame databa