CN105817029B - Mixed search algorithm based on road draw in chess type in six sub- chess game playing systems - Google Patents
Mixed search algorithm based on road draw in chess type in six sub- chess game playing systems Download PDFInfo
- Publication number
- CN105817029B CN105817029B CN201610145217.0A CN201610145217A CN105817029B CN 105817029 B CN105817029 B CN 105817029B CN 201610145217 A CN201610145217 A CN 201610145217A CN 105817029 B CN105817029 B CN 105817029B
- Authority
- CN
- China
- Prior art keywords
- valuation
- chess
- value
- node
- road
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
- 238000010845 search algorithm Methods 0.000 title claims abstract description 11
- 238000011156 evaluation Methods 0.000 claims description 18
- 238000004088 simulation Methods 0.000 claims description 13
- 101100129500 Caenorhabditis elegans max-2 gene Proteins 0.000 claims description 6
- 230000005484 gravity Effects 0.000 claims description 5
- 101100083446 Danio rerio plekhh1 gene Proteins 0.000 claims description 3
- 230000015572 biosynthetic process Effects 0.000 claims description 3
- 238000003786 synthesis reaction Methods 0.000 claims description 3
- 238000004519 manufacturing process Methods 0.000 claims 1
- 238000013138 pruning Methods 0.000 abstract description 13
- 238000000034 method Methods 0.000 description 11
- 241000209202 Bromus secalinus Species 0.000 description 8
- 230000008569 process Effects 0.000 description 6
- 230000005059 dormancy Effects 0.000 description 5
- 230000007958 sleep Effects 0.000 description 4
- 230000008859 change Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 241000243251 Hydra Species 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- QRXWMOHMRWLFEY-UHFFFAOYSA-N isoniazide Chemical compound NNC(=O)C1=CC=NC=C1 QRXWMOHMRWLFEY-UHFFFAOYSA-N 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 239000011800 void material Substances 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Classifications
-
- A—HUMAN NECESSITIES
- A63—SPORTS; GAMES; AMUSEMENTS
- A63F—CARD, BOARD, OR ROULETTE GAMES; INDOOR GAMES USING SMALL MOVING PLAYING BODIES; VIDEO GAMES; GAMES NOT OTHERWISE PROVIDED FOR
- A63F11/00—Game accessories of general use, e.g. score counters, boxes
- A63F11/0074—Game concepts, rules or strategies
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/22—Indexing; Data structures therefor; Storage structures
- G06F16/2228—Indexing structures
- G06F16/2246—Trees, e.g. B+trees
-
- A—HUMAN NECESSITIES
- A63—SPORTS; GAMES; AMUSEMENTS
- A63F—CARD, BOARD, OR ROULETTE GAMES; INDOOR GAMES USING SMALL MOVING PLAYING BODIES; VIDEO GAMES; GAMES NOT OTHERWISE PROVIDED FOR
- A63F11/00—Game accessories of general use, e.g. score counters, boxes
- A63F11/0074—Game concepts, rules or strategies
- A63F2011/0086—Rules
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Software Systems (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses the mixed search algorithms based on road draw in chess type in a kind of six sub- chess game playing systems, the present invention is to extend the intermediate node stage when scanning for game theory using Alpha-Beta pruning algorithms to have used the valuation mode based on " road " to the valuation of both candidate nodes, is carrying out having used the valuation mode based on " chess type " when valuation to leaf node.Both valuation modes are used in mixed way, are applied in Alpha-Beta pruning search, in conjunction with the advantages of the two.Hybrid Search mode with it is single based on the search pattern on " road " compared with, game level is improved in the identical situation of search efficiency, with it is single based on the search pattern of " chess type " compared with, improve search efficiency in the case where identical game is horizontal.
Description
Technical field
The invention belongs to the research field of chess game game playing by machine, in particular to it is based in a kind of six sub- chess game playing systems
The mixed search algorithm of road draw in chess type.
Background technique
Game playing by machine is one of Major research field of artificial intelligence, is exactly to realize man-machine battle or machine using computer
Machine battle, enables a computer to apish thought process.Game theory till now, is at home and abroad rapidly developed from beginning,
Many scholars and expert contribute in game field, and once supercomputer Hydra triumph Britain Chinese chess champion Michael is sub-
When this.Therefore, the final goal that game level is game is improved before the deadline.But due to by hardware system and six
The limitation of sub- chess complexity itself, it is horizontal to improve game, only from the aspect of the search of game theory and valuation two.
Six sub- chess English name connect6 derive from quintet game, are by our Taiwan National Chiao Tung University information engineering departments
Wu Yicheng is taught to propose.Because six sub- chesses are compared with quintet game, fairness is up to the present had more, simple in rule, playing method is complicated
The features such as.Become within 2006 International Olympic event, becomes the championship of Chinese University Students game playing by machine for the first time within 2007
Event.Six sub- chess laws of the game are first next chess pieces of black, and then every square wheel flows down two chess pieces, is first linked to be six company persons
For victory, do not prohibit hand limitation, is draw in chess if chessboard does not decide the winner still below.But sometimes can also be separately plus regular, when draw in chess,
By judgement formed five even number come deciding, to reduce the situation of draw in chess.The complexity of six sub- chesses is embodied in 19x19
Chessboard and judgement to chess type cause game theory branching factor excessive there are 361 points on chessboard；Chess type has 11 kinds, including
Six connect, and living five, it sleeps five, living four, sleeps four, living three, dim three, sleep three, living two, sleep two and single, wherein every kind of chess type is divided into again
A variety of situations, therefore its state space complexity can achieve 10 172 powers, complexity is equivalent to go；Chess game simultaneously
It is not only more, but also change more complicated.So the characteristic and difficulty of six sub- chesses itself, more attract countless fans to study,
It wherein mainly include two large divisions, valuation strategy and searching algorithm, since there are difficulty, valuation strategy is largely using quiet at present
State valuation, searching algorithm use minimax algorithm and AlphaBeta pruning algorithms.These result in game-tree search efficiency
Low, valuation is not accurate enough, to influence game playing by machine level.
The chessboard of six sub- chesses is bigger for other chess kinds, and node is relatively more, and the valuation analysis of chess game is also compared
It is complicated.The game playing algorithm for being usually applied to six sub- chesses is Alpha-Beta pruning algorithms, derives from minimax algorithm, although
Beta pruning has been used, has improved search efficiency to a certain extent, but game horizon effect is still unobvious.It is carried out when to game theory
The number of nodes of the algorithm search can increase with the incremental exponentially type of search depth when search, and search efficiency can be substantially reduced,
For its reason other than the exponential increase of node, a very big factor to affect is first obtained by way to get there generator in expanding node
All legal ways to get there are obtained, then all ways to get there are carried out with the mode of valuation, that is, is present in pseudo-code GetNMove and obtains expanding node
In function.Two kinds of valuation modes are primarily present, one is the valuation based on road, another kind is the valuation based on chess type, is respectively had excellent
Disadvantage.In search technique, used at present is still single way of search, therefore Alpha-Beta searching algorithm cannot be same
When reach the high-caliber degree of high efficiency.
Therefore, it in six current Zi Qi game softs, is used when using AlphaBeta pruning algorithms search game tree
Be all single valuation mode search, i.e., what is used intermediate node and leaf node be all similarly to estimate in search process
It value mode or is all based on " road " or being all based on " chess type ".It can be obtained by being completed to test: if only with
The single Alpha-Beta pruning search based on " road ", search efficiency wants high, but chess game valuation accuracy is not high；If only
Using the single Alpha-Beta pruning search based on " chess type ", valuation is more accurate, but search efficiency is by the very day of one's doom
System.So regardless of using which kind of mode, it all can Shortcomings, it is therefore desirable to which this is optimized.
Summary of the invention
The purpose of the present invention is in view of the drawbacks of the prior art, provide in six sub- chess game playing systems of one kind based on road draw in chess type
Mixed search algorithm.
To achieve the goals above, the invention adopts the following technical scheme: in a kind of six sub- chess game playing systems based on road and
The mixed search algorithm of chess type, comprising the following steps:
Step 1: setting nodal information, width value, depth value, Aapha value, Beta value and player；
Step 2: judging whether it is leaf node, if it is leaf node, thens follow the steps six, otherwise executes step 3；
Step 3: expanding node obtains and needs the node that is extended, using being based on the mode on " road " to needing to be extended
Node carries out valuation, and chooses number of nodes according to the width of setting, and the node as extension returns；
Step 4: begun according to search result simulation；
Step 5: by depth minus one, Alpha value and Beta value are exchanged, and changes player, then executes step 1；
Step 6: using based on valuation is carried out to leaf node by the way of " chess type ", the father of present node is then returned to
Node；
Step 7: judge present node depth whether be setting depth, i.e., whether be root node, saved when traverse piece
When the child node of point, search terminates.
The step 3 is based on " road " mode and carries out the evaluation function that valuation uses following formula:
Wherein, TotalScore: total valuation of current situation is represented；NumberOfMyRode: represent current situation we
The number on every road；ScoreOfMyRode: the value on our every road is represented；NumberOfThreatRode: current chess game pair is represented
The number on the road Fang Ge；ScoreOfThreatRode: the value of other side Ge Lu is represented；PosValueOfNode: current way to get there is represented
In two positional values begun on chessboard.
The step 3 is based on " chess type " mode and carries out the evaluation function that valuation uses following formula:
TotalScore=ScoreOfMy+ScoreOfThreat+ScoreOfExtra+W (2)
ScoreOfMy=FirstStoneOfMy+SecondStoneOfMy+AddPowOfMy (3)
ScoreOfThreat=FirstStoneOfThreat+SecondStoneOfThreat+Add PowOfThreat (4)
Wherein, ScoreOfMy: the valuation of our current situation is represented；ScoreOfThreat: it represents other side and we is produced
Raw threat value；ScoreOfExtra: additional synthesis valuation is represented；W: the unbalance factor of specific gravity is represented；FirstStoneOfMy:
Represent our first valuation to begin；SecondStoneOfMy: our second valuation to begin is represented；AddPowOfMy:
Represent the added value of our two relating headings of beginning；FirstStoneOfThreat: representative simulation other side first begins pair
The threat value just generated；SecondStoneOfThreat: the representative simulation other side second threat value begun to our generation；
AddPowOfThreat: the threat value that we is generated on representative simulation other side's relating heading.
The setting of the unbalance factor W of specific gravity uses following formula:
W=W1+W2 (5)
K1=max1/min1 (6)
K2=max2/min2 (7)
Wherein: W1: representing our first valuation that begins (FirstStoneOfMy) and first valuation that begins of other side
(FirstStoneOfThreat) difference generated when unbalance；W2: our second valuation that begins (SecondStoneOfMy) is represented
The difference generated when unbalance with second valuation that begins (SecondStoneOfThreat) of other side；K1: it represents first chess piece and estimates
The unbalance factor of value；K2: the unbalance factor of second chess piece valuation is represented；Max1 and min1: we and other side the is respectively represented
The big person and small person of the valuation of one chess piece；Max2 and min2: it is big with the valuation of second chess piece of other side to respectively represent us
Person and small person.
During Alpha-Beta pruning search, the valuation of node and the estimating of leaf node to be extended for centre
Value may exist four kinds of hybrid modes:
(1) twice all using the mode for being based on " road ".
(2) twice all using the mode for being based on " chess type ".
(3) " chess type " first is used, it is rear to use " road ".
(4) " road " first is used, it is rear to use " chess type ".
For first way, situation valuation lacks accuracy, causes to search for not accurate enough.Although second way valuation
Accurately, but search efficiency is very low, and depth can receive when being 4, with regard to unacceptable when depth is 6.For the third mode
For, even more worthless interleaved mode.Finally one is interleaved mode of the present invention, " road " and " chess type " are combined
Search efficiency and searching accuracy double dominant, search depth can be to 10, or even to 20, width also be can be set more
Greatly, such search efficiency is guaranteed, and game level significantly improves.
The present invention is that the intermediate node stage pair is extended when scanning for using Alpha-Beta pruning algorithms to game theory
The valuation of both candidate nodes has used the valuation mode based on " road ", has been used when carrying out valuation to leaf node based on " chess type "
Valuation mode.Both valuation modes are used in mixed way, are applied in Alpha-Beta pruning search, in conjunction with the excellent of the two
Point.The valuation mode based on road and the valuation mode based on chess type have been subjected to order by merging use in entire search process,
The advantages of each comfortable corresponding search phase shows oneself, balanced search efficiency and game are horizontal on the whole.Hybrid Search
Mode with it is single based on the search pattern on " road " compared with, improved in the identical situation of search efficiency game level, with list
One search pattern based on " chess type " is compared, and improves search efficiency in the case where identical game is horizontal.
Finally under Ubuntu system, using QTcreator as developing instrument, designed six sub- chess game playing system software.
And the mixing valuation mode based on road and based on chess type is applied in the system, improve the search efficiency and game water of system
It is flat.
Detailed description of the invention
Fig. 1 is the flow chart of mixed search algorithm.
Specific embodiment
Technical solution for a better understanding of the present invention carries out more detailed retouch to this programme below in conjunction with attached drawing
It states.
Step 1: setting nodal information, width value, depth value, Aapha value, Beta value and player；
Step 2: judging whether it is leaf node, if it is leaf node, thens follow the steps six, otherwise executes step 3；
Step 3: expanding node obtains and needs the node that is extended, using being based on the mode on " road " to needing to be extended
Node carries out valuation, and chooses number of nodes according to the width of setting, and the node as extension returns；Estimated based on " road " mode
Value uses the evaluation function of following formula:
TotalScore: representing total valuation of current situation,
NumberOfMyRode: representing the number on our every road of current situation,
ScoreOfMyRode: representing the value on our every road,
NumberOfThreatRode: representing the number of current chess game other side Ge Lu,
ScoreOfThreatRode: representing the value of other side Ge Lu,
PosValueOfNode: two positional values begun on chessboard in current way to get there are represented.
The valuation of entire situation subtracts other side's valuation along with chessboard positional value for our valuation, when advantageous to us I
Square valuation is bigger, and when unfavorable to other side, other side's valuation is smaller, and valuation total in this way will become larger, and just illustrates current way to get there
It is relatively good.This valuation mode feature based on " road " is exactly that evaluation function is fairly simple, while not needing to carry out chess piece again
Complicated analysis, so valuation efficiency is relatively high.Node is chosen according to the width value of setting, the node as extension returns.
Step 4: begun according to search result simulation；
Step 5: by depth minus one, Alpha value and Beta value are exchanged, and changes player, then executes step 1；
Step 6: using based on valuation is carried out to leaf node by the way of " chess type ", the father of present node is then returned to
Node；
When carrying out valuation to leaf node, the valuation mode based on " chess type " is used, the evaluation function of situation is such as
Under:
TotalScore=ScoreOfMy+ScoreOfThreat+ScoreOfExtra+W
ScoreOfMy: representing the valuation of our current situation,
ScoreOfThreat: representing the threat value that other side generates us,
ScoreOfExtra: representing additional synthesis valuation,
W: the unbalance factor of specific gravity is represented.
The valuation trend of the evaluation function is exactly to choose not only advantageous to us but also can eliminate the prestige that other side forms us
The side of body, even if ScoreOfMy and ScoreOfThreat reaches bigger value simultaneously, and by considering consequent comprehensive analysis,
It is assessed on the whole.Wherein the valuation of ScoreOfMy and ScoreOfThreat includes following formula:
ScoreOfMy=FirstStoneOfMy+SecondStoneOfMy+AddPowOfMy
ScoreOfThreat=FirstStoneOfThreat+SecondStoneOfThreat+Add PowOfThreat
FirstStoneOfMy: representing our first valuation to begin,
SecondStoneOfMy: representing our second valuation to begin,
AddPowOfMy: representing the added value of our two relating headings of beginning,
FirstStoneOfThreat: representative simulation other side first begin other side generation threat value,
SecondStoneOfThreat: the representative simulation other side second threat value begun to our generation,
AddPowOfThreat: the threat value that we is generated on representative simulation other side's relating heading.
Valuation needs based on " chess type " begin and carry out chess type analysis, including first is begun and begin with second, then
It is to be scanned to judge chess type (except the scanning of relating heading) to each four direction begun, if two are begun in the presence of association
Direction needs individually to carry out Information Statistics to relating heading, then carries out valuation again.Association refer to two begin exist simultaneously in
Level, vertically, on one of left oblique and right oblique four direction, and distance is less than or equal to 5.
For the value of ScoreOfExtra, including the urgent work number of generation, the size of border interference and expansible space.
These different factors can all have an impact total valuation.Compel work for having, can successively judge to compel to write number, such as compel to write
When number is more than or equal to 3, the way to get there that must be won is found, valuation is adjusted to maximum.For border interference, principle is closer to side
Boundary is unfavorable position it is necessary to suitably reduce valuation.Analysis for expansible space is to need to increase when being located at void spaces
Add valuation, if two begin between vacancy it is seldom, just do not have ductility.
The considerations of for W value, formula is as follows:
W=W1+W2
K1=max1/min1
K2=max2/min2
W1: our first valuation that begins (FirstStoneOfMy) and first valuation that begins of other side are represented
(FirstStoneOfThreat) difference generated when unbalance,
W2: our second valuation that begins (SecondStoneOfMy) and second valuation that begins of other side are represented
(SecondStoneOfThreat) difference generated when unbalance,
K1: representing the unbalance factor of first chess piece valuation,
K2: representing the unbalance factor of second chess piece valuation,
Max1 and min1: respectively representing the big person and small person of we and the valuation of first chess piece of other side,
Max2 and min2: the big person and small person of we and the valuation of second chess piece of other side are respectively represented.
When the unbalance factor is more than or equal to 2.5, illustrates to simulate our first valuation to begin and other side first is begun
Threat value difference it is larger, i.e., currently begin only to a Fang Youli or be only there is attack not defend or be only to have
No attack is defended, current way to get there is not required, so needing appropriate reduction valuation.
Evaluation function based on " chess type " needs to consider several factors, and function is also more complicated, while also needing to chess piece
Complicated analysis is carried out, the judgement of chess type, so valuation efficiency is lower, but valuation accuracy is very high.
Step 7: judge present node depth whether be setting depth, i.e., whether be root node, saved when traverse piece
When the child node of point, search terminates.
In initial stage (expanding node) of search, due to the initial stage in deep search to the valuation accuracy requirement of situation not
Height, no matter how valuation is returned as long as assuming to guarantee that best way to get there is fallen in the width range of setting by last
It traces back, just can determine that Best Point.Therefore need to be arranged bigger width and depth, and at this moment, if only using single base
It is that much can not meet this demand in the valuation mode of " chess type ", because the analysis and valuation of chess type are more complicated, search effect
Rate is low.So another valuation mode based on " road " can meet this requirement, and the valuation based on " road " be also can
To guarantee certain accuracy, therefore, the section that extension is treated based on " road " mode should be used in the case where not losing valuation accuracy
Point carries out valuation, and the width of search and depth can be made to increase.
It,, can not because the chessboard of six sub- chesses is very big due to having reached pre-set depth when searching leaf node
Chess game can be simulated terminates (state to decide the winner), so certain depth can only be simulated, then estimates to situation
Value, and final analog result is represented with this valuation, the result is certainly accurate without simulating the result to decide the winner, so leaf
The valuation accuracy of node is most important.If at this time still using based on to leaf node valuation, being can not by the way of " road "
Guarantee very high accuracy.So just valuation should be carried out to chess game by the way of being based on " chess type ", at this time to search efficiency
It is required that be not it is very high, scanned for using the valuation mode based on chess type, the accuracy of search can be improved.
In trace-back process, prune approach is still used.By valuation mode and base based on road in entire search process
Order by merging use has been carried out in the valuation mode of chess type, the advantages of each comfortable corresponding search phase shows oneself, from total
Balanced search efficiency and game are horizontal on body.
Hybrid Search mode with it is single based on the search pattern on " road " compared with, improved in the identical situation of search efficiency
Game is horizontal, with it is single based on the search pattern of " chess type " compared with, improve search efficiency in the case where identical game is horizontal.
System design
Valuation based on " road " is respectively used to win six sub- chesses in Alpha-Beta search with the valuation based on " chess type "
The valuation set to expanding node and leaf node is played chess, devises six sub- chess gaming platforms on this basis.It mainly include following several
A module:
(1) chessboard indicates
(2) way to get there generates
(3) evaluation function
(4) it searches for
(5) beginning library
(6) interface
Chessboard is indicated using two-dimensional array；Way to get there generation module is for generating all legal ways to get there；Evaluation function module
Including the evaluation function based on " road " and based on the evaluation function of " chess type "；Search module, which includes that simulation is continuous, compels work and Alpha-
Beta pruning search；Beginning library is used for the matching of beginning situation；The execution process of system: it is input way to get there first, then judges
It whether is opening stage, if it is the valuation and search with regard to not done depth, because to do deep search meaningless for beginning, if
It is not beginning, just carries out simulating continuous urgent work, continuous urgent write is for judging whether that the way to get there that must be won can be found, if urgent work
Success, will export way to get there, if it fails, carrying out defensive Alpha-Beta beta pruning deep search again, exports best way to get there.
Evaluation function based on " road "
In the evaluation function based on " road ", it is necessary first to solution be each road value setting.General setting is to increase
The setting of long formula, i.e. number are bigger, are worth bigger, for example it be 90,3 tunnels be 200,4 tunnels is 800,5 tunnels that 1 tunnel, which can be 30,2 tunnels,
It is 1000,6 tunnels for that must win, maximum such as 1000000 should be set as.
Followed by the setting of positional value of the chess piece in chessboard, rule of thumb it is found that center of the board position is that beginning is best
Position, closer from central point, value is higher, for example it is 10 that middle position value, which can be set, then successively subtracts one along four direction.
Finally it is exactly to be scanned to chessboard, counts us and to the quantity on each road of square chess.In order to avoid invalid is swept
It retouches, only counts effective number, and the range of scanning should be reduced simultaneously, is i.e. diminution chessboard, enabling enough minimum chessboards includes
All chess pieces on chessboard.
Evaluation function based on " chess type "
In the evaluation function to individually begin carry out valuation when, will successively carry out valuation from four direction, then
Specific scanning mode is then to judge corresponding chess type to be scanned in the range of being 5 comprising a distance of beginning.And it is right
It is then that Information Statistics, analysis, then according to the connection of adjacent chess piece point first are carried out to the direction in the judgment mode of chess type
Do not judge, finally provides a kind of matched chess type.Setting for value table is that there are two valuation tables, such as two following
Table:
Table one:
Table two:
| Chess type | Value |
| Dormancy one | 10 |
| Dan Guan | 20 |
| Dormancy two | 200 |
| Jump two | 450 |
| Living two | 600 |
| Dormancy three | 1300 |
| Dim three | 2100 |
| Living three | 2400 |
| Dormancy four | 5500 |
| Living four | 9000 |
| Dormancy five | 8000 |
| Living five | 10000 |
| Living six | 300000 |
Table 1 is the valuation of our each chess type, and table 2 is the threat value of other side's chess type, can foundation after judging chess type
This table returns to corresponding valuation.If there is association in two chess pieces, it is also necessary to which the information on relating heading is recorded and divided
Then analysis provides the judgement of chess type and added value.
Determination for factor size unbalance in evaluation function, this reality example, can be according to oneself warps using 2.5
It tests and the value is set, appropriate value can also be chosen by experiment.
Chessboard indicates
The chessboard of first six sub- chesses is the rectangle of 19x19, each crosspoint is exactly the position of a chess piece, is to use here
One two-dimensional array chess [19] [19] indicates position of the chess piece on chessboard, and initial value is all 0.For the color of chess piece,
Used herein is an integer, and 1 represents black, and 7 represent white side.Such as when chess [1] [2] are equal to 1, represent the position
Setting is black piece.Begin will the position initial value 0 assign corresponding 1 or 7 the above, be only preferable implementation of the invention
Example, not does any type of limitation to the present invention.All technology and methods essence according to the present invention is implemented to above
Any simple modification, equivalent change and modification made by example, in the range of still falling within technology and methods scheme of the invention.
Claims (4)
1. the mixed search algorithm based on road draw in chess type in a kind of six sub- chess game playing systems, it is characterised in that the following steps are included:
Step 1: setting nodal information, width value, depth value, Aapha value, Beta value and player；
Step 2: judging whether it is leaf node, if it is leaf node, thens follow the steps six, otherwise executes step 3；
Step 3: expanding node obtains the node for needing to be extended, using the mode based on " road " to the node for needing to be extended
Valuation is carried out, and chooses number of nodes according to the width of setting, the node as extension returns；
Step 4: begun according to search result simulation；
Step 5: by depth minus one, Alpha value and Beta value are exchanged, and changes player, then executes step 1；
Step 6: using based on valuation is carried out to leaf node by the way of " chess type ", the father node of present node is then returned to；
Step 7: judge present node depth whether be setting depth, i.e., whether be root node, when having traversed root node
When child node, search terminates.
2. mixed search algorithm according to claim 1, it is characterised in that: the step 3 is based on " road " mode and carries out
Valuation uses the evaluation function of following formula:
Wherein, TotalScore: total valuation of current situation is represented；NumberOfMyRode: our every road of current situation is represented
Number；ScoreOfMyRode: the value on our every road is represented；NumberOfThreatRode: it is each to represent current chess game other side
The number on road；ScoreOfThreatRode: the value of other side Ge Lu is represented；PosValueOfNode: it represents two in current way to get there
A positional value begun on chessboard.
3. mixed search algorithm according to claim 1, it is characterised in that: the step 3 be based on " chess type " mode into
Row valuation uses the evaluation function of following formula:
TotalScore=ScoreOfMy+ScoreOfThreat+ScoreOfExtra+W (2)
ScoreOfMy=FirstStoneOfMy+SecondStoneOfMy+AddPowOfMy (3)
ScoreOfThreat=FirstStoneOfThreat+SecondStoneOfThreat+Add PowOfThreat (4)
Wherein, ScoreOfMy: the valuation of our current situation is represented；ScoreOfThreat: represent what other side generated us
Threat value；ScoreOfExtra: additional synthesis valuation is represented；W: the unbalance factor of specific gravity is represented；FirstStoneOfMy: it represents
Our first valuation to begin；SecondStoneOfMy: our second valuation to begin is represented；AddPowOfMy: it represents
The added value of our two relating headings of beginning；FirstStoneOfThreat: representative simulation other side first begin other side production
Raw threat value；SecondStoneOfThreat: the representative simulation other side second threat value begun to our generation；
AddPowOfThreat: the threat value that we is generated on representative simulation other side's relating heading.
4. mixed search algorithm according to claim 3, it is characterised in that: the setting of the unbalance factor W of specific gravity is adopted
With following formula:
W=W1+W2 (5)
K1=max1/min1 (6)
K2=max2/min2 (7)
Wherein: W1: representing our first valuation that begins (FirstStoneOfMy) and first valuation that begins of other side
(FirstStoneOfThreat) difference generated when unbalance；W2: our second valuation that begins (SecondStoneOfMy) is represented
The difference generated when unbalance with second valuation that begins (SecondStoneOfThreat) of other side；K1: it represents first chess piece and estimates
The unbalance factor of value；K2: the unbalance factor of second chess piece valuation is represented；Max1 and min1: we and other side the is respectively represented
The big person and small person of the valuation of one chess piece；Max2 and min2: it is big with the valuation of second chess piece of other side to respectively represent us
Person and small person.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610145217.0A CN105817029B (en) | 2016-03-14 | 2016-03-14 | Mixed search algorithm based on road draw in chess type in six sub- chess game playing systems |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610145217.0A CN105817029B (en) | 2016-03-14 | 2016-03-14 | Mixed search algorithm based on road draw in chess type in six sub- chess game playing systems |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105817029A CN105817029A (en) | 2016-08-03 |
| CN105817029B true CN105817029B (en) | 2019-09-03 |
Family
ID=56987179
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610145217.0A Expired - Fee Related CN105817029B (en) | 2016-03-14 | 2016-03-14 | Mixed search algorithm based on road draw in chess type in six sub- chess game playing systems |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105817029B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107050839A (en) * | 2017-04-14 | 2017-08-18 | 安徽大学 | Amazon chess game playing by machine system based on UCT algorithms |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106339582B (en) * | 2016-08-19 | 2019-02-01 | 深圳市云安晟科技有限公司 | A kind of chess and card games automation final phase of a chess game generation method based on game playing by machine technology |
| CN106407670A (en) * | 2016-09-06 | 2017-02-15 | 中国矿业大学 | Game algorithm-based black and white chess game method and system |
| CN110275739B (en) * | 2019-07-26 | 2023-03-28 | 陕西师范大学 | Method for judging hand prohibition and chess shape attribute of gobang within constant time |
| CN112755520A (en) * | 2021-01-19 | 2021-05-07 | 寇英翰 | Chess force improving method based on Alpha-Beta pruning algorithm |
| CN113946604B (en) * | 2021-10-26 | 2023-01-20 | 网易有道信息技术(江苏)有限公司 | Staged go teaching method and device, electronic equipment and storage medium |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP4514771B2 (en) * | 2007-05-18 | 2010-07-28 | 株式会社エスグランツ | Coupled node tree longest match / shortest match search device, search method and program |
| CN101571892A (en) * | 2009-04-27 | 2009-11-04 | 姚斯宇 | Method for choosing landing points of I-go in I-go human-computer game-playing software |
| CN105302963A (en) * | 2015-10-30 | 2016-02-03 | 沈阳航空航天大学 | Machine game optimization method |
| CN105797380B (en) * | 2016-03-14 | 2020-06-19 | 广州趣丸网络科技有限公司 | Method and system for realizing information interaction among group members in mobile terminal game |
-
2016
- 2016-03-14 CN CN201610145217.0A patent/CN105817029B/en not_active Expired - Fee Related
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107050839A (en) * | 2017-04-14 | 2017-08-18 | 安徽大学 | Amazon chess game playing by machine system based on UCT algorithms |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105817029A (en) | 2016-08-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105817029B (en) | Mixed search algorithm based on road draw in chess type in six sub- chess game playing systems | |
| Torrado et al. | Deep reinforcement learning for general video game ai | |
| Cowling et al. | Information set monte carlo tree search | |
| Nielsen et al. | Towards generating arcade game rules with VGDL | |
| US11247128B2 (en) | Method for adjusting the strength of turn-based game automatically | |
| CN105727550B (en) | Point lattice chess game system based on UCT algorithm | |
| CN101571892A (en) | Method for choosing landing points of I-go in I-go human-computer game-playing software | |
| Yang et al. | Guiding Monte Carlo tree search by scripts in real-time strategy games | |
| CN112685921B (en) | Mahjong intelligent decision method, system and equipment for efficient and accurate search | |
| Nissen | Analysis of a complex conditional reaction in chimpanzee. | |
| Guerrero-Romero et al. | Map-elites to generate a team of agents that elicits diverse automated gameplay | |
| Hu et al. | Reinforcement learning with dual-observation for general video game playing | |
| CN114146401A (en) | Mahjong intelligent decision method, device, storage medium and equipment | |
| Smith et al. | Artificial intelligence approaches to build ticket to ride maps | |
| Thawonmas et al. | Aggregation of action symbol sub-sequences for discovery of online-game player characteristics using KeyGraph | |
| TW400506B (en) | A method of executing black-white chess in computer | |
| Yin | Playing Chinese Checkers with Reinforcement Learning | |
| Takeshita et al. | Reduction of the search space to find perfect play of 6× 6 board Othello | |
| Sannidhanam et al. | An analysis of heuristics in othello | |
| Takeuchi | Comparison of Search Behaviors in Chess, Shogi, and the game of Go | |
| Sun et al. | Pattern matching and Monte-Carlo simulation mechanism for the game of NoGo | |
| Wang et al. | Research and implementation of einstein würfelt nicht! algorithm | |
| Guangyao et al. | Study on the algorithm based on the important region of board | |
| Castro-Wunsch et al. | Beemo, a Monte Carlo simulation agent for playing parameterized poker squares | |
| Setiawan | Playing the SOS Game Using Feasible Greedy Strategy |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20190903 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |