GAMIFICATION OF CHESS FIRST MOVE IN MULTICORE ENVIRONMENT FOR ONE TO MANY RELATIONS

Transcription

1 GAMIFICATION OF CHESS FIRST MOVE IN MULTICORE ENVIRONMENT FOR ONE TO MANY RELATIONS Shital Bhabad 1, Sarang Joshi 2 1, 2 Department of Computer Engineering Pune Institute of Computer Technology, Pune, India ABSTRACT: Gamification is one of the recent strategies used for problem solving. It makes use of game design element in non-gaming context to improve user engagement. Gamification engages and motivates a group of people to solve problem in distributed environment. Demanding mobile applications such as multiplayer gaming, video streaming are stretching the capabilities of the single core smartphone processor. In order to increase the performance and stay within the budget of mobile batteries, it is inescapable that all mobile processor will have multi core processor. In multiplayer chess game, use of multicore processor computing power can be advantageous as this can helps in increasing the performance of the application. This research paper performs analytical processing and related mining to select first favorable move in chess gamification. Estimation of the goodness of the move can be computed by proportion of the games won with this move. Estimation can be made by analyzing the chess games stored in the MongoDB database. Keywords: Data Mining, Gamification, MongoDB, Multicore, Queuing Theory [1] INTRODUCTION Chess is a major application area of research in machine learning and artificial intelligence [3]. Chess game can be divided in to an opengame, middlegame and endgame. In chess game, Group of initial moves is called chess opening by white and chess defense by black. In the self-play chess game most of the research has concentrated on the middlegame, endgame and evaluation function. Terminal state (win, loss and draw) of chess game depends on a sequence of states during game play. Initial state includes selection of chessmen to make first move to start with game. It is necessary to devote a lot of attention in the opening stages to: 1. Development of Chessmen: main aim of the opening is to mobilize the chessmen on squares so they will have positive impact on the terminal state of the game. 2. Control of the Center of Board: allows chessmen to be easily moved to any part of the board and can also have a cramping effect on the opponent. 3. Prevention of the pawn weaknesses: avoid pawn move so it will not result in isolated, doubled and backward pawn [1]. Chess is not a game of chance so main objective of opening in chess is to obtain better and comfortable position when playing as white and to equalize when playing as black. Most of 1

2 Gamification Of Chess First Move In Multicore Environment For One To Many Relations the research in the area of chess game has focuses on an evaluation function to improve selflearning chess game. In chess game selecting a sequence of favorable moves that leads to game win state is a challenging task. Newbie chess players usually do practices by analyzing various existing chess opening. It becomes complex when they analyze and correlate hundreds of chess openings. In the game of chess, deeper the search of game tree more accurate is the results. Computational power of smart phones limits the number of play game tree to search. Parallel programming with the use of multicore environments can result in increase in speed and performance of the application. In multiplayer chess game, there are number of task that can be performed in parallel this can result in improved performance. If task are performed in sequential order, time required to respond to n players grows exponentially and this will degrade the performance of the application. In this paper a new strategy is considered to choose favorable move in initial stage of multiplayer chess game. This strategy makes use of chess game database to analyze and selection of favorable move based on terminal win state of game. An efficient mining strategy need to be used for extracting knowledge about favorable move from the chess game database. Multicore computing capabilities are used to make evaluation of different move to select best move and to service the n number of players at the same time where n is same as the number of core in the smartphone. The paper is organized as follows. Section II discusses the related work, section III describes proposed work for selection of first move using techniques of data mining and making of multicore environment to enhance performance of the application, section IV describes programmers design including mathematical modeling of the proposed system. The paper concludes in section V. [2] STATE OF THE ART Knowledge of the goodness of the move is base for the chess play which leads to winning state. Different methods are used to implement a self-learning chess program. Existing system provides solution for self-learning one to one chess game. Sarang Joshi et al. proposed system for selection of first favorable move using chess game database analysis based on win terminal state [1]. Christian Wirth et al. reported that for chess learning, chess evaluation function can be learned by preferences derived from game annotation. Pairwise preference data is used to train SVMRANK ranking support vector machine which reinterpret the preference statements as constraints on the evaluation function. Result did not yield better one to one chess playing program for computer [2]. Johannes Fuernkranz reported that playing strength of computer chess game increasing easily with the use of an opening book [3]. Aleksander Sadikov et al. proposed a method for dividing the endgame into stages automatically. Stages are characterized by different objectives of play. For each stage, stage specific evaluation function applied that used by minimax search when playing endgame. This method resembled human experts play in achieving goals reliably but not as quickly as possible [4]. R. M. Rani reported that chess opening can be improved by using data mining techniques such as association rule. This technique assigns variable to different sequence of 2

3 moves. Sequence association rule generated to calculate support and confidence which helps in finding best subsequence chess move which can lead to winning position [5]. Yan-Fang Fan et al. reported that in each match database stores the movements of the opponents and the position values. This helps in avoiding the same mistake next time by carrying out the game tree search on the opponents move that computer previously misjudged. Return value assigned to the above move so computer can search correct move next time [6]. Chrilly Donninger et al. describes an approach that evaluates the goodness of a move based on a heuristic formula that founded by the experimentation. This value is then added to the regular alpha-beta search [7]. Murry Campbell described the working of deep blue. Deep blue extract useful knowledge from a database of 700,000 grandmaster chess game and make use of opening book and extended book. Deep blue also make use of multicore which yield it as fastest chess program [8]. Giovanni Mariani et al. proposed an Application-specific Run-Time Management (ARTE) framework to tackle the problem of managing computational resources in an application specific multi-core system. ARTE uses M/D/1 queuing model at run-time to predict the applications response times [9]. [3] PROBLEM PP The proposed system designs the multiplayer chess game for smart phone. This game will be able to select favorable move which leads to the win state. The favorable move will be selected by analyzing database of chess game and extracting the knowledge from the database by using the data mining techniques. Proposed system achieves the concurrency in multiplayer chess game by making use of multicores of the smartphone processor. Idea of multiplayer chess game is explained as, one mobile player can be able to play with n other mobile player at the same time. Chess game is n one to one chess games played against n opponents. In this research paper we are taking into consideration selection of first favorable move for the n player. State diagram is drawn to visualize the different cases and flow of action in sates for selecting first favorable move in multiplayer chess game. Cases and states in figure:1 are described as follows: S1: start of the chess game S2: first move (playing as white) S3: opponents fist move (playing as black) S4: either start game and make move (playing as white) or respond to the opponent (playing as black) S5: your first move based on opponents move S6: favorable move selection from available distinct moves S7: make first favorable move. Case 1: Player1 has to start with the first Move. This is best case scenario in which player1 has to make move to all other n players. If player 1 requires t time to make first move to any one of the opponent player then to make first move for 3

4 Gamification Of Chess First Move In Multicore Environment For One To Many Relations n 1 t t all opponent is t nt which is n times multiple of the time required to make first e e move to only one opponent and t e is time required to select favorable move. Case 2: Player1 has to make his first move This is worst case scenario in which player1 has to make his first move depending on all other n players corresponding first move. If all the opponents make their first move at the same time then the waiting time of the opponents is t +t e, 2t + 2t e,3t + 3t e,...,nt + nt e as time required to make move for player 1 is t = n(t + t e ). Case 3: Player 1 has to start with first move as well he has to make his first case In this scenario player1 requires k t +t e units of time to make first move to k opponents and (n k)t + (n k)t e units of time to make his first move to the remaining (n-k) opponents. In the above cases nt units of time is required to make first move if the smartphone with single core processor is used. If smartphone with m core is used then time required to make first favorable move is m times less than that of single core processor. Figure: 1. State diagram of cases for selecting first favorable move Case 1: m n n i m where i = 0,1,...,n m t where t p t p is time required to make move to n players in multicore environment, t s s is time required to make move to n players in single core environment. Case 2: n m m n n t p = t s where t p is time required to make move to n players in multicore environment, t s is time required to make move to n players in single core environment. Case 3: m n n ( i m) where i = 0,1,...,n (m t p ) 1 = t s where t p is time required to make move to n players in multicore environment, t s is time required to make move to n players in single core environment and this is the worst case scenario. In best case t p = 2 units. 4

5 Table 1 shows how performance and speed of application improves with the help of multicore smartphones. Selection of the first move in chess is NP-Hard problem and this problem can be solved by making it as NP-Complete by using method proposed in [1]. In chess, there are 20 C 1 possible ways to select first move. The selection of the favorable move makes use of method proposed in [1]. To select most favorable move, it is necessary to evaluate all 20 possible move. Time required to evaluate single possible move is say t unit then time required to evaluate all possible move for the first favorable move selection is 20 t units. Each different first move can be evaluated by a different processor/core to see how the game would continue from that point. Table1. Relation between number of cores and time required to serve number of player. At the end, these results have to be combined to find out which is the best first move. This saves the computing time and response time of player as well as waiting time of opponent player. Figure: 2. Relation between number of cores and time required to evaluate total number of possible moves. Assigning different player or different move evaluation processes to different core in multiprocessor computing environment is a challenging task. In the proposed system this challenge can be solved by using queuing theory and FCFS scheduling technique. For the multicore processor tasks can be assigned by using either M/M/1 or M/M/m queuing model [10]. For M/M/1 Queuing model each core is considered as a separate server and has separate associated queue of processes while in M/M/m queuing model each core is considered as a separate server and has single queue of processes. 5

6 Gamification Of Chess First Move In Multicore Environment For One To Many Relations [3.1] MATHEMATICAL MODEL With reference to the mathematical model presented in 4.1 of [1]. Let S be programmers perspective where class S = {s,e,t,p,cm,m,c,u,p pass,dd,ndd,succ,fail,f m,f i,sh m, φ s } and φ s represents the rules governing the chess board game and mobile mi must have multicore processor, s = T = 0,1 be the start (initializer or constructor) and e be the end (destructor) of the game. P = {p 1,p 2,.,p n } represents the total number of players participated in multiplayer chess game. C = {c 1,c 2,.c m } represents the number of cores in mobiles. M = {m 1, m 2, m 3,,m n }. U = {u 1,u 2,,u n } represents the username of player p i. P pass = {p 1,p 2,.,p n }represents the password of player p i. CM = {P,R,BI,KN,q,k} represent the chessmen [1], F = {P,KN} represents the chessmen participating in the first move where P = {p 0,p 1,.,p 7 } represents pawn, KN = {kn 1,kn 2 } represents knight. f p = P P Figure: 3. Mapping function of player on itself Initially p i (ri,cj) where i = 1,j = 0,1,2,,7. p i (r i-1, c j ) p i (r i-2, c j ) represents each white pawn p i can move either one or two cells forward only. Initially kn wi (ri,cj) where i = 1,j = 2,6. kn wi (r i-2, c j+1 ) kn wi (r i-2, c j-1 ) represent for the first move each night kn wi can move two ways. For the initial move, all possible moves are (p i (r i-1, c j ) p i (r i-2, c j )) (kn wi (r i-2, c j+1 ) kn wi (r i-2, c j-1 )) = 20. In multiplayer chess, each mobile player p i invite or accept connection from other players p n-1. SH m = {U, P pass } = {(u i,p i ), (u j, p j )} where i 6= j and i,j = 1,2,,n. Figure: 4. Time for calculating first favorable move in sequential manner 6

7 To improve the performance of the game it is necessary to use multicore programming. This will help in more realistic and interactive chess game. Figure: 5. Time for calculating first favorable move in multicore environment Each mobile phone m i where i = 1,2,.,n has m cores and each core is associated with the processing of n number of processes (here processes are game players) so mapping function required on M,C and P. DD = C = total number of cores in mobile. f m = M (C P) Figure: 6. Mapping function of mobile core. Player p i generates the request and assumes that request follows Poisson process with a mean generating rate of λ i and this request can be assigned to any one of core c k depending on the number of resources and types of demands Figure: 7. M/M/m model of queuing system 7

8 Gamification Of Chess First Move In Multicore Environment For One To Many Relations Case 1: different demands to different resources For each request r i can be assigned to different cores c k. In chess this scenario will come in to picture when all the opponents have already made first move and player has to make his first move. Case 2: different demands to same resource. For each request r i can be assigned to same cores say c 1 then rest of c k-1 cores are idle. In chess, this scenario will come into picture when player has to make first move to all opponents and history of none of opponents play strategy is available in the database. But in chess, above two scenarios will be executed in worst case. Case 3: In multiplayer chess, the scenario is the combination of above two scenarios. In multiplayer chess game n one to one game are played concurrently with (n+1) players. As it is online multiplayer game, some games are started by player and some are initiated by opponents. [4] EXPERIMENTAL SETUP The Experimental setup for this proposed system will be building of development and testing environment. Proposed system will be developed in eclipse and tested on real time smartphone with multicore functionality. The particulars about platform and technology used: Base Operating System: Fedora 19/ Windows 8 Databases: MongoDB Tools: Eclipse Kepler Language: Java Plugin: Android According to these specifications and structure, the system is built up. [5] CONCLUSION Game of chess is one of the active research areas in the fields of artificial intelligence (AI). Game of chess is not a game of chance as in chess every move depends on the move of opponent. To increase the intelligence of chess it is necessary to apply more efficient mining techniques on the database of chess games. More the efficient prediction of next move more powerful the game will be. Prediction of next move will be more accurate if mining result applied on huge amount of data. Large amount of chess data available and grows with time. To address problem of growing chess literature it is necessary to use BigData such as MongoDB. Proposed solution helps to program chess which may lead to win state. Proposed system makes use of multicore computing capabilities of smartphones to improve performance and response time of the multiplayer chess game. In future, we are planning to extend the searching and mining techniques to be applied to all the moves in game of chess. 8

9 REFERENCES [1] Sarang Joshi and Shital Bhabad, Mobile Interoperability Algorithm for First Move in Chess Gamification. In the International Journal of Engineering Technology and Applied Research, pp 1-6, 2014.J. Clerk Maxwell, A Treatise on Electricity and Magnetism, 3rd ed., vol. 2. Oxford: Clarendon, 1892, pp [2] Johannes Fuernkranz, Recent Advances in Machine Learning and Game Playing. In Journal of Oesterreichische Gesellschaft fuer Artificial, [3] Aleksander Sadikov and Ivan Bratko, Learning long-term chess strategies from database. In the Special Issue on Machine Learning in Game in Machine Learning, pp , [4] R. M. Rani, Analysis of Sequence Moves in Successful Chess Opening using Data Mining with Association Rules. In Journal of world academy of science, Engineering and Technology Vol. 48, 2010, pp [5] Yan-Fang Fan, Xue-Jie Bai, Rui-Ying Liu and Sheng Xing, The Research of Chinese Chess Based on Database with Self Learning. In Proceeding of the ninth International Conference on Machine Learning and Cybernetics, Qingdao, july [6] Chrilly Donninger and Ulf Lorenz, Innovative opening-book handling. In Journal of Advances in Computer Games, Springer, [7] Murry Campbell, Knowledge Discovery in Deep Blue. In the Communications of the ACM, Vol. 42, No. 11, Nov 1999, pp [8] Giovanni Mariani, Gianluca Palermo, Vittorio Zaccaria and Cristina Silvano, ARTE: An Application-specific Run-Time management framework for multi-cores based on queuing models. In Journal of Parallel Computing., [9] Kishor S. Trivedi, Probability and Statistics with Reliability,Queing and Computer Science Applications. In Second Edition, Duke University,Durham, North Carolina, pp [10] MongoDB Documentation, release 2.4.6, [11] 9