Research overview of computer play

Research overview of computer play

Original: Jay Burmeister

Janet Wiles

Department of Computer Science and Psychology, Queensland University, Queensland, Australia

Departments of computer science and psychology the university of queensland, QLD 4072, Australia

Compilation: George Zhong

This translation is purely volunteer. If you have any case, please see the original text: http://www.psy.uq.edu.au/~jay/

(The original text is also added, so this translation is not complete, but will change with the original changes).

In addition, the translator's own work will also go online, please pay attention.

table of Contents

1.0 overview

2.0 Go knowledge 2.1 gas and minute 2.1.1 self-exert 2.2 pieces of chess 2.3 眼 2.3.1 Sin life 2.3.2 Clothing 2.3 Seki 2.4 Robbery 2.5 Take 2.6 Lon 2.6.1 Print 2.7 Contact 2.8 Chess Block 2.9 Chess 2.10 Datches 2.11 Make children and section level

3.0 Go Programming Challenge 3.1 Comparison of Go and Chess 3.2 The complexity of Go, 3.3 Why do Go cannot be programmed like an international chess?

4.0 Computer Go's History 4.1 Academic Cook 4.1.1 Zobrist 4.1.2 Ryder 4.1.3 Reitman and Wilcox 4.1.4 Other 4.2 Programs 4.2.1.1.............................................................. 2 GO4 4.2.2.1 Started with 4.2.2.2 Evaluation Function 4.2.2.3 Performance and Time 4.2.3 Handtalk (Talk) 4.3 Computer Go Competition 4.3.1 Show Cup

5.0 Structure of Go Program

6.0 The performance of the current program 6.1 Between the program 6.2 man-machine confrontation

7.0 Improvement in the future

8.0 Internet on Go and Computer 8.1 Computer Go Resource 8.1.1 Anonymous FTP Archive and Mirror Sites 8.1.2 The Internet Go Server (IGS) 8.1.3 Chess Storage Format 8.1.4 The Computer-Go Mailing List 8.1.5 The Computer-Go Mailing List 8.1.5 Computer Go Ladder 8.1.6 Computer Go Large 8.1.7 Computer Go Encyclopedia 8.1.8 Chess Board 8.2 Go Resource 8.2.1 Chess, Getting Started Guide 8.2.2 Go Getting Started Guide and Demo 8.2.3 The Go News Group 8.2. 4 Go FAQ

9.0 conclusion

Appendix terminology compilation and English control

reference

1.0 overview

This article is intended to introduce the reader to the status of the computer Go and the relevant resources on the Internet.

For the readers who are not familiar with Go, I will introduce the basic common sense and rules of Go, familiar with these readers can skip it, or when necessary.

Part 6 will tell readers, due to their own characteristics, the current level of Go programs is great compared to chess programs, and they need further investment for power development; Part 3 gives the basis for their theory and practice, and pointing out Go. The program is impossible to develop as the chess program (see Table 1).

Part 4 briefly reviews the history of computer Go, and introduces some academic books, computer Go games and results, with some Go programs.

Part 8 introduces Internet resources about Go and computer Go, including online game servers, FTP sites, mailing tables, etc. 2.0 Go knowledge

Go is a kind of two-person chessboard intelligence game without any luck ingredients. It appears in the East 2,500 to 4,000 years ago, it is in Japan, China (including Taiwan) and South Korea (North Korea), like chess in Western countries.

Go's chessboard is a mesh point consisting of some latitude and latitude, usually 19 x 19, called 19-way chessboard, but 9 and 13-way chessboard also have used, chess pieces are black and white two colors, black chess first, play the game , Chess pieces in the grid intersection. Finally, it is a winner that occupies more intersections.

2.1 gas and mix

The empty spot of the chess pieces referred to as the gas of the piece (as shown in Figure 1). The no-angled chess will be lifted from the checkerboard, called a mix (as white in Figure 2). The raid is also called chess, it is a very pleasant thing. Of course, the opponent does not necessarily, in addition to this, the chess pieces falling on the board cannot move.

2.1.1

If the chess is falling, it is in a state of no gas, that is, it is to be subjected to it. This is the chess register is not allowed (X), as shown in FIG. (Allow blocks in the rules)

2.2 block

The colored chess pieces are adjacent to the wire or weft, called a connection. (Figure 4) The chess pieces connected together have formed a chess, of course, a single piece of chess may be called a chess. The gas of the block is the sum of the gas of all the chess pieces, (as shown in Figure 5 L); that is, to mention a piece of chess, must surround it? ⒉ 皇? A href = "# 2.1. 1 "> Self-exerted, because the other's son must be mentioned first, it has the gas. There is a two-eyed piece of lugs that can never be taken, (as shown in Figure 9), because the other party cannot fall into the eyes at the same time, but the film cannot be pulled in a single eye. Self-exertion.

2.3.1 dead

There are two-eyed chess, usually, if even if the other party can't be mentioned, it is active chess; even if it is, it is impossible to avoid it, it is a dead chess.

2.3.2 false eyes

Sometimes it is from two or more chess, then it is necessary to see if the connection between them is real; there is certain one or more of them is currently not a conventional connection point, if these points are The eyes are occupied by the other party, and the eyes of this chess will also occupy each other. If there is at least one other empty connection point that is enclosed, there is at least two empty connection points, the eye is true because the mandrel is always connected.

2.3.3 SEKI

2.4 robbery

In order to avoid endless chess cycles, the chess rules stipulate that the players cannot make the chess game with the same method as the last hand. Usually this situation occurs when it is robust. As shown in Figure 10, the white is falling in 1, and it will be black, and it can be black in 2 because of the white 1 is black. In turn, it can come back in 1 position; this is the case where it is forbidden When the white 1 is raised, the black cannot immediately take 2, and it is necessary to loot in other places (usually the other party needs to pay for it). After the other party should rob, it will be robbed in 2, which is robbery. .

2.5 play

The master can sometimes use the guidance of the chess bureau to develop according to their own deployment. At this time, his starting makes the opponent only the only one.

2.6

The monoson is a method of eating, as shown in Figure 12, no matter how, black chess is destined to be mentioned, because white 1 is playing, black only 2 fled, white and then 3 play, finally this trapeter continues to the board side Black chess is mentioned. 2.6.1 Extended

When the chess pieces are enrolled, they can be admitted to the trapezoidal route of the monk, so that the other party will fail, referred to as an extraguation.

2.7 Contact

The connection between the two pieces in the Go rules refers to the connection on the latitude and latitude, also known. However, there are many ways to connect (see Figure 13) in the actual chess game (see Figure 13), such as, off, two off, pointed, flying, big flying, etc. If the two chess are not liaison, it is called disconnected.

2.8 chess block

Two or more tight contact with each other, a chess block, as A and B, and C1, C2, C3, and C4 in Figure 15. They can be seen as a chess due to mutual contacts. The chess block is the basic unit of the game's player analyzing the chess game. Its most important attribute is whether it is alive, or whether it can make two eyes, or you can contact a live chess block that has already had two eyes.

2.9 chess

Two or more contacted loose hierarchical plaques make up the game, as the AB chess block in Figure 15 forms a game with the C1C2C3C4 chess block. There is no other chess piece between the two chess blocks, which seems to be a large region, but this is not guaranteed; so the conversion of the player in the field is important for the player. The other party invades this region called it.

2.10 Vo

When the bureau is finished, the dead start from the chessboard, the number of the number of the number of the number of the number of the number and the number of the number of spaces, respectively, and a winner.

2.11 Make children and section level

Go has a unique macro and segment level system. The chess hand is divided into 30 levels. After the 10 to 15 games, the player is approximately 20, and it is a segment after level 1, and the amateur segment is 1 to 6, and the occupational segment is from 1 to 9.

With the improvement of the level, the upgrade is getting harder, from 20 to level 10, it takes 1, 2 bureaus every week, then look at some chess books, and to achieve the level of professional segment, it takes many years of professional training. The gap between occupational segments is relatively small, the gap between the first and nine sections is approximately equal to the difference between the amateur segment.

In order to balance the gap between the master and the game between the master, it can be granted first. In the rules of the Stype, the award is a number of steps that will be taken first, and in China / Japan rules, the award is a fixed position. The number of granules is determined according to the gap between the chess, such as the 10-level player can teach 15 players. The principle of its determination is to make the master's winning rate of 50%, or can be granted according to each 10 sub-estimation. If you use a small chessboard, the number of granules is correspondingly reduced. For example, a 9-way 12 is equivalent to a 13-way 12 or 19 road award.

When there is no peep, the first effectiveness is approximately 5 mesh. Under the way, it is generally black. If the two-party level is quite, the next black is to stick white 5. In the formal competition, it is 5. China Rules shall be sticker 2 and 3/4 due to the numerous mid law.

3.0 Challenges for Go Programming

Compared with chess programming, the development of Go programming is still in the primary stage. In the field of programming research in chess, there have been a situation in which powerful research power in the 1960s and 1970s has never been there in the field of Go programming. There is an essence between the two, making the chess are favored by people, and even if there is also power to invest in Go program, the results cannot catch the progress of chess programming. There are several reasons here. On the one hand, due to the complexity of Go itself, on the other hand, the programming technology acquired in the chess field cannot properly be properly applied to the Go.

3.1 Comparison of Go and Chess Chess

1. The type of Go chess is simple, only one type, and there are 6 kinds of chess; but the play of Go's chessboard is 19x19, far greater than the 8x8 chess board of chess.

2. The size of the chessboard and the relative freedom of launch, making the play of the Go chess game about 300, greater than the average 80 steps of chess. 3. The relative freedom is the relative freedom, so that each step of the challenge (branch factor) can be as high as 200, while the exchange of chess steps is only 40 options due to the restrictions of the chess pieces; with the progress of the chess game, the chess pieces are reduced , Make the branch factor decrease, and there is no such advantage in Go. The branch factor also affects the various stages of the chess game (layout, medium tray and official). In the chess, there are many famous systems, usually in 10 steps; and in the Go, the layout is very much, and the depth has less than 3 hand. Of course, there is a modular calendar on the corner angle, but the selected choice still needs to consider the configuration of surrounding chess pieces, etc., it takes certain skills.

4. Go and the Chess Chess Bureau have two ways: the implementation of the Natural Final Office, the final of the chess is relatively easy to discriminate, that is, it will die, and the finals of the Go, is also approved by both parties, but beginners are hard Distinguished whether or not there is no chess, so the chess games they will often be more than the master, and the Go program will also make the same problem. If you calculate the wins in accordance with Japan's approachable method, you will reduce your destination because of the fill in the empty mixture or the mid of the other party.

5. Go and chess chess pieces have a certain threat. For chess, the threat of chess pieces is only the plaid it can come, and in the Go, the chess is no longer moving (only When the subtraction is performed), but can have an impact on other chess pieces on the chessboard, such as an extrajury can change the monolithic relationship of a piece of chess.

6. The chess game of chess changes immediately, and in the Go, this change is not significant, there is a large capacity memory analysis chess game. Even a beginner can also see the results of the monograms after 60 steps, this is a very deep but very narrow algorithm. Only when the chess pieces were taken, the physical state of the board was significantly changed. In Go, the value of 30 destination chess is often taken as an exchange, and the final wins may only have 2 me.

7. In Chess and Go's chess game, chess should usually consider both global strategies and local fighting factors. In chess, due to the quantity and quality of the chess pieces on the chessboard, it is basically not considering the global strategy in the computer chess. In Go, the loss of local campaign is not completely related to the global victory. When starting, the player should consider the play of the game rather than the field. So most Go programs include a chess estimate algorithm, but in addition to actual resolution, this kind of algorithm itself cannot be evaluated.

8. In summary, chess programming can be solved by excellent assessment functions and trimming search tree, and such a method is inong, because the latter's branch factor is too large, there is no suitable assessment function The trim.

9. The thinking of human players in chess and Go is also different. Beginners can only look at a few steps, but in Go, there is a chess shape like a monk, and beginners can not be angry, because this series of starters are playing. In a typical international chess game, experts can calculate 10 steps, but only a few changes per step, in Go, there is a lot of changes in each step, which may also have a deeper, such as It is unable to be well grasped with breadth search and depth search.

10. The horizon effect, that is, after the depth search gets an evaluation, if a few steps are deepened, the assessment will be corrected, such as in the abandoned tactics. The horizon effect only appears in the international chess field, and in the Go, even in the primary player, there are also, for example, the monk.

11. Chess has been used as a research object of human thinking in the 1970s. Chase and Simon show that the expert chess is used by chess pieces; and Reitman1976 research results will be in the opposite, It is a chess piece cross thinking. Determining the liaison and segmentation of the chessia is the key to the death of the chess, REITMAN's research shows that only the relationship between the end of the Finance, the relationship between the chess pieces can clarify, this research shows that the evaluation function may bring the evaluation function in Go programming. problem. 12. In the actual situation of chess and Go, the opponent's level is a major factor affecting the game. Only two chess players in chess can be a considerable chess game, and in Go, due to the presence of the management system, the master's level can be compensated by a decent, in this game, it is always trying to go. Simplify the chess game, keep the advantage, and you have to have a mistake in the other party; Therefore, the level of opponents in many chess games can determine the strategy of players. The chess hand needs to worry about the reasons for some kinds of techniques, which can effectively attack each other's weaknesses. According to the computer Go program, it is a usual practice for someone's first on the bureau. The first bureau of the same person will lose to the computer, and the next few games will win, because he discovers the weakness of the opponent, and the computer program is not Any change.

3.2 Complexity of Go

A limited time soluble algorithm can be applied to the problem solving in the real world, and the algorithm that cannot be solved at a limited time is called an index time algorithm, which can only be applied to a very small problem. The problem soluble in the limited memory space of the computer is called P-space to solve problems.

Studies have shown that the question of the winner is a P-space is difficult to determine the P-space. (Lichtenstein and Sipser, 1980) Another study also involved the issue of robbery to indicate that it is an index time problem for any given Go chess game, determined victory and choice. (Robson, 1983) Therefore, it is necessary to use a fuzzy algorithm when solving some Go questions to achieve approaching the best solution, and simple search will soon lead to difficult.

In allis et al. (1991), the complexity compared to other chess is detailed in detail, and two terms are defined in the text: the solution is solved and completely solved; and demonstrate the complexity and decision complexity of search space Differences. If a chess can get the optimal solution through the program in any chess game, it is called soluble; if this positive solution can be explained by the human language, it is fully solved. Demonstration is done through a play of Go, the model specifies that both parties are turned to fall or give up, take up 181 points, which has the same search space with Go, but as long as the first party will not give up, will definitely Win, so there is no decision complexity.

Allis et al will give a logarithmic indicator of the complexity indicators and decision complexity of the search space complexity, and the two indicators showing Go is the most complicated. And it is not available. For the comparison of Go and Chess, the results show that the complexity of the 9-way chess go is approximate to chess, and the search space is 35 pairs of 50, while the 19-way chess go is much more complicated, so as mentioned above It is not available. Allis et al. 1991).

3.3 Why do Go cannot be programmed like chess?

In the chess program, heuristic search and evaluation techniques are used. The search tree is fixed depth. According to the evaluation of the chess game, the technology is used in chess, because the chessboard is small enough and from essentially Pay attention to local tactics rather than a global strategy.

Assessing the Chess Bureau in Go, I met questions in chess, Go and the latter pay more attention to the global strategy. Go is not like chess, and the play is gradually formed in a long bureau, and there are various direct and interpreted strategies in the bureau to win the victory, such as whereabout Field, expansion outdoor, attacking weakmas, defensive, joining, etc. Because the board is very large, the Go Chess Bureau also contains many localized types. If the chess becomes a battle, then Go is a war. Many local combat should be selected according to the global strategy, so the players must master local and global balances. Due to the various problems brought about by chess bureau, artificial intelligence priority search technology used in chess programming cannot be applied in Go. Go is more research value in artificial intelligence and cognitive science. Hans Berliner front chess world champion, famous American chess, famous chess programming experts have said "... Go will replace Chess as the research object of artificial intelligence" (Berliner, 1978).

4.0 Computer Go's History

About Go Programming Previously appeared in the end of the 1960s, initially purely academic, just a further in-depth study of Go as some research, rather than trying to explore Go's programming theory. Really started to invest in the development of business Go prunes or after the personal computer walks into the family and the computer Go's bonus is set up.

4.1 Academic Work

4.1.1 Zobrist

The earliest Go program is proposed by Albert Zobrist as a part of his pattern identification professional doctoral thesis (Zobrist, 1970) We mainly involve part of the computer Go.

Effect function

Zobrist introduces the effect function to divide the chessboard into black and white geographies. The effect function calculates the value amount of each intersection on the board, the black child is 50, the white pie is value -50 and the blank point is 0; the positive effect point is to give it an adjacent point plus 1, the same negative effect point Plus -1, such algorithms are recursively executed 4 times, and the checker board is finalized (as shown in Figure 17).

The program thus established a 19X19 array to express the status of the chessboard, including: the state of the intersection (black and white or empty), the black and white number of the adjacent point (horizontal, vertical and diagonal direction), chess The number of pieces of the block and the number of meters, the blank point adjacent to the blank point, and the number of empty and chess pieces containing.

The program also uses another effect function to express the effectiveness of the chess pieces, it is similar to the above function, but the initial value of the chess pieces is 100, the positive effect of the point to the increment of the neighborhood 3, while negative effects The point is -4, which is because the computer is black, so the situation of the opponent is estimated to be optimistic.

Choose to start

Establish internal expressions, the program is used to identify the match in its own mode library. The mode library contains a hand and its evaluation. After a match is obtained, it is necessary to use rotation and symmetrical techniques during the matching process. Finally, select the point to get the highest value value to start. Zobrist's procedures are roughly used in the way of accumulating evaluation values.

Zobrist tries to compensate for the weaknesses showing the procedure in the bureau. He divided the chess bureau into four stages: the corner of the layout, the corner of the layout, advanced in the middle, defensive and joining, official son. In order to make the program can have normal response in the chess game, different mode libraries are used at all stages.

Results and improvements

Another improvement of the program is that the restriction calculation is also inspired. In mode identification, the part can be counted three steps to make the program can master the lone and eating chess, connect, and cut, and make eye.

Zobrist's program performance is general, it defeats two beginners, but the skilled players are poor.

4.1.2 Ryder

Jon Ryder (Ryder, 1971) is the in-depth of Zobrist Research, which uses effect functions and accumulated evaluation. However, Ryder has added global strategies and local tactics, and enhances the depth of the calculation, and Ryder considers local tactics in the short-term target and uses a global strategy in the long-term target.

Since it is not practical to use a full search tree in Go, it is necessary to have a method, and Ryder's program uses many combination analysis methods such as local tactical analysis. Ryder regards Go to the field, the play-in-the-style balance between the game, and he puts these three formulas to maintain such a balance: the board organization displays the control of the board area, the situation is judged to choose the best direction of both sides. , Chess block analysis to determine its safety.

Start to choose

The first layer determines the local tactical state of all blocks, identifying the board, chess, and analyzes the geographical fields of both sides and the threat to the other chess block. Finally, 15 can be selected from all possible to prepare the next layer.

The second layer uses local tactics and simple global strategic theories to analyze 15 starts from the first layer, add two analytical values ​​to the highest value. If you can determine a "obvious" good game, the second layer analysis can be simply terminated.

Ryder's program uses a firing search tree using a forward trimming method, applying different scenarios to determine if there is a need to track a branch of the tree, which means that a certain hand is worth considering. Terminate the search; the program includes two eyes, and whether the escape is 5 gas or make the block chess. The risk of forward trimness is that the best launch may be ignored before being found.

Effect function

As Zobrist, Ryder also uses an effect function to provide quantification of each board. His effect function is similar to Zobrist, is also a black scales, and the white part is incorporated, and the value of a certain effect is accumulated by the effect of its neighborhood. The effect function of Ryder is simpler, and the propagation factor is fixed (Figure 18)

Ryder effect function

Estimation of a certain point is estimated or too low, it will bring problems to the judgment. It is estimated that excessive high will make the overall focus, and therefore ignore the overall situation, and it is estimated that it may be grasped by the opponent. The optimal attachment should be a point where the effect is estimated to estimate the peak of the curve. Ryder regards the effect of the chess pieces as the local properties of the chess, although he also admits that the chess pieces may have an effect on the side of the distance from the distance, such as an extrajug. In this case where this effect function cannot be accurately expressed, local tactical assessment is used to compensate for the shortcomings of the effect function.

The liaison is a link that the effect function cannot be assessed well. Ryder defines two terms: contacts and strong associations. At some point, it means that this point is the non-dead child, or that there is at least one egg-neighboring child and there is no blank point of the other. A extension of a fully contacted is a semi-container, refers to a blank point having at least two oriental orientation and at least one other adjacent point. Strong relationship means that a point is connected by someone who is connected by someone, and a point is connected to some of the diagonal and a total of at least one blank point (tip), and a blank point between a two son (off) this Two situations are also considered to be strong. The chess block is composed of a set of contacts or semi-contacted points.

The effect function portion is used to determine the play. The game is composed of a set of contacts, and its effect function is not smaller than the program predefined threshold, and is divided into strong and vulnerable.

The local tactical concepts used in the program include numerous gas, robbery, monograms, eyes and false eyes. Ryder believes that local tactical capabilities in the program need to improve, he called "Dodging".

Variable description

Variable description requires the following information: Basic objects need to be used in calculations, often require multiple access variables that need to be calculated in the calculation.

The chess moment is expressed by the pointer. It maintains a pointer to the playback of the game it is to each point; and each checker has a pointer to some point, called the base point, and other points are searched by the base point.

The global description of the chessboard is a 21x21 array (19x19 chessboard plus two borders), including the following information: area pointer, vulnerable pointer, strong pointer, chess block pointer, block chess pointer (black, white or empty).

Results and improvements

The result of Ryder is just a game, the opponent is a beginner who has just learned the chess and basic local and global skills. As the main aspect of Ryder Research, the conclusions that he publishes it in tactical analysis, and the search technique is already completely selected with the level of beginners, and only a slight improvement can improve the local tactical level of the program. The further improvement solution proposed by Ryder includes the reuse of the initial information, an entry that issues and the weak definitions, and Ryder pointed out that the player stage analysis of 10% time in the process is worth noting. The result of the Ryder program is not ideal, which is also its reason for the opening, the medium disk and officials do not have a clear distinction, so in particular in the official bureaucratic local tactics, Ryder believes that the future improvements should be strengthened .

4.1.3 REITMAN AND WILCOX

Walter Reitman and Bruce Wilcox began as research objects in 1972, and the procedures for research were named Reitman-Wilcox and InterIM.2, respectively. After the study is interrupted, Wilcox rewrites the InterIM.2 program and produces a commercial Go program Nemesis.

Reitman and Wilcox Applications Beginning in Go Go is focused on artificial fields, including pattern identification (Reitman et al., 1978; Reitman and Wilcox, 1975, 1978), Planning (Reitman and Wilcox, 1974) and humans in Go. It feels reproduced with the cognitive process. In fact, the purpose of the program is to play with the human chess player and recording the results by the master to discover the thinking characteristics of human thinking.

The three basic aspects of the Reitman and Wilcox play are cognitive, knowledge and synthesis (Wilcox, 1988). They try to join the description in the program to obtain a cognitive level, identify and store two types of Googin knowledge: the decision knowledge includes how to eat chess, how to circumvent, including the identification of the chess block And related quantitative analysis, etc .; through appropriate control structure integrated procedures, you can choose to start.

The Interim.2 program is established and maintained a selection update, a multi-layer network description, Reitman and Wilcox introduced this is analoged by skilled players. Experts and commentators for choosing the analysis is hierarchical, and the calculation is not involved in the whole, but the target is driven.

Interim.2 program

InterIM.2 adopts summary-feedback technology (Wilcox, 1988). The information contains low-level objects such as block chess, summarized and used in high-level descriptions such as chess blocks, and programs are selected in high-level objects to make decisions, and decision results feed back and correct low-level objects.

Flow Control

The program describes the chess game in a structure called GameMap, which includes a set of array variables called Gameboards. In each Gameboard array, there are pointers to various special data structures, such as String-Board contains pointers to all blocks on the board. The Gameboards variable is simple and complex: Typeboard, Stringboard, Linkboard. Special Game-Boards variables such as Lensboard, WebOard, Sectorboard and TacticsBoard are designed to describe the skilled human player's feelings and cognitive processes.

The flow control of the InterIM.2 program is switched between Move and Reflex, and when the false Move selection updates the GameMap's affected component, Reflex is responsible for selecting the hand in the rank.

Tactics

The tactical part of the program, that is, ProBe is a calculation of the target-driven circuit for solving a particular problem. Probe is based on such assumptions, that is, if a given problem, a set of intelligent selection logic returns the same result, and it can be considered that this is a positive solution. Such assumptions are not always effective, but Reitman and Wilcox think this is the small cost that is given to accelerates search, and the loss caused by human causes of misunderstandings cannot be listed on the same day. Probe is a process design of the simulated human go chess for selective calculation (Reitman and Wilcox, 1979) .probe can be used to solve a specific problem and give a proper start, you can get a hypothesis The behind the board situation. ProBe continues to calculate the next hand based on this assumption, know the result of the problem. If the result is a successful Probe returns to the initial advice, otherwise the overturn assumptions are recalculated, in non-limiting calculations, to traverse all the positive solutions that start. However, in Probe is not all, each of them is calculated, and each failure is limited to 2 to 3 times to return the result.

The choice and inspection of advice is the idea chain (Wilcox, 1988) that forms, evaluated and implemented, including a variety of expert knowledge, such as the target and sub-target, it is recommended to start The goal, the current chess game and the target, use all the goals and sub-objectives. Expert knowledge contains many special Go knowledge. The control structure implemented in Probe in Probe makes various expert knowledge, which is easy to test and expand (Wilcox, 1988). The direction of the goal-directed behaviour of Probe directs the search in a particular situation can be done to 60-80 steps (Wilcox, 1988).

Cognition and description

InterIM.2 maintains a lot of data structures when describing chess games. Point status (black, white or empty) is used to determine block chess, and a basic attribute of loft chess is dead. InterIM.2 can identify tips, off, two off, Xiaofei, Damai and other contacts, contact and divide into ordinary contact, may contact and die. The basic attributes of the contact include whether they can be connected or cut, and whether to intersect each other, and establish a liaison reason, it can also identify special contacts within the four lines of the disk. The boundary line is a long-distance contact for chess-form assessment and hand choice, and several chess that cannot be cut can be considered to be a piece of chess shape, and the mutual contact is called chess block, and liaison with chess, block. And the area containing the empty point and the counterpartum surrounded by the disk is referred to as a closed area. The motivation to start is also recorded to enhance the level of choice and facilitate debugging.

The effect is based on the Ryder algorithm, REITMAN and WILCOX have found that the effects of certain circumstances may be misleading, so it improves the concept of the effect. There is no contact information in the same value of the large block, and the boundary is formed by shrinkage of the effect, and the attributes of the chess pieces and their connection are not considered during this process, so the final result of the region may be curved. Or stripped. The independent regional structure finally merges to become a effect block, the more in the effect block, the more possible, it is possible to become a field.

In Interim.2, Reitman and Wilcox established a maintenance chess game description, that is, cognitive and data structures are sometimes referred to as cognitive expression, and the two cognitive elements used in InterIM.2 programs are particularly interesting. One is grid, The area formed by the effect of the chess effect is determined, the other is the lens, and the actual chess shape in the chess game is determined by the mode library of the program itself.

The grid is a number of gas circles built around the chess block, and the highest number of 9 floors will be done outwards, so that the other chess pieces, no matter where it is. The contacts between the chess pieces or the linked pieces and disk (up to 3 points) can also terminate the grid. The outermost grid point constitutes the boundary, where the grid boundary comes from the identified possible attack, where you can avoid the attack, the location of your own chess block, how to disconnect contacts.

Reitman and Wilcox believe that local chess shape and chess order are a basic aspect of Go, excellent chess hands master a variety of ordinary chess-shaped chess, such as contact offensive and defense. The lens is designed to use a mode library such as. The lens includes several domains, each domain contains a certain order, and its valuation, providing information and global information of information and tactics that are recommended. When a domain recognizes a chess shape, it is recommended to start and related information to the lens, otherwise it is ignored to save time, which makes the lens can be focused.

performance

The three pairs of Interim.2 programs were published in Reitman & Wilcox (1978). Two of them defeated 22 and 34 players. One game lost to a 4-level player, which is calculated The interim.2 program has a level 27.

Reitman and Wilcox were very satisfied with the first simulation of human game for InterIM.2 programs. It has a more conservative, defensive than attack, it defensive very well, in addition to the case where the program is wrong, a few chess are simultaneously attacked.

4.1.4 Other Academic Works

Some academic works using Go as objects are listed, which are all levels, generally not providing details about Go programming.

Lehner

Paul Lehner is a student of Wilcox, and he has obtained his Ph.D. in the Go's Strategic Planning Field (Lehner, 1981), his work is that Wilcox has found this-based alternative to this in Go player (Kerwin and Wilcox, 1973) ).

Alternative search can provide its strategy through one of many possible order. Lehner's conclusion shows that replacement search is not only enough to assess strategic planning, but also quickly eliminate inappropriate planning, he found that the application replacement search can reduce the number of search layers that must be in the precise computing system.

STOUTAMIRE

David Stoutamire (1991) Research machine learning with Go object, is a combination of hand choosing instead of using Go Tactics / Strategic Knowledge through expert knowledge base. STOUTAMIRE continues to develop a model-preferred classification technology to automatically obtain a good pattern expression; because the mode library is exponentially growing, a remedy has been developed during the increase in memory demand.

Sander

To avoid the use of search tree evaluation and selecting an index explosion brought by using, Peter Sander (1979) designs a program that chooses to start with the top or target-oriented organizational structure. Sander's program named KYU, created a summary organizational structure that changed the progress of the game progressed, and simplifies the strategy into the target, and the hand is selected. KYU can only go to 25 hands from the opening, and there is no subsequent calculation logic.

KYU's data structure is based on liaison records of block chess, chess blocks, and chess motives. You can access the block, chess blocks, and chess motives you belong or affected from any point on the board. After each checker, the data structure changes accordingly, this change is from the lower to the block, from the block to the group, from the group.

The structure cited in KYU is SANDER to integrate the multi-layer structure data expression and the hierarchical organization planning system, which combines strategies with the target, and he also believes that such a structure can also be applied to various objects other than Go.

Friedenbach

Kenneth Friedenbach research sensation and cognition in Go-like objects in solving interactions in complex issues (Friedenbach, 1988), which is based on chart discussion, abstract and applied to Go. He uses Go as a research object because of its philosophy and multi-layer organization.

Friedenbach's feeling - the cognitive model includes three parts, first, including five-layer static organization: chess object, similar block, related block, loose group (potential) and global attribute summary; second is due to the progress of the chess Analysis of the analysis of structural changes; finally a comprehensive consideration of various analyzes.

It is necessary to perform four types of analysis, tactical safety analysis runs in block chess, involving the attack and defense of block chess; strategic security analysis involves the attack and defense of chess blocks, including eye position and liaison analysis; strategic development analysis in the championship, involving Open area, equilibrium measurement, new chess block, and chess block mergers; global equilibrium analysis is the highest level of analysis, design chess game overall strategy. Friedenbach specifies a analysis of three conditions: start condition, termination conditions, and continuing conditions. Taking tactical analysis as an example, its start condition is a few hundred chess, and the termination conditions are at least three gas; its continued conditions include adding new chess pieces that eat measures and defensive side, in the analysis process If a hand has led to a change in the number of meters, you need to re-analyze.

4.2 Program

The following is given below for the details of the details available in the current Go program, others will be listed in Chapter 6.

4.2.1 The Many Faces of Go GO

The Many Faces of Go (MFG) is one of the best commercial Go programs. David Fotland started using spare time programming Go and published MFG in 1990 in 1981. Fotland has also continued to improve MFG and publish the topic on computer Go Programming and MFG in the computer dialogue list. MFG has developed into two famous procedures: G2 and COSMOS.

4.2.1.1 G2

The first Go program of Fotland uses the radiation effect function to determine the boundary. Unfortunately, it is easily defeated by a simple gas collection algorithm, so he has designed G2. The internal structure of G2 is divided into data structures, tactical analysis, and adopt (Fotland, 1986). G2 uses a full consideration to assess the effective start and focus on strategic deployment, and finally select the highest valuation.

data structure

Most of the code in G2 is used to update the data structure, the chess pieces (weigh) and organize a lug (weighing group) to finally form a chess block (called large group). The data is associated with each point on the disk, including the distance, neighboring point, and diagonal point list, neighboring point, and diagonal point list, block chess pointer, numerous number thereof, adjacent to pieces color (black, white or mixed), effect functions and respective The nearest neighboring point in the direction.

The same color chessia in the vertical and horizontal direction constitutes a block chess, block chess includes a color, a gas number, a list, an adjacent list, an adjacent other block chess list, a large group and its degree of freedom (hereinafter described).

G2 identification contacts include single-gate, two-level, tip and small fly, contact data including its own block, color, location, and state.

Chess blocks include indiscrunable contacts, or block chess with each other's dead chess, chess block data includes its composition block chess, other dead chess, number and gas number, number of eyes, and freedom.

Tactical analysis

Tactical analysis evaluates the position of the chessboard to give the choice to start, each chess block has a degree of freedom and makes a block of chess. G2's chess is mainly determined by tactical analysis and Fotland also spends the greatest energy (Fotland 1986). Since most of the time spent in tactical analysis, the trimming of the search needs to be fast and effective.

The degree of freedom is obtained by a multi-layer process: can be trouched by a plug-in, which is a dead; identifying an eye bit; Identifying an indivisible contact; liaisonless block identifying a chess block; by boundary identifier identifying boundary; The dead actuator gives the degree.

Tactical analysis Check chess each five gas is determined to be dead, and there is a piece of loft or above. Recommend interest in the calculation to determine the block status. Interesting attacks, including long-standing, interesting defensive, including penetrating, single, long, connection, etc.

Interestingly started to enlighten the valuation, the highest value is selected as further analysis, as a hypothesis, as a counterpart, the overall perspective. Similarly, it is also possible to use the best to use the backtrack and maximum - minimum process trim at each step. Therefore, it is necessary to establish a deep search tree until the chess block is determined to live, the search tree can reach up to 80 hands, so that the monographer can accurately calculate. "Best" The number (1 to 3) depends on the depth of the current start in the search tree: there are fewer leaves near the roots. Since the value of the termination node is only two (eating or escape), the largest - minimum process trim is very effective (Fotland, 1996). Eye analyzer is used to maintain information about the eye, half-eye (one hand) and the eye (two hands can be eye-catching), etc. G2 can identify some basic dead and basic ophthalmoids such as straight three, four, plum blossom, Ding 4, Qu four.

The chess block is composed of a block chess that cannot be divided into, and the tactical analyzer tries to cut off its contact to test its strength. The dead chess can also be used as a playful connection.

Boundary analyzer determines the boundary. G2 can determine the closed area and the incomplete closed area close to the disk. On the side, the chess pieces and disk or contacts can constitute a region, such a chess shape generally appears within the four lines of the disk.

The dead actuator analyzes each chess block. The freedom of the chess block is extended by its gas, the eye position, the controlled area, extends along the disk, and whether it is in contact with the opponent's chess, whether it is completely surrounded. The degree of freedom itself has 20 levels including "two eyes", "big eyes can be two eyes", "unspecified", "no eye" and "unable to do two eyes".

Latch

Use the strategic evaluator containing 65 rules to select the two-step valuation. The first step first evaluates the strategy, the second step is to evaluate all the hands. Information of data structure and tactical analysis is used in these two steps to assess chess bureaus. Fotland feels that the weakness of the data structure makes G2 are also weaknesses in this regard (Fotland, 1986).

Those rules can provide strategic starting as in the catering, do Shimari or Kakari, Joseki, expand, insert, in contact, combat, segmentation, cut, contact, attack, robbery.

In the first step, you should evaluate all recommended strategies. Assume that the fidumo is evaluated the results of the strategic player, the evaluation process includes the assignment of the inventory and returns it. The valuation of the inventory is determined by the freedom of controlling its block. If the strategic game, the chess game is not satisfied with the strategic goal, it is ignored.

In the second step, all the startings that have been taken in the first step are further analyzed. Its evaluation value is determined by the degree of freedom, size, and quantity of each piece of chess in the district.

The hands considering in two steps, its estimated, the highest score is selected as the next hand.

Size, performance and time limit

In 1981 and 1988, Fotland spent nearly four years of spare time to write G2, because the computer's speed could not satisfy the stronger Go program needs, he terminates the work on G2 (Fotland, 1996). G2 is approximately 11,000 line C code takes up about 700K code and data memory. During this time, Fotland's Go level is enhanced from 15 to 1. G2 roughly is 25-20 segments, and you can overcome beginners (Fotland, 1986). Fotland is also successful in computer Go games, G21997, in the Bay World's Coats, the first place in the US computer Go Championship.

4.2.1.2 Cosmos

Fotland converted G2 into Cosmos in 1988. The main difference between Cosmos and G2 was that instead of evaluating all legal moves, only suggested moves were evaluated. This transition was achieved by increasing the number of move suggestors to over 250. Quiescent search was added and The Connection, Eye, And Territory Evaluations Were Made More Accurate (Fotland, 1996) .data Structures

The data structures used in Comos were basically the same as those used in G2. The eye information included the number of eyes achievable in gote, the number of eyes achievable in sente, the number of eyes achievable if the opponent moved twice, a list of Vital Points, And EiE Type. The Information Contained in The Data Structures WAS Updated Either Incrementally (EG, LISTS OF LIBERTIES) OR AFTER EVERY MOVE (EG, INFLUENCE).

Tactician

Essentially, Cosmos had the same tactician as G2 but better move generation and sorting. In Cosmos, the tactician was only used to determine dead and threatened strings by trying to capture them. Its parameters were maximum liberty count, maximum, depth, and maximum search size. The maximum liberty count was 4, and therefore as for G2, strings with more than 4 liberties were assumed to avoid capture.The maximum depth allowed was 100. The search size determined the playing level (Stoutamire, 1991) and since forcing moves WERE NOT COULD BE ACCURATELY Read Even At Low Playing Levels.

Influnce function

Territory was determined by radiating positive influence from alive groups and negative influence from dead groups. Radiated influence did not pass through stones or connections and was inversely proportional to distance.

Board evataration

Life and death of groups was the primary concern dealt with by the evaluation process. Board positions were evaluated by a similar process to that used in G2 with the addition of quiescent search.The tactician was used to determine dead and threatened strings (those that could .................. ..

Eyes were analysed to determine their potential and then allocated to groups. False eyes were identified by checking the diagonals of eyes and some dead shapes were also known by Cosmos. Any group with enough eyespace to make two eyes was considered to be alive. Potential eye Space Could Be Gained by Extensions Along The Edge, Possible Connections, Being Adjacent To A Threatened Enemy Group, And by Controlling Territory Which Was Not Already Considered to Be An Eye.

There were 25 values ​​that described a groups strength in terms of its life and death status. The values ​​were divided into five main categories which included very alive, alive, unsettled, weak and dead. Determining the strength of a group included considering potential connections, potential eyes, and potential extensions along the edge. Positive influence was radiated from alive groups and negative influence was radiated from dead groups with black and white influence being maintained separately.

Weak groups which were contacted by the influence from friendly alive groups were considered not to be surrounded. Weak groups were further divided into those that could run and fight and those which would almost certainly die.

In general, board points were scored according to the radiated influence values. However, occupied board points, board points adjacent to a stone and board points between a stone and an edge were scored differently. Unsettled groups were scored according to who as to play next .MOVE SELECTION

Cosmos had over 250 rules that suggested moves which included fuseki moves (including shimari, kakari and joseki moves), edge moves, playing in the centre, playing safe when ahead of opponent, "squirming" around when behind opponent, pattern matching, saving weak friendly groups (including making eyes, running, or fighting semeais), killing weak enemy groups, cutting, connecting, contact fights, ko threats, and filling dame. Cosmos had a joseki library which contained around 5000 suggested joseki moves and a pattern database which had 60 patterns which were 5 x 5 in size. Whenever a match was made, the rule (code) associated with the pattern was applied (executed) and would then suggest a move. The addition of extra moves to be suggested could easily be accomplished .

The rules would supply a guess value (probable evaluation of the move), bonus value, minimum aliveness value, and would indicate which groups were being attacked or defended if any. The moves were sorted by their guess values ​​and, depending on playing level, a certain number were "played" Moves could be rejected before being evaluated if they did not "apply" A move applied if it accomplished what it was intended to do:.. if a move was intended to defend a group and the group ended up weaker than it started the move was rejected. Surviving moves were awarded a sente bonus if they achieved sente. The position resulting from the move was then evaluated. The evaluation score, the sente bonus and the rule bonus were summed together and the highest scoring move WAS SELECTED As the next move.size, Performance and Timeline

It took Fotland nine months to compress G2 down to 512 k and add a new user interface. In 1988 Cosmos was released by Ishi Press for IBM-PC as Cosmos, The Computer Go Partner and was released a second time in 1989. Due to the amount of time spent on the user interface, Fotland was unable to devote sufficient time to improving Cosmos' playing level and its performance in various world computer Go championships was modest coming 8th in 1988 and 7th in 1989 (Fotland, 1996).

4.2.1.3 Many Faces of GO

After a rewrite of the user interface and addition of professional graphics and new features, Cosmos was released as The Many Faces of Go in 1990. Additions to MFG included a limited full board lookahead capability, consideration of the value of sente, and a strategy function which is used to focus attention on the important areas of the board and to identify urgent moves. The number of patterns in both the joseki and pattern databases were increased. The pattern database was also improved by increasing the pattern size from 5x5 to 8x8, by storing many suggested moves for each pattern in a move tree (rather than just one per pattern as for Cosmos), and by a complete rewrite which encompassed algorithms, code, and data structures and lead to an increase in speed (Fotland, 1996). Data Structures

The dynamic data describing a board position is stored in three classes of data structures:. Incremental, locally recalculated, and globally recalculated The data in the incremental data structures is updated upon addition or removal of a stone and includes low level information such as lists of empty points, number of liberties etc. Locally recalculated data is updated as it is needed ie, it is only updated for regions of the board which have been affected by a particular move or move sequence, and includes connection strength, and eyes. Globally recalculated Data IS Updated for the Whole Board and Includes Group Strength And Influnce. The Locally and Global Recalculate Data Structures Are Maintained by The Evaluation Function.

The dynamic data is stored globally and is generated progressively in several passes of increasing abstraction. The dynamic data used in MFG includes point environment, string data, connection data, eye data, potential eye data, group data, territory, and score.An instance of the connection data structure stores information pertaining to connectivity between two strings. Each connection has a list connection points ie, a liberty of one string which is no more than three points away from the other string. Thus, the connection recognised by MFG include hane , one point jump (ikken tobi), knights move (kogeima), two point jump (nikken tobi), large knights move (ogeima), three point jump, and bent three point. Other special connections near the board edge are handled through the pattern database. The data stored in the connection data structure includes the strings involved in the connection (s), the number of and lists of one, two and three point connections (ie, as measured from the co nnection points), the type of each connection, and the strength of each connection (eg, already cut, cuttable, shared connection, connected with aji, connected solidly). Other than the type and strength data which is stored in a locally recalculated data Structure, The Connection Data is Stored In An Incremental Data Structure.

The eye data structure records information pertaining to either a block of empty point or dead enemy strings which are partially or completely enclosed by friendly stones. Data stored in the eye data structure includes colour, a list of the points in the eye, vital points, And Eye Type (EG, One Point, Two Point, Line Eye, Big Eye, And Dead Group Eye). The Number of Eyes (To A Resolution of 1/8 of An EyeVable Under Various Conditions " eyes if opponent moves twice, number of eyes if opponent moves next, and number of eyes if program moves next. A board point can only be recorded as belonging to one eye. The eye data structure is a locally recalculated data structure since eye data does not lend itself to incremental update. In hindsight, Fotland says that a pattern based approach to recognizing eyes would be preferable (Fotland, 1993) .The evaluation of group strength and the generation of attacking and defending moves involves identifying pot ential eye space and running points. The potential eye data structure contains type, value and location data for potential eyes. An examples of data stored in the potential eye data structure is "extend along edge" (type), value is number of new points of eye space, and location is the liberty to extend from. running points are stored by type (eg, running towards friendly / unfriendly stones) in several different lists. Potential eyes and running points are stored in a globally recalculated data structure.

Tactician

Every string with three or less liberties and many strings with four liberties are read by the tactician. Each string is read twice, once with White moving first and once with Black moving first. The tactician determines whether a string is captured (ie, can not live even if it moves first), threatened (ie, it lives if it moves first and dies if it moves second), or stable (ie, lives regardless of who moves next). The tactician relies on simple heuristics concerned with the number of liberties and connectivity; pattern matching is not used in the tactician.The tactician has two separate move generators; one to generate attacking moves and one to generate defensive moves The moves suggested by the move generators are sorted according to criteria which include second order liberties. , Cut Points, An Alpha-Beta Depth-First Search Is Employed with The Performance of The Search Depending on The Quality of The Move Sorting (Fotland, CGM1 5 Mar. 1993) .

Tactical searches are goal directed and are limited to a maximum number of nodes. For string captures this is around 100, however, when only one move is suggested, it is not counted towards the node limit. In this way, ladders can be read without problems (Fotland, cgm1 14 Oct. 1993). The number of nodes for a search are allocated to the branches according to the value given to the moves by the first ply move generator and thus different branches may end at different depths. The branching factor AT Each Successive Ply Is Progressiveness Constrained by The Tactician. The Branching Firl Fifts from Five At The First Ply To One or Two by The Fifth Ply.

Since the data structures in MFG are separated into incremental, locally recalculated and globally recalculated, low level local tactical searches can be executed quickly by only updating the incremental data structures. However, the possible goals of such a search are limited since the high level data structures are not recalculated. It is not possible, for example, for MFG to search for x liberties and two eyes (Fotland, cgm1 3 Nov. 1993) .The tactician is used to read connections and eyes. A cutting stone can be examined by the tactician to determine whether or not it will be captured. Stones on the diagonal of eyes can be examined to determine whether they can be captured. These types of search do not exceed around 12 ply since the program actually plays worse if more plies are considered (Fotland, 1996).

MFG uses caching to reduce the amount of time spent reading. Since strings are read twice, tactical results related to eyes, capture or threat, and cuts are cached rather than re-evaluated. Life and death analysis are cached as trees since they are smaller THAN The Trees Which Would Result from Tactical Searches (Fotland, CGM4 16 Aug. 1994).

Influnce function

The influence function radiates influence which is inversely proportional to distance. Influence is radiated to a fixed maximum distance of nine although when lower playing levels are chosen, the distance is smaller. Stones and connection impede the radiation of influence. Black and White influence are radiated separately (ie, they are not summed together to give a single value for each board point) so that their relative values ​​can be compared. Dead stones radiate negative influence and thus, since Black and White influence is maintained separately, dead stones reduce their own influence rather than increase their opponents influence. The initial influence value radiated from a stone depends on its strength.Influence is not used to determine group connectivity. The use of influence in MFG includes determining thickness and territory, identifying surrounded groups, and generating moyo building OR Reducing Moves. Another Way In Which Influnce Is Used Is To Determine A Group's Running ability. A group's ability to run towards a liberty depends on the relative values ​​of friendly and unfriendly influence nearby. The sum of a group's ability to run toward each of its liberties determines its running ability whilst the gradient of the influence determines the direction in which IT Should Run.

Evaluation

Board positions are evaluated in a multiple pass process The evaluation function consists of many components which include the tactical analyser and evaluators for connections, eyes, group strength, and territory A score for the position being evaluated is achieved by assigning a value between.. - 50 and 50 to each board point to indicate the level of Black or White control exerted on them. The overall score for the board position is the sum of these values ​​for each board point.In order to try and overcome the difficulties associated with the horizon effect, Fotland chose to use quiescent search rather than modify the evaluation function. in quiescent search, positions are evaluated when they become quiet rather than when they are in a state of flux (eg, such as in the middle of a piece exchange in Chess). Positions Which Are Not Quiet Are Characterianed by Oscillations in The Evaluation Values ​​Returned for Successive Moves Which Makes It Diffult To Make Good Decisions Based on The evaluations. MFG uses quiescent search as part of the evaluation function for full board evaluation to overcome this problem by either calling itself recursively (ie, generating further moves) until a position becomes quiet or by using patterns suggested as obvious local answers to make the position QUIET. MFG WILL GENERATE UP TO SIX PLIES OF MOVES IN A Quiescent Search Which Improves Its Performance By Up to Four Stones (Fotland, CGM1 30 Sep. 1993).

Move Suggestion

Moves are suggested by a rule-based expert system and include fuseki (eg shimari, kakari, and joseki moves), moves on the edge (eg, invasions and extensions), playing in the centre, conservative play when ahead, risky play when behind , pattern matching (eg, cutting and connecting, surrounding and escaping, invasion and defence, endgame, killing and saving groups, and shape-based moves), group defensive moves (eg, making eyes, running, and fighting semeais), contact fights (eg, blocking, extending for liberties, and hane), ko threats, and dame.A limited form of full board lookahead is achieved by storing move sequences in the pattern and joseki databases. Move sequences attached to patterns in the pattern and joseki databases are played onto the board and evaluated at their endpoints. Thus if a particular joseki is suggested as appropriate, the stones in that sequence are played and the resulting board position is evaluated. This enables MFG to "play" to the end of a lookahead SEQUENCE AND minimax the score returned by the evulting board position.

Pattern Database

Each pattern in the pattern database has a a move tree associated with it. The pattern database contains around 1200 patterns of size 8x8 and around 6900 suggested moves with the average number of moves stored for each pattern being about 5.

Each point in a pattern is marked as black, white, empty, white or empty, black or empty, black or white, or anything (ie, do not care). Patterns are categorised according to whether they are in the middle of the board OR WHETHER THEY Contain Edges and / Or Corners.

A successful match of a pattern depends on both matching the stones in the pattern with those on the board and matching any attributes associated with the pattern. Each pattern has as many as 10 attributes which pertain to stones and include whether they were dead or alive, minimum number of liberties, and connectivity between stones.To match over a 1000 patterns on a 19x19 board in 16 different orientations requires around 3 million primitive operations. Since performance is a big issue, the patterns are compiled into a bit array which facilitates fast bit -parallel comparisons. By using an 8-bit hash function and bit-wise operations, the number of primitive operations is reduced to around 350000. This number of operations still takes enough time to make it too expensive to match all patterns more than once per Move.

Patterns from the pattern database are evaluated after being played on the board in their entirety ie, they are evaluated at the endpoints of their move trees. A small subset of patterns are used to read local sequences based on shape. Fotland has recently begun evaluating the Stregth of One Point Jump Connection Using Patterns (Fotland, 1996).

A fast pattern matcher is necessary for a strong Go program according to Fotland. Pattern matching can be used to suggest and sort moves, and to classify connections and eyes. Reading is employed to determine whether eyes are false. Pattern matching on its own is not Adequate for Life and Death and Group Stregth Determination (Fotland, CGM1 8 Apr. 1993). The Eye and Connection Patterns Are Hand Coded Decision Trees.

One method that Fotland tried to increase the number of patterns in the pattern database was to take them from professional games.When replaying a professional game, if MFG did not consider the move played by a professional or the move played was low on the suggestion list, Fotland would manually cut-and-paste an 8x8 section from the game into the pattern database using a self-written graphics pattern database editor. Fotland discontinued this method since most of the missed moves were tactical (ie, based on read-out not shape) or depended on a larger context than the 8x8 section that was placed in the pattern database. Fotland found that the cut-and-paste approach to patterns was more profitable from even games played on the IGS (Fotland, cgm1 13 Jan. 1994) .joseki Database

The joseki database contains around 45000 moves. Joseki sequences are stored as directed acyclic graphs (ie, they contain no loops) having empty corners at their roots. Moves in a joseki sequence can be one of several types including urgent, complex, trick, and bad. The urgent move have priority whilst the complex and trick moves are only played if MFG is behind. The bad moves are not for the program to play, rather they are used by the program to know how to respond to bad moves played by its Opponent.

Joseki patterns are evaluated by being played on the board in their entirety and evaluating the resulting board position. MFG selects joseki sequences that are appropriate given the status of the other corners (Fotland, 1993).

STRATEGY

A strategy function is used to focus attention on important areas of the board before each move (Fotland, 1993). The dynamic data is examined to determine what phase the game is in, the relative score, the value of taking sente, and whether there are any urgent offensive or defensive moves that need to be made.Urgent offensive and defensive moves are examined initially and if sufficiently urgent are played without any further considerations. Urgent moves can be suggested by the pattern matcher or by examination of groups which are determined to need urgent attention. Friendly groups which need urgent defence include big groups and cutting groups. Opposition groups which need urgent attack include big groups, cutting groups, and groups adjacent to friendly groups needing urgent defence.

Move suggestion is affected by the phase that the game is in. Certain rules will only fire in certain phases of the game. The fuseki phase lasts until joseki sequences have been pursued in all the corners. The middle game lasts until at least move 120 and .

The types of moves suggested are also affected by the relative score. MFG plays more conservatively when it is far ahead of its opponent and makes unsound invasions when it is far behind its opponent. The relative score is classified into many states including "way ahead" (Over 40 Points), "AHEAD (20 - 40 Points)," About Even "," BEHIND ", and" Way BEHIND ".

Move Selection

Since the evaluation function is very slow (Fotland, cgm4 16 Aug. 1994), a limit is placed on the number of full board evaluations which can be performed each move depending on the playing level. MFG can only look at 5 to 10 moves in detail (depending on playing level) before selecting the move it finally plays. A safeguard against errors in the evaluation function is that only reasonable moves are suggested for full board evaluation (Fotland, 1993) .By employing the strategy function, the move suggestion expert System, and the Joseki and Pattern Databases, MFG Fully Examines A Small Number of Moves, Playing The One Which Receives The Highest Score from the evAolization function.

The score returned from a full board evaluation is also modified by the addition of sente if applicable. Thus if MFG plays a sente move, the value of sente is added to the evaluation for the resulting position. The value of sente varies from 7 points during .

Size, Performance and Timeline

With some help from Ishi Press and a professional artist, Fotland rewrote Cosmos' user interface and released it as MFG in 1990. Until the present time (July 1996), MFG has had 3 releases with various improvements. MFG contains around 40000 lines of C Code With Around Another 20000 LINES of User Interface Code.

MFG has performed well in various world computer Go championships including 4th in 1993, 2nd in 1994, and 3rd in 1995. According to Fotland, MFG gains about one stone in strength each year (Fotland, 1996). MFG plays on both the IGS and the NNGS as ManyFaces. MFG participates in the NNGS rating scheme and was rated 9 kyu (as at July 1996). It was also awarded an 8 kyu diploma from the Nihon Ki-in in Japan based on its performance against a 1 dan amateur at 9 Stones and A Sample Play Against A Professional In September 1995 (Fotland, 1996). 4.2.2 GO4

Micheal Reiss started writing Go programs in 1983 and his current program, Go4 , has evolved from various predecessors over that time. Reiss has met with some success in computer Go competitions and Go4 came 2nd in both the 1st and 2nd FOST cups.

Reiss' programming philosophy is to use simple algorithms on a large amount of data rather than complex algorithms on a small amount of data Go4 calculates everything from first principles;. Complex concepts are calculated from simpler concepts which are in turn functions of even simpler concepts. Thus, Go4 does not have an elaborate rule-based expert system at its core. A positive consequence of this approach is that Reiss' relatively weak ranking (2 kyu) compared to other top Go programmers is not a deficit. It also means that Go4 usually responds well to the strange moves (by human standards) that are quite often played by computer Go programs and which are difficult to foresee when designing a rule-based system. The major drawback of this approach is that it is computation intensive and requires a Powerful Computer.

At the foundation of Go4 is a massive amount of connectivity data from which almost everything else is eventually derived. Go4 selects a move by first generating about 50 candidate moves that are analyzed by an evaluation function which estimates territory. The move scoring the highest evaluation is Played as the next move. thus, go4 uses 1 Ply Global Lookahead Alth Tactical Lookahead and search IS Used in evatating each move. 4.2.2.1 Candidate Move Generation

Candidate moves are generated by a process of pattern matching to determine the best 50 moves to examine in depth. Go4 has around 15 high level patterns which include terms for the probability of an eye, safety, and territorial value. Whenever a match is made, the board point (s) suggested as good places to play have an appropriate value added to them. The 50 highest scoring board points are then examined by the evaluation function. The board point which receives the highest score from the evaluation function is played as the Next Move.

In some circumstances, Go4 effectively evaluates more than 50 candidate moves per turn. Pattern matching is not applied to regions of the board where the stone safety (see Stone Safety in section 4.2.2.2) and territory (see Territory in section 4.2.2.2) do not differ from those of the previous move. Thus after evaluating the 50 highest scoring board points after pattern matching, the next move is selected from the combined set of evaluation scores of the 50 board points just considered and the evaluation scores of any board points In The Undurbed Regions Which WERE Considered During The Selection of The Previous Move.

4.2.2.2 Evaluation Function

The evaluation function is a six step process: 1. a connectivity probability map is generated, 2. groups are determined, 3. eye space is determined from which individual eyes are identified, 4. the safety of each stone is determined, 5. a radiation function combines the stone safety data and connectivity probability map to radiate safety to empty points, and 6. territory is determined. Reiss describes the evaluation function as pessimistic since whenever it is applicable, the evaluation function considers that the opponent is the next to move .Connectivity probability map

Each time a hypothetical move is played on the board, a connectivity probability map for all friendly and opponent stones (including the hypothetical stone) is generated and stored. The connectivity probability map for a stone contains the probability of connecting to nearby friendly stones already on The Board or To An Empty Point if it is occupied by a friendly stone. stones and empty points which can be reached directly (nobi), by a one point jump (ikken-Tobi Link), A One Point Jump (Ikken-Tobi Link) Two Point Jump (Nikken-Tobi Link), A Small Knight's Move (Kogeima Link), or a Large Knight's Move (OGEIMA LINK) Are Examined (Total of 32 Possible Points).

The probability of connecting two stones is calculated empirically by playing out the sequences of moves needed to determine whether the stones can be connected or whether the opponent can cut them. This process requires the use of lookahead, particularly when ladders are generated by cutting moves ( eg, cutting a kogeima link generates two ladders, one for each cutting point). If all possible cutting stones are captured, then the probability of connection is 100%. However, if the cutting stones can not be captured, a hand-tuned algorithm estimates the connection probability.A massive amount of data is generated in calculating the probability map for each stone on the board for every hypothetical stone being evaluated. The process is very computationally expensive and requires a powerful computer (Go4 currently runs on a Pentium-Pro 200).

Group determination

Stones and strings are catalogued into groups:.. Any stones which are 100% connected are considered groups A tactical search on all strings with less than 4 liberties is used to determine which strings are definitely dead Dead strings are not included as part of any group .

Eye Identification

Identifying Eyes IS Achieved In A Two Step Process: 1. Eye Space Is Determined, and 2. Individual Eyes Are Identified from The Eye Space.

A floodfill type algorithm is used to identify contiguous regions in which opponent connectivity probability is low. Points in those regions whose 8 neighbouring points are mostly clear of opponent connectivity probability are considered to be eye space.

Individual eyes within a groups eye space are identified by shape. There are several grades of eyes with the grade being a function of the opponent's connectivity map. Eyes with a grade above a certain level (see Stone Safety below) are considered to be true eyes .Stone safety

The safety of each group is determined by the number of true eyes it has The safety process is repeated 5 times;. Each successive pass is more pessimistic about which points will eventually become true eyes (ie, the threshold is raised for selecting which grade of Eyes WILL EVENTUALLY BECOME TRUE EYES). The final safty value for a group is the average of it 5 pases.

RADIATION FUNCTION

The safety value for each stone is radiated to the nearby empty points in proportion to the connectivity probability map for that stone The inverse of a stones safety. (100% - safety) is radiated in the same manner and contributes to the opponents radiated value ie .

Territory

Black and white territory is estimated from the Radiation Values ​​of the Empty Points. The Difference Between Black and White Territory Is Returned As The Evaluation of The Hypothetical Move for Which It Was Generated.

4.2.2.3 Performance and Timeline

Reiss' Go skill has progressed from 10kyu in 1987 to 4 kyu in 1992 to 2 kyu in 1996. Reiss has programmed full-time on Go4 for about the last year (1996) and also programmed full-time for one and a half years around 1986. Other than for these two periods since starting Go programming in 1983, Reiss has programmed part-time with the most effort usually expended in the two months leading up to a competition.There have been many commercial releases of Reiss' programs by Oxford Softworks WITH Another Soon to Follow. NOTABLE Releases Have Been Go Player Professional and an Updated Windows Version, Go Professional II, Released in Japan.

Reiss characterized Go4 as the weakest of the current strong programs at tactical situations requiring life and death analysis although it is good at creating moyos and gaining territory in the middle and endgame. Since Go4 tends to play well in the centre of the board, Prof. CHEN PLAYED 200 Games Against Go4 in An Attempt to Test Modifications to Handtalk Which Were Designed At Playing Better in The Center.

According to Reiss, other weaknesses are the absence of global lookahead, lack of a concept of "shape", and the reluctance of Go4 to invade. Since the evaluation function pessimistically considers that the opponent moves next during tactical searches, Go4 does not usually evaluate the likelihood of invading stones forming eyes as being favourable. to compensate for not invading well, Go4 always plays it first move at the 3,3 point and also tries to live in as many places as possible to stop its opponent creating moyos early.

Reiss uses many empirically hand-tuned algorithms within Go4 although no learning algorithms are used. Carefully crafted changes to these algorithms can improve the strength of Go4 eg, by making the connection algorithm more accurate, eye and territory evaluation also become more accurate, resulting in an increase in playing strength. Before the 2nd FOST Cup in 1996, Reiss tuned the performance of Go4 by playing 700 test games against Handtalk, resulting in Handtalk's only defeat during that competition.4.2.3 Handtalk

Handtalk has been one of the strongest programs in recent years and is written by Zhixing Chen, a Chinese professor from the University of Guangzhou. Although Handtalk becomes stronger each year and was awarded a 4 kyu diploma by the Nihon Kiin in 1996, Prof. Chen Does Not Believe That Go Program Will Reach Shodan by 2000 IF SHODAN IS TAKEN AT ITS International Level (2 - 3 Dan Japanese).

Details of the architecture and algorithms used by Prof. Chen are not abundantly available. Some of the details which have emerged from discussions that various people have had with Prof.Chen at various competitions are provided below.

Handtalk is written completely in IBM-PC assembly language and is very small (around 250k) and very fast. Only a very small number (around 4) candidate moves are evaluated using a static evaluation function which seems to be accurate and stable. Handtalk also Uses An Influnce Function Whose Influnce Decreases AS 1/2 ^ Distance.

Handtalk does not use very much global search. Pattern matching is used to some extent although it is unclear whether the patterns are contained in a database or if the pattern matching is rule-based. Handtalk does not use joseki much and has only about 10% as many josekis as MFG.According to Prof. Chen, Handtalk's strong points are its fighting, group safety estimation, ability to recognize life and death, the end game, and board evaluation. Other people have also commented on its strength at estimating a group's running ability, sense of shape, the middle game, and winning semeais. Prof. Chen characterizes its weaknesses as surrounding, ko, and attacking territory. Recent changes include fighting less often than in previous versions and an improvement in gaining territory in the centre which Was achieved by Testing Alterations in Around 200 Games Played AgaInst Go4 .

4.3 COMPUTER Go Competitions

There are various computer Go competitions throughout the world including the United States Computer Go Championship, the North American Computer Go Championship, the Computer Go European Championship, and the International Go Contest. There is also a computer Go competition conducted as a part of the Computer Olympiads helod in england.

% A cash prize of around $ 6000 is offered to the winning program in the World Computer Go Congress as well as the chance of challenging dan ranked junior opponents to win the Ing prize The challenging program plays 3 junior players. (Ages 11 - 13) which Arena Ranked Around 3 to 6 DAN.

4.3.1 The ING Prize

% Acer Incorporated and the Ing Chang-Ki Wei-Chi (Go) Educational Foundation have been joint sponsors of the International Computer Go Congress since 1985 (Shih, 1989) and also sponsored the First Computer Olympiad in 1989. The three top contestants from the North American; European; and Japan and Asia-Pacific region are invited to the International Go Congress in Tapei A prize of approximately $ 1.5 million US will be won by the programmer (s) of the first program to beat a nominated human competitor without handicap. ON A 19x19 Board. This Prize Is Known as The ING Prize.5.0 Structure of Go Program

6.0 Performance of the current program

6.1 confrontation between procedures

6.2 human-machine confrontation

The strongest Go program is approximately 8-13 points to people. Because people can adjust chess wind to the weakness of the program, it is very difficult to give a program level, and a program may be only 5 in the initial pair. Level players have a 4 child, but after a few games will increase to 8 to 9.

In 1994, the World Computer Go Champion GO Intellect loses three games in the bureau of 14 sub-bureaus on the young players. Because the rules allow for the son to put it itself, this means that 15 sizes are black ( Computer) first 16 hand, then leave it. In 1991, the World Computer Champion Goliaath was full of 14 sub-bureau in the three games in the 15-child game.

7.0 Improvement in the future

8.0 Go and Computer Go Resource on the Internet

(This chapter is taken from Accessing Go and Computer Go Resources on the Internet Posted "The Proceedings of The Second Games Programming Workshop in Japan" Many chapters selected from the README file of BSDServer.ucsf.edu/go, due to the update of the site, can not guarantee it. All parts can be accessed)

Individuals of researchers and amateur design in Computer Go can benefit from Internet resources, which is mainly used by researchers in computer play.

8.1 Computer Go-related Internet Resources

8.1.1 Anonymous FTP Archive and Mirror Sites

The main Go FTP Archive site and its mirror are freely providing relevant Go and computer Go data in anonymous FTP. The README file under the Go directory generally contains a brief introduction to all its files. Here are some sub-directory names and their presentation.

AGA US Go Organization

Clients Internet Go Server Customer

CoMP computer Go program

Games annotation chess

IGS IGS Chess Bureau in PostScript or Smart-GO format

INFO ordinary text file information

MGT MGT program, readable smart-go file

Printing for printed chess games

PRO unbounded occupational pair

Problems Go problem, mainly about Smart-GO format

PROG and Go-related procedures, but not the game's COMP directory includes:

Amiga, Macintosh, IBM PC (DOS and Windows), Linux, And X Windows utility

Archives of computer dialog list

Computer Go Encyclopedia

Computer Go Character and Program

Various reports on Go programs and computer Go, articles and papers

Main site BSDServer.ucsf.edu/go

Mirror site ftp.pasteur.fr/pub/go rzis1.rz.tu-bs.de/pub/go igs.nuri.net/go

8.1.2 Internet Go Server (IGS)

The Internet Go Server (IGS) provides a means for people to play real-time interactive games of Go against other people or against Go programs over the internet. The IGS serves as a virtual Go club which is particularly useful for players who do not have Local access to other go players or a go club.

First time users can login as "guest". Full access can be gained by registering for an account. Information about registering can be obtained by typing "help register". The help command provides information on how to use the interface. Once logged on to The IGS, You CAN Play A Game Against Another Player, Watch Games In Progress, or Even Comment On Games in Progress.

Interfacing to the IGS CAN Be Difficult and As A Result, Client Programs Have Been Written for Various Platforms E.G. Atari St, Amiga, Macintosh, Next, MS Windows, IBM PC, UNIX (ASCII) and X11.

Client Programs Available in The Clients Directory: BSDServer.ucsf.edu/go/clients/

Help on Client Programs: www.well.com/user/mmcadams/igs.howto.html (FOLLOW The "getting software" LINK)

Help files on using the IGS: bsdserver.usf.edu/go/igs/igs_ug.ps.z bsdserver.ucsf.edu/go/igs/igs_ug.txt.z bsdserver.ucsf.edu/go/igs/igshelp.z Www.well.com/user/mmcadams/igs.howto.html ltiwww.epfl.ch/~warkent/gO/etiquette/et.html

Telnet Connection to the IGS: Igs.Nuri.net:6969 Hellspark.Wharton.upenn.edu:6969

8.1.3 Game Record Formats

Several game record formats exist although currently the two most commonly used are Smart Go Format (SGF) and Ishi "Standard" Format. Information, executables, converters etc. related to game record formats can be found in the info, mgt and prog directories of The ftp archive seats.bsdserver.usf.edu/gO/Info/ bsdserver.ucsf.edu/gO/mgt/ bsdserver.ucsf.edu/gO/Prog/ www.cwi.nl/-jansteen/go/games/sgf/ SGF.html

8.1.4 List of Computer Go Mail

Computer Go Mail list specifically for programming Go and related papers, e-mail to this address will forward to its member.

Many individuals engaged in research or programming in the computer Go are discussed, and their questions from abstraction to specific, which is a good place to inspire thinking.

The list of computer play mailing can be added to Computer-go-Request@comlab.oxford.ac.uk by sending E-mail to the following address

The past information archivebsdserver.ucsf.edu:/go/comp/ (compGo-mail - *. Z)

Current information archiveftp.comlab.ox.ac.uk:/pub/documents/computer-go/mail-archive

8.1.5 The Computer Go Ladder

Discussions on the computer-go mailing list resulted in the establishment of an informal competition between computer Go programs called the Computer Go Ladder. The primary purpose of the ladder is for the enjoyment of the participants with the secondary purposes of stimulating both interest and work on Computer Go Programs and providing a Way of Measuring The State-of-The-Art of current programs.

There are HTTP links from the Computer Go Ladder page to pages related to some of the programmers and some of the programs involved in the Computer Go Ladder, world computer Go competition results, the IGS and various other associated pages on the Internet. The Computer Go Ladder is maintained by Eric Pettersen (Pett@cgl.ucsf.edu).

CGL.ucsf.edu/go/Ladder.html

8.1.6 Computer Go game and results

Nobi.Ethz.ch/martin/cgresults.html (Martin M 黮 LER) www.reiss.demon.co.uk/webgo/gores.htm (Michael Reiss)

8.1.7 Computer Go Encyclopedia

There are several different versions of computer Go Daquan Igs.nuri.Net/go/comp/compbibs.z Nobi.Ethz.ch/martin/references.html (Martin M 黮 LER) www.reiss.demon.co .uk / Webgo / Bib.htm (Michael Reiss) www.psy.uq.edu.au/~jay/go/comp-go.bib.html (Jay Burmester) 8.1.8 Chess Library

IGS stores chess board libraries in PostScript and Smart Go format. Jan Van Der Steen can query thousands of chess music in conducting a GoBase.

Igs.nuri.Net/go/igsarch/ (IGS Archive) BSDServer.ucsf.edu/go/games/ bsdserver.ucsf.edu/go/pro/Search5D.html (5 Dan IGS) Games) www.cwi.nl/gobase.html (Go Base) www.geocities.com/siliconvalley/park/4872/index.html

8.2 Internet Go Related Resources

8.2.1 Chess and Tutorial

There are various versions of chess and regulations on the Internet.

BSDServer.ucsf.edu/go/rules bsdserver.ucsf.edu/go/rules.sg bsdserver.usf.edu/gO/aga/aga.rules.z bsdserver.usf.edu/go/aga/ingrules.zip BSDServer. Ucsf.edu/go/igs/goe.rules.z bsdserver.ucsf.edu/go/igs/goe.Igs.z

8.2.2 Go Studying Guidance

There are some learning guidance and chess and chess introductions for beginners. .

BSDServer.ucsf.edu/go/rules.ps.z www.cwi.nl/~jansteen/go/rules/human.html www.cs.vu.nl/~willems/go.html ltiwww.epfl.ch/~ WARKENT / Go / Rules / Rules.html LTIWW.EPFL.CH/~wARKENT/GO/Rules/9x9Game/1.html (ASCII) Hyperg.iicm.tu-graz.ac.at/gtl (Teaching Ladder Page) Section 2 Of this document

8.2.3 Go Newsing Group

The Go Newsing Group is a good discovery, sharing the resources of the information about Go, games and IGS, which is not mainly for computer Go, but there are also some relevant discussions and information.

Rec.games.go

8.2.4 The Go Frequently Asked Questions (FAQ)

The Go Faq Contains Questions That Are Frequently Posted to the Rec.games.go NewsGroup and incduDes Sections on Basic Go, The IGS, And Computer Go Related Issues.

BSDServer.usf.edu/gO/faq rtfm.mit.edu/pub/Usenet/news.edu/pub/Usenet/news.edu/pub/Usenet/news.edu/pub/Usenet/news.edu/pub/Usenet/news.Answers/games/go-faq www.cs.utexas.edu/users/orb/go/gofaq.html 9.0 Conclusion

Appendix Universal Go Terms and English Control

The following Go terms and English control are prepared from Fletch Holmquist, Bill Taylor adaptation. .

ATARI: An Immediate THREAT TO CAPTURE; A SINGLE LIBERTY Remains. A Verbal Warning IS Offen Issue PLACING An OPPONENT INTO ATE.DAME (useless): a neutral point, territory for neither; a liberty.

DAN: Advanced Grade.

FUSEKI: The Opening Moves of the Game Where Influnce and Territory Outlines Are FORMED. (Litrally: `no stones'

Gote: Defensive Play, Loss Of Initiative. (Literally: 'Lower Hand'.)

Hoshi: (`Star Point '), 4-4 Point.

Ikken-Tobi: One Point Extension.

Joseki (Established Stones): KNown Sequences of Moves Near The Corner Which Result in Near Equal Positions for White and Black.

Keima: Knight's move extension.

KO: REPETITIVE CAPTURE. (Literally: `eTernity ')

Ko Threat: Intervening Move (That One Hopes Will Force A Reply) Before A Ko Can Be Recaptured.

Komi: Score Adjustment USUALLY PENALIZING Black for Playing First. OFTEN 5.5 Points.

Kosumi: a Diagonal Play Next To One's Own Stone.

KYU: Learner Grade.

Nobi (Stretch): An Extension Away from an Opponent's Tsuke, Cross-Cut, ETC.

OGEIMA (Large Knight's Move): Three Across And One Vertical (or Vice Versa).

SEKI: a Situation WHERE Neither Player May Place The Other In Ate With PLACING HIMSELF IN ATE. Stalemate, with no territory awarded.

SENTE: Threat Forcing Direct Response, Creates Initiative. The Right to Choose Where To Play Next. Opposite To Gote. (Litrally: `Upper hand".)

Shicho: ladder play.

Shicho-atari: ladder Breaker. A Stone Played In The Path of a potential shicho, threatening to make it it.

Tobi: JUMP.

Wei Chi: The Chinese name for Go. (Literally: "Game of Encirclement")

reference

References

(The following is the part that is not available in General Go Bibliography)

Berliner, H. J. A Chronology of Computer Chesces and ITS Litrate. Artificial Intelligence, 10, Pages 201 - 214, 1978.Fotland, D. The Program G2. Computer Go 1, 1986.

FoTland, D. Knowledge representation in the Many Faces of Go. Available on The i NTERNET AT ftp://bsdserver.ucsf.edu/gO/Comp/mfg.gz, 1993.

Fotland, D. Personal Communications, 1996.

Stoutamire, D. Machine Learning, Game Play and Go. Technical Report TR 91-128, Case Western Reserve University, Cleveland, OHIO, 1991.

Computer-go mailing list references (cgm) Fotland, D. E-mail communication on the computer-go mailing list. Mailing list archive available on the Internet at ftp://bsdserver.ucsf.edu/Go/comp/ cgm1 compgo- Mail-1.z CGM4 CompGo-Mail-4.Z