Scribd
Upload
188 views

Unit 4- (Part B)Game Theory

Ppt

Uploaded by

Mansi Tuteja
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
188 views

Unit 4- (Part B)Game Theory

Ppt

Uploaded by

Mansi Tuteja
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
You are on page 1/ 51
lame Theos Game : Game %s defined as am ackuily among two o% move persons as per set of gules at the end of Which each person gels some bene Kit 07 bears loss. The set of ules omd puoceduse delves the game A com betitive Staton ts called game. The tem qeme | Bebuesents a conlic between fuse ov more pasties. Game Theory: Game theory Studies triesactve — decrsion-waking| whese the outcome bor each pawl erbant oF pladgess debend gn the achons of all- Th You ore playes in such a game, when | choo swig your course ob ackon gy skategy, you rust dake unto account the choices of others. Ti se ‘Scanned with CamScanner | Use of Game Theowg : Gome therry com help combemtes | | - : make strategies chotce atthe | ox outside of, theiy Og mization S, @s becially against | | combibibors. | Diphesent stations ave presented through suinble qames jrat set Ub hypothetical scenaxio meant fe simulate | weal- world conditions ond predict o players behaviour Gare Yreory con help to predict how feeble behwe — | | when they ase tna com betihve siluaton- | Types o} james : | (Tloo-person_ and wn- pexson games: En 4wo- bevson ames, the pleqers — may have esscble choices | | 4 the number o to them each play of the game bul ot tages es o bi. Hence, tt ts Called a tooo enon. game. In case ob more than. two person, the yore, L's _genetally called m= person gare. | ‘Scanned with CamScanner Cidzeso_sum game : A Zero- sum. game {s one cn behich ne ‘ sum oy the payments fy all the competi tows oe evesy ossible outcome of the gome dhe sum ob Ane points wen, equals the Sum Of finds Ne: Git Tigo persom_zexo sum: A game with wo blayers, where fhe gain Of one playes eqpals the \oss ok dhe othes, {s ee 0S 0 two-bersen zexo - Sum qome- Tk is also calle 7 rectangular gone because thers ‘pay. bh matrix is Beclomqular form a Wo 10 =D A oc oc “Mp -I0 30 10 1D = 0 ‘Scanned with CamScanner Strate g : The term ‘Strategg’ is defied as a tom ele set of plans of actun spect Sym ecisely What the ex ah do undes oe Maud future contigency that might occux dug the play ob the gome, he, $ bas Ob o& ployer ts the deciston ule he uses fos om a choice, from his list of courses of acton. shategy con be Classthred as: @® Pure strategy ti) Mixed strategy - . trot ts called uve tf one Knows In Pose _shalegy © A PONT Sane play lahat ths 16 cerdain fo be adobled , Trseshective of the shotegy the other players ‘might chose. i ab ‘ oun turse h Mixed strategy : ry obhal grote ined 4 Prete a fo each shategy, its een of betng chosen. The Shotegy So determined is called mixed strategy because ‘Scanned with CamScanner ibis o pwebabiliste combinaton of the available choice ok skategy. Mixed shategy is dented by dhe set S; Se {XX en xm$ 4» where KX is the probability of choosing Cousse So Xt ket Xgt oe tXmed i Tk is evident Yhat a Pere shategy isa Sbeeral ob mixed strate gy. case ‘Scanned with CamScanner y| Pay- of f : Pay ~ 6 | 6 Column maxim 5D ! 5G 4-6 ee ee es Ce » Min (Maximum) = % = Me (5,3, 1596) Since, Yc Val ste exits o Saddle point. Nola of the 7 ; fodeay and is The Position of saddle pévet is the oe. 3 (4, 8:) \ \ MW " ‘Scanned with CamScanner & Detesrnine which of the following, ase Strictly delerminable and fair gent for" each player im he case oy Player @ (i) 8B, Be 5 As ool Ployes A ‘i : | oF Player A Min, (Meximurn’) = Mine ib) = Maxmin = Max( 5.4) = Bot a-S FO * The Goma. fs Stet Volue Ca he gone Alimal shoteag is He posthio (Ave) « Column. ence 4 4wo- pesson zevo -sum games + Given. the cman. oF Stoic Hy ie Player 8 3! 8, Rew ymin al al ALY -3 jes 1 Max (minimus) 2 Mex (3) =! Min (Maximum) =Min (Wot) 2 1 Yous $0 Te gone ve shy delerminabe| volun of He game > \ Othimal shade gy is Wee Posten (Av Ba) . ‘Scanned with CamScanner Gronnes (lishout Sadle points (Mixed Steategy) A game without saddle peent com be solved by Vorwious solukon methods . (2x2 Games without Saddle point Wi Gsacphicat Method for 2xn oF mxd Gomes GW) Dominance Pro pent . ‘Scanned with CamScanner 2x2 Games Without Saddle _bocnt Constdes a 2x2 4wo-pexson Zebo-sum game wethout omy saddle point, having the boob rmabix foo blages 4 Ob , B Ba ayo | pa on Ora 8s 2 1 Az) ond Sgr The optimum mt xed strate gies, sae[P ra ae y h ee al =ypcl- ees | (assan) = (auras) > Piro Pe='- pr = Boa - Orn ity = qe ku (aut dor) - (412442) Tre value of the gore (7) : dy Aaa, ~ Ara ar a (au +422) - Car +21) ‘Scanned with CamScanner G+ Solve the following pey-off mahrx. Also delesmine the Opt mal shategies god Value of ‘the geme. ‘Ted BB A As] 8, B.) Solution? AY Fay se 5 poe la | Hr [aan Aas So deo aes ce (au azn) - (41.8401) (B+)-(143) 5 4 pr tpa 2! a pekb = re =¢ Se ee WP Gyatial= (anata), (A= C8) 7 ur 2! ~~ Ge Ih = 1-2 ao ‘Scanned with CamScanner Vale of; the game Gute ae (au 44x) = (aw eda) = bat = 'x8 (Baa) -C43) 5 JF Optimum mixed —shotegy 5 sa(boP) ond te (4-2) ‘Scanned with CamScanner [Q Ina qome of ‘malching coins with wo blayess, suppose A wins | One unit of value when these ave 4wo heads, wins noting when there ase feo foils amd losses + Umit of value vohen. these ase ene head amd one oil: Delesmene the bay. off, ~matrix, the best shadegies fs ech player and the volue of the gome be Ae Soludrom: The oyoth eater fr the cs gow by, bb. Ce ee Pr eed | Plager A (¢ | ae oe q Tle oO o =o se a aa 4 cy > roe og 2% Obhimam stsodeg» sae 2 eile | PG) bE Va of gore oes) + 372) + 3 BE | : 4 = pias i | mY eo | v7 | Sohwe a previaud foi rndat ‘Scanned with CamScanner Grobhical Method fox 2xm ov mx2 Games Q: Solve the following 2x3 qrme grebhicallg: xu eae ny i 8 Playes A | _ : 2 © “ohakon: Since the foblens ches mt fests @ ony Saddle point, Le} the player A play by the mixed strategy Sne( * ae) woth fae Pr ca ue The guth plotting fre ne aloove bay of madee iS giver by- Axis 1 ‘Scanned with CamScanner es A ysishes fo mawemize his -minéanum excepted considered the heighest petet of intessecton — H expected bay Mb eytation « T od passing dreeugh- H define the ‘welevent Now since poy obb» we tm the bower envelope ob As The lines : weoves that I ond TL alone a fo play The Selthrone + 37 2 yedues tre otuginal — 2™ cee eer ree . a i 7 -3 a S & aed eee a 2 pa -2 mons “AL 4 : 4 ee Tu op himem shotegy fee peta) eae eis gre a 2 Pa (lee aay He spe(Hof) oo 84: (40%) a (3x3) +(Se) 5 HA2 2 yp SS vele. of te Gem See =U a WV ‘Scanned with CamScanner Goephical Method for mx2 Gomes S 6x2 gome problem graphically . -3 |-3 5/3 6 |-\ rql 2|® 0 |-5 6 sen protien dues at poses port. Truce fore det peje : J. strate: play by ere a ie a oaainst bay: a: ‘Scanned with CamScanner Since plajer B vsshes fo minimize he maximum exbected bari ConsideeL the — loowest point 2% fe upper boundary of Bis exh ected pry! ~ OFF equation « infer sectin of Urs 2 amd 4) The pont H resents the Tnuni max ected value of the game for player B. on Me ae et Cx2 game wdues to 2x2 Hoff mateix. ’ x2 9) on ; Py 77 = ey os i a se Te set ey BA nt 8 8 re - qn bys by 2(9 3 10, % 10, ) Psa racy o (4 -=) (ax-3) +42) ee ote Vous of Ye game = Soe 24S ese D4 2x2 Formula ont ‘Scanned with CamScanner | | Dominance Property ] The pornciple of dominance states that dy ome Shategy | a ex dominates oves fhe other strategy en all condifjons then ‘the [eter strategy cam be tgnosed. A shateay dlominales |Over jhe others only tb th is \breLevabl hex ¢ 1 condihons- P fesable over other tm all ed ee ee of the | | bye matrix by deleting those strates tes, Which ate deninated by others. The qeresol Sele bow dominamee UAE | | If alt the elements of & 1, 504 KM ow one less then, ov |“ eqpol -& the comesponding elements af omy other Tow » seq 07" | tow, then KM gow is dominated by the ah so wan, Say jh Coluron , AF | corres boncking element Ca any | Shen | Kin column is dominated, Ui 1p all the elements of % col quaker Ahan 08 egal to tne ether Column, $ xt column, by th column: ‘Scanned with CamScanner Gi) Derimated tows amd calumns rmay be deleted -b we duce tre size of the bey -ob modsix as the optimal skate gies will Yemain unabrected- ©) Tf some lineas combinations of Some Bours comninate i row, then the Ub wow will be deleted - Sten lax oxquments fella fox columns. ‘Scanned with CamScanner 1@. Sele te followeng game. lager & | Mie = 2 1O | Player Am 6 2 + \S ast 1B Yrind raw au les then oF equal Second Bw, therefore the ard vow: So we com delete Solution: Since ath the element im the fp the caves bonding elements Hhind you iS) dominafed by Yard tow (domineded vou) | Se meb & Now, dre teduad re: te amoterx US | a al | Ad 6 2 ane clmente of ed Clem oe rods Hoo or ey | He correspondnig elements of dhe first column, which gave Arek colrmn Sa we com delete i he al So) Oy) i ta i WE s (ie beb os : ie a) (~ i ‘Scanned with CamScanner The Opkmum shategy fv A omd B wu grea by, Ce Sa2( a \ ) hm &B Sp = \ y : ig J es 5 (xo) + (2s) a ON ‘Scanned with CamScanner pgafty OF GAMES jit. INTRODUCTION re ee erreean thal competitors and rivals always try to oulguess one another. This ts true of conte nthe marketplace, yale in a polite campalin or armies In a battlefield, e.g., oligopolistic ket st ra is a mutual Inter-dependence el ti 7 atts price cul oF an advertisement eamage er slebenlence of frns where every action mies a rection a quality change). In or : a ire theory has been developed, known as the theory of jeune ee erie Randle a behaviour game theory attempts to study “decision making in slualions where tu mst behaviour of {he opponent, the approach of game theory Is to se iunter-strategy F oown ‘best’ moves and lo formulate the appropriate e measi cane vo Heme are locked up tn a war to maintain ther markel sieee brea cece eee ees For wall iavite reaction from the second firm tn the nature of a price-cut. This will, In Leng. effect the sales and profits of the first firm which will again have to develop a counter-strategy to meat the challenge from the second firm. The game will, thus, go on. Game theory helps in determining the ‘best course of action’ for a firm in view of the expected countermoves from the competitor 's. A game is thus a ‘competitive situation’ where the market players pursue their own interests and no player can dietate the outeome Game theory came into existence in carly 20th Century. However. the mathematical treatment of the Game Theory was made available only in 1944. when John Von Neumann and Oscar Morgenstern published rook “Theory of the Games and Economic Behaviour. Author's approach was based on the principle of ‘best out ofthe worst, Le., he utilised the idea of minimization of the maximum losses. We now consider some examples of various types of situations where game theory can be used to analyse them, One of the most common examples in an industry is the management-worker bargaining over wage payments, conditions of work, fringe benefits, and so on. Other things being equal, what one gains, the other loses. Bilateral monopoly is an analogous state of affairs, involving bargaining betwcen a monopolisue buyer ofa product and a monopolistic seller. Market firms in which there are few, perhaps only two, (duopolistic situation) competing firms also involve in direct conflict. The price and production policies of one firm influence the sales and profitability, and therefore, the policies of other firm can even think of problems of conflict in political fields as real life examples wherein analytical skills of theory of games can be applied. For example, the conflict between alternative political parties or in a single party between alternative power centres. One of the important contributions of game theory to international strategy is to clarify the phenomenon of the exercise of military power and strategy without the use of actual arms. 11-2. BASIC TERMINOLOGY As an area of academic study, theory of games provides a series of mathematical models that may be usefull in explaining interactive decision-making concepts where two or more competitors are involved under conditions of conflict and competition. The models in the theory of games can be classified depending upon the following factors: 1. Number of Players. If a game involves only two players (competitors), then it is called a two-person game. However. if the number of players Is more, the game is referred to as n-person game. 2. Payoff. Outcome of a game when different alternatives are adopted by the competing players are called the payoffs. The game's payoff can be eXher (a win, lose or draw) or (winning or losing a sum of inoney). 3. Strategy. A strategy for a player is defined as a set of rules or alternative courses of action available to him in advance, by which player decides the course of action that he should adopt. Strategy may be of two types : ss ‘ s rele strategy. (a) Pure strategy. If the players select the same strategy each time, then It ts referred to as pure strategy In this case each player known exactly what the other ts going to do, Le. there Is a deterministic situation and the objective of the players is (o maxiinize gains or lo minimize losses. sen . ced. e aye a combination of strategies and each player always kept guessing Oe eee ee ae eee selected by Ute other player ata particular eceasion ten this 6 known 3S (0 which course of action is to be selected by the other pla ies 3 pal cul ocean et LS era ss mbked-stratgy Ts, there 1s probabil si a ahs faIns or to minimize losses. Thus, mixed strategy Is a se among ae ay which puts the player in the most preferred position, action or a epi Ategy. Any deviation from this strategy ‘0 or more intelligent and rational Instead of making inferences from the k to determine a rival's most profitable (4) Optimum strategy. A course of ol imespeetive of the nurateay of his competitors {s called an optimum strate Tesults ina decreased pay-off for the player. ‘Scanned with CamScanner OPERATIONS RESEARCH : QUANTITATIVE TECHIQUES FOR NANRGENENy OPERATIONS RESEARCH : QUANTITATIVE TECHQUES FOR MANAGEUEyy 9 . It Is the expected payoff of play when all the players of the game follow their optnnun suatjen The gue called fairifthe value of the game Is zero and unfair if itis non-zero. 5. Payoff Matrix. The payoffs in terms of gains or losses, when players select their particulars strategies (courses of action), can be represented in the form of a matrix, called the payoff matrix. Given below is the payoff matrix when firms A and B are competing firms and are trying to decide whether or not to undertake advertising campaign. FIRMB Advertise Don't advertise Advertise 20; 10 30; 0 FIRMA Don't advertise 12:16 10:5 If firm A advertises, but firm B does not, A makes a profit of Rs. 30 and B makes no profit; whereas if firm B advertises but A does not, then A and B make a profit of Rs. 12 and Rs. 16 respectively. 6. Two-person Zero-sum Game. A game of two persons, in which the gains of one player are the losses of the other player is called a Two-person zero-sum Game, Le., in a two-person zero-sum game, the algebraic sum of the gains to both the players after a play Is bound to be zero. Two-person zero-sum games are also known as rectangular games since they are represented by rectangular payoff matrix. Suppose player A have m strategies Ay, Ag, .... An and player B have n strategies By. By. .... By. Here, itis assumed that each player has his choices from amongst the pure strategies. Also it is assumed that player Ais always the gainer and player B is always the loser, ie., all payoffs are assumed in terms of player A. Let y be the payoff which player A gains from player B if player A chooses strategy A, and player B chooses strategy B, Then the payoff matrix to player A is : BB B, Ai} a a ain Aa] dx Gan On PlaerA :| : Am| nt Gna oO Thus in a payoff matrix, the number of rows of the matrix corresponds to the number of courses of action (choice) of the player A which is always written on the left of the matrix, and the number of columns corresponds to the number of courses of action of the player B which is always written on the top of the matrix. Such matrix always shows the payments or gains to player A. If there is a negative entry in the table, {t means that the payments are to be made by A to B. Thus for the player B, the matrix always represents the losses, while for the player A, tt represents the gains. Accordingly, the player A ts called the maximizing player since he would like to maximize the gains, while the player Bis called the minimizing player since he would like to minimize the losses. 7. (@) Maximin Principle. According to this principle, the player adopts a pessimistle attitude and plays safe, Le., his strategy is always that which results in the best out of the worst outcomes. In other words the Player A. (the maximizing player) decides to play that strategy which corresponds to the maximum of the minimum gains for his different courses of action. This is known as ‘Maximin’ principle. (ii) Minimax Principle. Similarly, the player B (the minimizing player) would also like to play safe and in that case he selects that strategy which corresponds to the minimum of the maximum losses for his different courses of action and this is known as the ‘Minimax principle. In other words, the maximizing player adopts the Maximin principle, while the minimizing player adopts the Minimax principle for the best of his plays of the game. 8. Characteristics. A competitive situation is known as a ‘Game’ If it has the following characteristics : (0 The number of competitors called players is finite. (i The players act rationally and intelligently. {t Each player has available to him a finite number of choices or possible courses of action called strategies. The number of choices need not be the same for each player. ‘Scanned with CamScanner Gis) isles of cach player and the amount of gain or loss on in player in advance. 1 (strategies) simultaneously. ‘ation. THEORY OF Games THEORY OF GAMES (0) All retevant information, te. the ered strate idual’s: move (strategy) are known (0 € {9 The players setect their reapeetive courses of act Leumi {u0 The players make individual decisions without direct 60 the minimizing paver ties to many (oo The Maximizing player attempts to masimnlze Nis gals ard Zing player tries to minimize his losses meveratieml (vii) The payoff is Axed and d ed In advance. of certainty are as follows ay DECISION MAKING Under certainty. Under risk. Under uncertainty Non-competitive- Competitive situations. situations (Game Theory) Pure strategy Mixed strategy: 4 (Saddle point exists) 2x2 Strategies game 2xm orn x2 men Strategies game Strategies Arithmetical method L t Graphical Method Linear Programming method 11-3. SOLUTION METHODS OF PURE STRATEGY GAMES (With Saddle Point) {acase of a pure strategy game, the miximizing player arrives at his optimal strategy on the basis of the toked wncrierien. while the minimizing player's. strategy Is based on the minor The game is fames eng the Maxtmin value equals minimax value, The dilicrenee bevween pure and mixed strategy asad that pure strategy games possess a saddle point whereas mixed strategy games do not. The presence see point is identified by comparing the values of each player's maxi sts ey. aie us lake an example, to illustrate the 2-person zero-sum pure-strategy game. Suppose Firm A has two towne ne. Firm B has three Strategies. These strategies and their corresponding payoif are as given in the Frm B Row By B By minimum a 2 Ri A 2 18 ma Ay 16 10 ee S-Maximin | Column Maximum i 18 8 = Minna | In th layer and Firm Ba minimizing player. Now. if Firm A chooses ear Ue Frm As dhe mening player and ry nnn lye, Now ses ETE ce ose Ba as 8 coun ill choose By as a counter-strategy, thus giving Firm A a payolf Malt aso. one ses Peace osition for Firm A would be the latter strategy where it gets a higher So, the most-preferr ‘Scanned with CamScanner OPERATIONS RESEARCH : QUANTITATIVE TECHNIQUES FOR MANA ENT From the point of view of Frm B.strateny B, dominates the stentay Be a au yalues of Bs column sr es than the B, column, Thus, effectively Firm B has on 8 a - Now, 1 anter strategy Ag. and Firm B will lose 16 units. On the other hand, i CHOSE BE Fin Av aan choose Ay but the toss to Firm U this Ume equals 8. Thus, the most prefect position for Firm Bis when it employs the second strategy, vlz., By. The extensive calculations of moves and counter-moves of the competitors, as discussed above, can be simplified with the help of maximin and minimax eriterta. This erlterla Includes : (a) Finding maximin value (for the maximizing player (@ Find the minimum value in each row (shown as row minimum In above table) denoting minimum possible game from each strategy of A. (i The maximin value is the maximum of these minimum values (value 8 in the row column of above table). (8) Finding minimax value (for the minimizing player) : (@ Find the maximum value in each column (shown as column maximum in above table) denoting minimum possible loss from each strategy of B. (i The minimax value is the minimum of these maximum (value 8 is the column maximum in above table}. (6 Saddle (or equilibrium) point. In case the maximin value equals the minimax value, we get a saddle point. The saddle point is the solution or value of the game. Once saddle point exists for a game, we can then say what the optimal strategy would be for two players. (In our example, it is Ay for Firm A and By for Finn 5B) Thus the strategies of A and B corresponding to saddle point are the optimal strategies of A and B. ‘We can summarize the analysis of pure strategy games as follows : 1. Develop the payoff matrix. ‘ 2. Identify row minimums and select the largest of these as player one’s maximin strategy. 3. Identify column maximums and select the smallest of these as the opponent's minimax strategy. 4. Ifthe maximin value equals the minimax value, the game is a pure strategy game and that value is the saddle point. 5. The value of the game of player one is the maximin value, and to player two, the value is the negative of the minimax value. . Remark: tis not necessary that every payoff matrix will have a saddle point. Saddle point may not exist ina game Example 11-1. For the following payoff matrix For ‘firm A, determine the optimal strategies for both Firm A and Pirm B and the value of the game (using maximin-minimax principle) + onies J Firm B 3-104 6 7 8 2 4 FumA 6 8 6 4 12 1M 4 2 Solution. It is evident from the given matrix, that Minimum of frst row =min (3, 467) Minimum of second row = =imin_ (cl, 2 4,12) Minimum of third row min (16, 6 14. 12) Minimum of fourth row min (1, 4 2) . Maximin value =max (1, 6-4, Again, Maximum of first column =max (3, -l, 16, vy Maximum of second column =max (-1 8 8 411) Maximum of third column = =max (4, 2 6, 4) Maximum offourth column =max (6, 4, 14.) Maximum of fifth column = =max (7, 12) 12, 1} ‘ Minimax value min (16, 11,6, 14, 12) ‘Scanned with CamScanner HEORY OF GAMES ee ste ine Pia te ake thus the optimal strategy for Firm A ts its third strategy Te sadtdte Paint ew an Bis ite thie togy. Ne optimal strategy and the opumal mount in the saddle point position {8 also called vatue of the game. For this game, value of the ‘The Payot miayer A. The value of game 18 always expressed fromthe point of view of the player whose game 16 6 fe Heted in thte rows. Strate ve is strictly determinable. Also, since the value of the Frompte 11-2. Solve the game whose payoff mame given by : ar Player B's Strategy yy BB By Ay 16 me Is not zero, the game {s not fair. “60 560-58 Payer A's Strategy Az | -20 28 ig pg As 24 8 o 24 . ving the rule of finding out the saddle point, we obt Solution, Apr Bectangle a8 showin bo ‘ain the saddle point which is enclosed II] below ; TABLE 11-1 Player A Player B Strategies Row Minimum| Strategies By B By By | t | A 16 -60 5 se | | | Ay -20 28 -18 24 Ay 24 -8 ° 24 ey | | Column Maximum Bl We observe from the above tal J ble that there exist two saddle points solution to the game is given by : at positions (2.4) and (3, 4). Thus, the {0 The optimum strategies for Player A are Ay or Ay. (i) The optimum strategy for Player B \s By, and (ti) The value of game 1 Is -24 for Player A and Player 24 for B, Since v0, the game is n \ot fair, although it is strictly determinable, aample 11-3. A company management and the labour union are negotiating a new 3 Each of these has 4 strategies : 1. Hard and aggressive bargaining I. Reasoning and logicat approach 1M. Legalistic strategy IV. Coneiliatory approach Costs to the company are given for every pair of strategy choice years settlement. Company Strategies 1 u a 1 20 Union Strategies It 25 m 40 12 M4 8 2 10 5 v 5 4 Mn ° What strategy witt the two sides adopt ? Also determine the value of the game. elution. We first find out the saddle point by encircling each row minim, 1 and putting squares around (ich column maxima. Thus, we obtain the saddle point which is enclosed by a circle and a square both as Shown below : ‘Scanned with CamScanner OPERATIONS RESEARCH : QUANTITATIVE TECHNIQUES FOR MANAGENEW, OPERATIONS FESERR aS eve© VV oS oaaS—aOOMAEMENT Company Strategies 1 " Mm wv Row minima 1 20 15 @) 35 Ga) }—Maximin 1 25 ry ® 10 ® Union Strategies m 10 5 ® ’ "s+ |e Column mama [Go 7 3S T Minimax Since Maximin = Minimax = Value of game = 12, the company will always adopt strategy III—Legalistic strategy and union will always adopt strategy I—Hard and agergssive bargaining SELF-PRACTICE PROBLEMS (200. —<—<—<——d 1. Solve the following games by using maximin (minimax) principle whose payoff matrix are given below. Include in your answer () strategy selection for each player, (i the value of the game to each player. Does the game have a saddle point? Player B Player B i) BL Bh By (h By Bm By By Al 10 5-2 Al -5 3 1 10 PlayerA Az] 6 = 7 3 | PlayerA | 5 5 4 6 4, 4 8 4 | Al 4 2 0 55 2. Solve the following games whose payolf matrix is given by fa) Player B (ib Player B BB, BB BB, BBB a[to7 3 4 are 3 tao Player A A|5 6 4 5 Player A 2 7 a l7 2 0 3 mje 43 3 8 ALS 6 2 21 @ Amp c) BB BB By A c ApS + 4 6 7 roa 8 All 8 2 4 12 un) -2 4 rma Pl 8 6 ue Fim X | og 3 Allow 4 2 1 wl 6 2 3. Given below ts the pay off matrix in respect of a two-person zero-sum game : B's strategy By B, By By Bs A 8 10 -3 -8 -12 Ma 3 6 O 6 i A's strategy Ag 7 5 2 8 17 a, | -1l 12 -10 10 20 As 7 0 0 6 2 ofthe [a Wate down the maximin and minimax strategies. (i I ia strictly determinable game ? (i) What is the value of game ? (iv) Is this game a fair one? ‘Scanned with CamScanner yay_OF GAMES ume that two firms are competing for market share for a particular product, Each firm 1s considering what 4. ional strategy (o employ for the coming period. As H60! ae ioponsinn 5 3 no Bee wees tortion ‘ $ 2 {9 Which firm would be winner, in terms of market share? {a Would the solution strategies necessarily maximize profits for either of the firms ? 5, Find the optimal strategies for two stores from the following payoff matrix showing gain or loss of customers for store I. Action of Store ¥ A B c 1 ° 20 -60 Action of StoreX 30 -10 -20 m 70 -80 -30 6. Two Firms A and B (manufacturers of dk letergent powder) are planning to make fund allocation for advertising their lucts. The matrix given below shows the pei Feentage of market share of Firm A for its various advertising policies, T Fim B | | Strategies No advertising Mediwn advertising Large advertising | ! No advertising 60 50 40 | | mms Meduunaaresng 0 = | | Large advertising 80 60 50 | Determine the optimal strategy for Firm A. 7. Following is payoff matrix in terms of increase in votes to X loss to Y) using three different strategies available to wa ae eae Sd¥erusing. Find optimal strategy to be adopted by X for the campaign and the number of voles X aah sary with this strategy. Candiate Y Strategies [~ 1 u 1 A 300 200100 Candidatex Bp 600 = 500.400 c 600 400-600 8. Find the saddle point and the value of game for each of the following two games: Player B Player B (a BB, By Be Dh B BR BB, af[9 3 1 8 0 A o 0 5 3 A 2 Payer a | 6 5 4 6 7! prayer a at 2s A|2 4 3 8 B A} -4 -3 0 -3 6 als 6 2 2 1 AL 5 1 -5 -2 -6 ©. Determine the range of values of p and q which will render the entry (2, 2) a saddle point for the game : Player B B, B, By a[2 4 7 Player A Ag | 10-7 As|[ 4 Pp 8 ‘Scanned with CamScanner OPERATIONS RESEARCH : QUANTITATIVE TECHNIQUES FOR ANGE HINTS AND ANSWERS eee : value of game = 4 smal strategies : (Ay, By). val sentes: 2 lOptna steteien Uy vale of kane = 3 (9) Optima states: Ay (c) Optimal strategies : (Ap, B,). value of game = 6 . nin strategy : Ay, minimax strategy : By ; * ate because a saddle point extsts at (Ap, 9), (i) zero: (le) Yes. bare the game value is zero. 4, Optimum strategy for both A and Bis ‘much proportion’, value of game = 1 5. Optimal strategies : (Il, Q, value of game = - 20 os 6. Optima Strategy for both Firms A and B ts ‘Large advertising’ and value of game = 50. fame = 600. 7. Optimal strategies : (C, N. (C, II): Value of game = 60 - 7 - 8, (a) Optimal strategies : (A,, B), value of game = 4; (b] Optimal strategies : (Az, B). value of game = | 9. Saddle point will exist at the position (2, 2) only when ps7 and q> 7. 11-4. PRINCIPLE OF DOMINANCE Has Gometimes reduce the size of a game's payoff matrix by eliminating a course of action which iss» inferior to another course of action that it can be left out of the set of choices. Such a course of action Issa] fo ,be dominated by the other. The concept of dominance 1s especially useful for the evaluation of two.pest zero-sum games where a saddle point does not exist. In general the following rules (or properties) of dominance are used to reduce the size of payoff matrix: rere iy the elements in a row (say, ith row) of a payoff matrix are less than or equal to the SehessPonding elements of the other row (say jth row) then the player A will never choose the ith strategy or in other words the ith strategy is dominated by the jth strategy, Rule 2. If all the elements in a column (say, rth column) of a payoff matrix are greater than or equal to the Staton ore glements of the other column (say, sth column) then the player B will newer chee hon Strategy or in other words, the rth strategy is dominated by the sth strategy, Rule 3. A pure strategy may be dominated If it is inferior to average of twwo or more other pure strategies. In short, the Dominance Principle can be summarised as follows : By). value of game = 4 1. When all elements in a row of a payoff matrix are I another row, then the former row is dominated by the I 2. When all elements in a column are greater than or by the latter equal to the corresponding elements in — colu:nn, then the former column is dominated by the latter and can, therefore, be deleted from the) matrix. | Example 11-4. For the following game, find optimat principle of dominance: strategies of A and B and value of game usitd Player B By By By 6 Player A a o Solution. Comparing strategies A, and Ay, A, glves more a ol " gain than Aj in all conditions (for all strated! 5). Ay dominates over Ay. Hence we can ignore Ayand the efter Pay olf matrix shall be as shown bel? Player B Strategies Ay PlayerA Ay Ay ‘Scanned with CamScanner con OF GANS — inthis above reduced matrix By Is dominating By, 80 By Is redundant and tgnoring this, the effective payoff spatri will Be : Player B Statens By, ow minum Ay @ A “4 A 2 Column Maximum 10 " Hence we find that the saddle point exist and A’s optimal strategy 1s A, whereas B's be B, and value of game ts 7 6. It implies that A cannot gain more th below ¢ 6 in the case. Example 11-5. Following is the payoff matrix for player A : Player B oon ow optimal strategy will ran Rs. 6 and B cannot reduce his loss Player A eels Using dominance property, obtain the optimal strategies for both the players and determine the value of the game. Solution. Player B Steps: 1. Since all elements of column I are I tess than or equal to respective elements of Player At Column IV, we can delete the column IV. m v 2. Now as all elements in row IV are less than [2 5 : ; rorpeciive clements in row Ill, we can delete the 4, ; : 3 : Player B t Ls mM less than if-2 7 3 3, Now ment of column I is m| 6 ; : ‘espective element of row Ill, we can delete both IL are less than respect 4. Now as each element of row I and row the rows I and I. Player B oom Ww PayerA ‘Scanned with CamScanner OPERATIONS RESEARCH : QUANTITATIVE TECHNIQUES FOR MCE ‘Thi imum 5. Now in this row (row ad n a Bplays |, ie, A =Iland B= 1, ana apy Value of game fs then = 6. Optimal strategy for A= III, B=1 docs not exist. Prior to applying do, . Dominance property ts used when the saddle point rm property, find the maxinunn aid tino test whether Ue problem Is in equilibrium (Ce., stable the saddle point 's when A plays mg. nance with S saddle point). Player B rT nr ure Vv Vv Row minimum ho 4-3 8 4 ]2 "ss 6 3 7 8 | 3. Maximin=6 Player A vs ™)e 7 9 8 7 |( Minimax=6 Mla 2 8 4 3]fa Column maximum. [6] 7 9 8 8 Saddle point exists. The value of the game is 6. A will choose strategy III and B strategy I. Beample 11-6. Using the principle of dominance in Game Theory and solve the following game : Player B bbe by by aps 10 9 14 PlayerA a2! 10 11 8 12 %lI3 12 14 13 Player B by bp by by af 8 10 9 14 PlayerA @/10 11 8 19 HL1IZ 12 14 13 Im the above game, b, 1s dominated by by. Eliminating by, the reduced matrix is : Player B by by bs a 8 10 9 Player A ag [ 10 paT :| lis 12 14 Strategy a, is dominated by ay, so elimination of 4 reduces the matrix to : Player B by by by 4110 11g @|13° 12 14 Now, we see that averages of by and by dominate B, as is set out below : Player A Player B b, Btbs 12 flo 95 Player A lis 13 cal ‘Scanned with CamScanner THEORY OF Games G3] —SSLOF Games Eliminating by, the matly reduces (0: Player B ba by %I 8 PlayerA alia 14 is 2 x2 game. This ts now manageable as Ht ts 2 x 2 Player B by by Row minunea Hil og 8 Payer tlie By ® Column maxima G2] a4 (a3, bs) is the saddle point. Hence, players. 1-4-1. Nash Equilibrium. All games do not h matrix given below, Firm has no domiin advertises A must advertise, while If B deces value of the game ts 12 as tt presents the best payoff for both the ave a dominant strategy for ever layer. In cas ff nt strategy. Its best decislo von wheat Fae Payor spends on what Firm B does: if B SS NOt Lo advertise, A benefits by now advertising, PAYOFF MATRIX : ADVERTISING GAME fom B Advertise Don't Advertise Firma Advertise 30:0 Don't Advertise 40:5 to advertise as it gains A equilibrium is a set of best it can. given the strategy of the ceaonents. Since player is satisfied that it has nag ible, he has no incentive to deviate m the chosen strategy. Thus, The distinction between Nash rategy equilibrium may be well understood. 'n dominant strategy case, gach player chooses his best strateyy, irrespective of the Strategies of other players While in case of Nash equilibrium cach player chooses a strategy that is his best choice, subject to what gles the opponent chooses. teonipe chat al players Asis obvious from the discussion so far, Nash equilibrium ts possible on ly if we assume that all player: “nderstand the ‘ame ona are rational. This may not be true always. If you are cautious and are also . owing the game or is likely to be irrational then you may ‘tncerned that ent may not be fully I then y ‘hoose a sate enema 2 os te minimum gain that can be earned, known as maximin strates "5. SOLUTION METHODS OF MIXED STRATEGY GAMES Games Without Saddle Point) : equa imum of column @., maxtnum of row minima is not equal to minimum o , Pacing) apis of game where saddle pened sifateny roblems. In order to solve such a nme, each player nist cannot exist are taken and starts to play in a random manner and in auch a way that hs ave age Fant gn ConceDt of chance move an f the game should be optimal, even though he may lose mor in any iidad pase Buber of Pither words the player adopts a certain strategy ving probability fr wi dhe shee soe ste ving naan 2 eae hand then shifts over 1m of his all proba 2 a ch that the su ; > one and pay sein probability ps and 30 ame becomes optimal (ce., maximum for the maximizing player a gg Be number of plays of the gal 'm for the minimizing player). ‘Scanned with CamScanner LOUANTITATIVE TECHNIQUES FOR MaNAgey, RESEARCH + Eur ‘OPERATIONS A mixed strategy game eat methods : nd 3. Lincar Programming Method, y thor rec by followit oF Graphieal Method e solv game with the following pa 1 Algebrate Method 2. 00 person zerorsuin game with the foolowing payoff mata ; 115-1, algebraic Method, Consider the Wworpersn ZTE », B, Probability strategy | BL a aa ‘ nl 1- Player A an, an p ’ t Probability | yest elements of the mat mi iS gar is to have no saddle point, the two larges rategies. Our task is to aioe Rees ll ea pers ese Ste ae the probabilities with which both players choose thelr co ‘Ay with respective probabilities p and 1 — pang let Player B via iy his, shategles Bi and By with respective probabilities q and 1 - q (since the sum of two 1 Of the termine B's strategy A's Expected pay-off | ‘The expected payofls to player A when B B, plays any one of his strategies By or By up +4 (1~ ph throughout the game, are given by: a apt on (1A Now in order player A is unaffected with whatever choice of strategies B makes, we must have P+ (1 -p) =a2p+ag(1- p) => - zp ~ aay Pay = aya) + (aaa Gy) ‘Similarly by and equating the expected payoffs of the player B, makes, we have * py ps Ad +2 a= anq+an(1-q = = Qo~ ay 9° (ayaa) + (aa aad ‘The value of game v is found by fA eg. V= ap + ay (1- and ‘The solution of the game ts : A play's (p,1-p), where p = B play's (4, 1-4), where q = and value of game y = Summary of Mixed Strategy summarized below : 41) aaa ~ yp ay, (au ~ aa) ¥ (an aia) Games. The Ways ~ aya) + (a22- ag,)Ip Oy - aia l-p= Tar Gia) + (ag ~ ayy) = aay ana (A= Gal + (aga Substituting the value of i P| which on substitution and reat rrangement Weoee rangement becomes : a= Oy n> a) (aa) (4u- @a)+ (aa- aad pita =a, feu aa) Can ay Az ~ a, for whatever choice of strategies player A Wan ~ aan) + (aa = aiallq = ay2- a2 ‘pressions for the expected gain 1, Set up the payoff matrix, 2. Apply the maximin criterion to strategy 1s netded. 3. Subtract the from the larger one 4. Interchange each of these patrs of subteseted numb, Procedure See the game has “8 found tn Step 2 above, for analysing 2 2 mixea strategy games '5 @ saddle pot xed smaller payoff in cach row from the Point. If it has none, than a mis larger Ser one and the smaller Pay-off in each colt! ‘Scanned with CamScanner eon OF GAMES 5, Put each of the interchanged numbers over the sum of the palr of numbers. 6. Simplify the fraction to obtain the required strategies. 7, Value of the game can be found either by looking at the game from A’s point of view or B's point of view: Example 11-7. Solve the following game : Payer B a Ai [256 Payora 4! [78 a Solution. In the given matrix, since there 1s no one value which ts smallest in its row and largest in the columnn. there fs no saddle point in it, and the game is not that of the pure strategies. Le.. neither the player Anor player B would play the same strategy throughout the game but they would resort to what is known as mixed strategies. The plaver A would play his first strategy A; with probability pthen he would shift over to his second strategy Az and would play with it with probability (1-p), since the stim of the two probabilities is one. Similarly if the player B selects his first strategy B, and play with it with probability q. then he would play second strategy Bz with probability (1-q) since the sum of the two probabilities is equal to one. in the given payoff matrix, let the player A play his first strategy A, with probability p. then he would play his second strategy Az with probaility (1 — p). Suppose now that the player B selects his first strategy B, and plays with it throughout the game, then the expected payoff of player A for this game would be the algebraic sum of the expected values of the two events, Le., 25p + 10 (1 — p) But if the player B selects his second strategy By and plays with it throughout the game. then expected payoff to player A for the game would be 5p + 15 (1-5). B's Strategy Expected payoff to player A ‘This can be put up in adjoining tabular form : B, 25p + 10(1-p) B 5p+15(1-p) Similarly if the player B selects strategies B, and B. with probabilities qand (1 —@) respectively, then the Gected payoif to player B, when the player A adopts his strategy A, and plays with it throughout the game. tyuld be 254 + 5 (1 ~q), and the expected payoff to player B, when the player A adopts his second stratexy As throughout the game would be 10q + 15 (1 —q). A's Strategy _ Expected payoff to player B 7 inhi Which may also be put in the tabular form as : A 25q +5(1-q@ Ay 10q+ 15 (1 iow in order that the player A may be unaffected with whatever choice B makes, the optimal plan for the Player A should be such that the expected Pay-offs for each of B's strategics should be equal, Le., 25p +10(1-p) = Sp+15(1-p) = 15p+10=15-10p and 1-p=1 Hence, the player A would play his first strategy Ay with probability and his second strategy Ay with Probabuity 2 Similarly by equating the expected pay-offs of the player B, for whatever choice of strategies player A ikes, we have 25q +5(1-q) = 10q+15(1-q) = 20q +5=15-5q s q=3222 and 1-q=1-2=3 2 2 respective! Hence, player B would play his strategies B, and B, with probabilities 3 and respectively. ‘Scanned with CamScanner OPEMATIONS ACEEARCH : QUANTITATIVE TECHNIQUES FOR WAAceyRyy —eeoMO9 ENT ‘The value of the game Is determined by substituting the v player A or Bas shown below : Value of the game = Expected payoff to player A © of por qin any one of the expected values tn 25xbsi0x3 3 or (i Sxb 415x413 Similarly value of the game = Expected payoff to player B E 2 is, (0 25x2 45x32 =13 or (id 10x2 + 15x 13 Thus the value of the game fs equal to the expected payoff to player A or the expected payoff to player p, ‘The value of the game, v= 13, Example 11-8. Two firms are competing for business under the conditions so that one firm's gain is another firm's loss. Firm A's pay-off matrix is given below Firm B No Medium Heavy advertising advertising advertising No advertising 10 5 2 FumA — Medium advertising 13 12 15 Heavy advertising 16 4 10 ‘Suggest optimum strategies for the two firms and the net outcome thereof. Solution. Clearly, the first column is dominated by the second column as all the elements of the first column are greater than elements of second column. This eliminating first column, we get Firm B Medium Heavy advertising, By advertising. By Noadvertising a, 5 2 Pima — Medium advertising Ay 1215 Heavy advertising Ag 410 Again, first row is dominated by second and third row as all t respective elements of second, and third row. he elements of first row are less than the Hence eliminating fir matrix. ‘st row, we obtain the following 2 x 2 pay-oif Firm Medium Heavy ea advertising, By advertising, By fedlum advertising, A 5 hima Ay 12 15, Heavy advertising, Ay M4 10 The reduced 2 x2 map Semepe 2 x 2 payoff matrix also does not have the saddle point. Thus, both the Firms A and B use For Firm A. Let p; and ps be probabilities of sele the expected gal st 12p, + 14py = Spa 10Ps = 12p, +14 (1 - ps) = 15 pes 10(1 - py) “ 7Pa For Firm B. Let q, and q; be probabilities of selecting stra Bs ig Strategies By (Medi ind strategy Meawy advertising) respectively. Then the expected loss to Firm B aay nat ges to Boas B, strategies should be : 4ope$ and Pre iapy=d. 12q9 + 18q3 = 14q) + 10455 a2 + qy = 1 12q2 + 15(1- qa) = 14q + 10 (1 - gq) ‘Scanned with CamScanner sto Firm B: ; pected (05 124150 212.55 15.2 Pt 1x pce INE optimum strategies for the two firms ar 7 to Medium ae advertising advertising gg !@A™ an > 7 ertising ‘ i No Mectiny ium advertising advertising qa idvertising 2 oO 5 Se [ 5 | sod the value of the game is V = & FimB B By Ag [ 12 15 14-102 44 pm | 10=4, Pla) =a 3°97 4 Tie 10 15-12=3,7, 15-10=5 } Pad 5 PB.) = 55377 + Hence, Firm A should adopt strategies Az and Ay with 57% of time and 43% of time respectively, (or with sand 43% probability on any one play of the game respectively). Similarly, Firm B should adopt strategies and By with 71% of ime and 29% of time respectively (or with 71% and 29% probability on any one play of game respectively). Expected gain of Firm A Expected gain of Firm B ] a 2xd+14x 3-2, Firm B adopts By) (9 12x3+15x3=%2. Firm Aadopts A to (@ 15x44 10%3=% , Firm Badopts By) 143+ 18xF=P. Firm Aadopts A Example 11-9. Two breakfast food manufacturing firms A and B are competing for an increased market we. To improve its market share, both the firms decide to launch the following strategies : ‘A;. By = Give coupons ‘Ap, Bz = Decrease price ‘Ag, By = Maintain present strategy Increase advertising , B. The pay-off matrix, shown tn the following table describes the increase in market share for firm A and “ease in market share for firm B : By Be By By A 35 35 25 5 a, 30 20 15 0 Firma 10-50. a $60 10-15 nd value of the game. Semin the optimal strategies orcach em i a. oun nthe problem The gtlution. The first step Is to search for 4 Sl) ‘ye reduced in size by rules of dominance. Since each ni Step ‘ts to observe If the payofl Mresponding ‘elements of Second row, second row is dominated by °f second row is less than the ‘Scanned with CamScanner OPERATIONS RESEARCH : QUANTITATIVE TECHNIQUES FOR MANAGE y OPERATIONS RESEARCH + QUANTITATE Teens ROR MANACENSH irst row because payolls are less attractive for firm after deleting the second row, the reduced mat rh Ing the sec y A, first row because payoffs ar FumB By 5 10 15, c more than the corresponding elements in frst ach clement of second column ts I antyeraion eecond column fs dominated by fst calumn because payoffs are less attractive {more Ipc lumn, therefore ‘ohun 3 fst column bec ; fr B. Thus after deleting the second column, the reduced matrix become: Firm B By By By A, [35 25 5 FirmA 43 |40 010 Avl55 10° 15 Further compare rows 2 and 3 and then columns } and 2 and delete the less attractive row and column from a's and B's point of view. The reduced payoff matrix is as shown : FimB By By Prob. 25) py Fi 1 A aAlio 15 | Pe Prob. py py 8 Seetee 2 * 2 payoff matrix also does not have the saddle point. Thus, both the firms A and B use mixed strategies. For Firm A: Let py 1 and p, be probabilities of sel advertising) respectively. by lecting strategy A, (Give coupons) and a, (Increase ‘Then the expected gain to Firm A when Firm Buses its By and By strategies is given 25p, +10p, and 5p, + 15p, : py + pp=1 For Furn A, the probability p, and Pz Should be such that expected gains under both conditions are equal, te. 25P1 + 10P,= Spi +15p, = 25p,+10(1- PY=5p,+ 15(1-p) * 25p, =5 or p, = and am Pel-p=s For Firm B : Let q, and qz be probabilities of selecting strategies By (Maint: Moerease advertising) respectively. Then the expected lose finn Buen, firm A should be + 5a2 = 109) + 18q5 q+ q2=1 25q, +5 (1-q)= 10q, + 15 (-~q) 25q,=10 or q at Hence optimal strate coupons) and strate present strategy) ani ‘ain present strategy) and By ses Its Ay and Ay strategies 25q, o and q@ = Bles for both the manufacty BY Ay (Increase advertising) 20% id strategy By (Increase a The expected gain and loss to firm A a1 Expected gain to firm A: 25 p, 5 p+ 15p, Expected loss to frm B: 254) + 5q_ = 25(2/8) + 5 (9/5) 10qy + 1542 = 10 (2/5) + 15 (3/5) = 13 urers are that. firm A should adopt strategy A (Give Ume while fi in m B should adopt strategy Bs (Maintal \dvertising), 40% time. " “ ind firm Bean =13 ‘Scanned with CamScanner THEORY OF GAMES THEORY OF GAMES Example 11-10. A steel company (s negotiating with {ts unton for revision of wages to its emy loyees. The management. with the help of « mediator. has preparod a pay off malroe SO aes ‘olus sign represents wage increase, while negative sign stands for wage decrease. Unton has also constructed a table which is comparable (0 that developed by management. The management does not have the specific knowledge of game theory to select the best strategy or (strategies) for (he firm. You have to assist the management on the problem. What game value and strategies are available (0 the opposing group ? Additional costs to Steet Co. (Rs) Union strategies uy uy Us Us a 42:50 42-70 4350-020 ‘steel Co, strategies © +2:00 +160 +080 40:80 +140 +120 +150 +190 co +3:00 +140 +190 o Solution. Since the company represents the ‘minimizing player’ and the union the ‘maximizing player’. the given payoff matrix is recast as follows by interchanging rows and columns. Union Company strategies strategies q G G a uy 250 2-00 1:40 3:00 wy 2-70 1-60 1-20 1-40 Us 3:50 0-80 150 1-90 uy 0:20 0-80 1:30 ° This game obviously has no saddle point. Since all the entries in the third row of this matrix are greater than, or equal to the corresponding entrics in the fourth row, the fourth row is dominated by the third row and hence can be deleted. Thus, we get a & & cs UY [ 250 2-00 1-40 3-00 ] u,} 270 1-60 1-20 1-40 u; | 350 0-80 1-50 1-90 G & In the above matrix, the first column is dominated both by the second Uf 200 140 and the third columns, and the fourth column is dominated by the third Us [ 1-60 13 column. Deleting the dominated columns, the matrix is reduced as : Us oe0 aa & &% Further, the second row is dominated by the first. Deleting this row, we yy 2-00 140 fet the adjoining matrix of the order 2 x 2, and obtain the solution to they, | oes. 10 | fame analytically. Ifp, be the probability with which the union adopts policy U, and q be the probability of adoption of Cy the company, we have 1501-40 2x15-08x14 _ U8 pan 1 - = Pree eit arts Wee ran Iss aNd VET TSS + FAI 1800 22 9); for the union it is 0, $j. 0) and the game Thus, optimum strategy for the company 1s (0, "alue is 188° (representing increased wages) eee 1. Use dominance principle to solve the game whose payolf matrix is given by Firm B Player A si on om trategies by by by bg bo 5 af[4 6 5 10 7 I 5 7 9 @ Fmaale 7 4 8 9): 4 Player B hn -30 2 -L 4! g 8g 10 9 8 -4 -8 -10 ‘Scanned with CamScanner OPERATIONS RESEARCH : QUANTITATIVE TECHNIQUES FOR MANAGE, Strategies bj by By Dy bs af2 2 3 5 6 - aml 4-2 5 4 38 ‘ PaverA a) 7 3 4 6 8 ala 2 4 3 3 2. Find the optimum strategies of the players and the value of the game from the followign payoff mat, ‘two-person game : by 4 5 6 1 TH concerning Player Y Strategy! Strategy It Strategy [4 1 (a) Player Xstrategy | 2 3 Player B Company P B, By Py Pa A[7 3 5 [8 + wy raver aS [7 3] lacompany 5 & [s 4 + For the following two-person zero-sum game, find the optimal strategy for each player and the value of the game; Player B Player A bb by A A AS af o -1 2 B[-5 -3 1 (Q PlyerAt|-5 2 4 |. (b Players B) 2 -1 2 ™\-2 -3 4 4. Use the ‘concept of dominas the optimum stategies and the val lue of the game : Player ¥ Svateges 1 ol iil rf2 2 4 PlayerX Ul 4 3 4 Mis 1 og ince" to reduce the followin, Vv -2 Bl-2 3 4 v -3 18 game to 2 x 2 game and then, from the reduced game. find Union position By By By Adamant Compromise Yield Ay Negotiate { _» 1 2 Ag Delay 5 2 3 (9 Solve the management's e if Zobe tg managem en's problem, ( What should be the union's Strategy ? (ii) Discuss the implications of 6. Find optimal strategies for two companies matr showing gain or loss of carternacy {0 ComPantes competing with each other on the basis of the following pay-off — Company B Company A Quantity Discounts CSR ere Events Sponsorship Quantity Discounts 20 75 30 CSR Initiative -10 ae 28 | Event Sponsorship 40 50 30 (0 State the required assumptions for finding optimal strategies for two companies. (4 Find out value of the game. use dominance rule, ‘Scanned with CamScanner THE producing a new toy under Hence f 2.9 compe snufetrer are prodeng 0m 4 paten olde, dumm they have te pon of runnin the plant fr pte ata ay trae does the cost. One of the manufacture . has se day veh below, in whieh m the market that eco enpture anid matntain the diferent production schey ts 1e est Manufacturer § —— In order to meet the roduction increases so limates the percentage eer Manufacturer A Si? 8 hrs 60% 2:16 hrs om Sy: 24 hrs 83% —_——__ | (9 Atwhich level should each produce ? (i What Percentage of the market will B have? HINTS AND ANSWERS 1 (a) Player A should strategy as, player B should strat (6 Optimal strategies : 2. (a) X= (1/4, 3/4), (2/4, 2/4), v. 12, (b) A: (1/5, 4/5), 15, 3/5), v= 23/5, (dS: (8/5, 2/5), P:12/5, 3/5), v=~ 11/5 S (dA=(7/8, 1/80). B: (8/8, 5/8, 0), (9B: (0, 5/8,3/8). A: (478, 4/8, 0) {9 X:(0.2/7.5/7, —¥:10,5/7,0, 2/7, 0, 7. (0 Optimal strategy : (S,, Cy, (@ At the optimal tevel of 24 hours per day, 11'5-2. Graphic Method for 2 x n and mx 2 Games. Since the optimal strategies for both the players assign sarge Probabilities to the same number of pure peasietts. Ht 8 obvious that if one player has only tee Srarsties the other will also use two strategies’ Grashinet method fs helpful in finding out which of the two Suafeeies can be used. It is also useful if the nature of the Game Is of the form 2x n and m2. The graphic method consists of two graphs (0 the payoff (gains) available to player A versus his strategies options, and {f the payoff (losses) faced by player B versus his strategies options. ice at the level of 24 hours per day. Bwill have 60 per cent of the market share. Consider the following 2 x n game which has no saddle Point and whose payoff matrix is as follows ; Player B BB Probability A fan a... nm An lay am oan] op Probability gq, gyn B's Pure A's expected payoly strategies Rayer A has two strategies Ay and Ay Ifpy and py ate the y2 pr a (1 = py) = (aq, an1) Py +0, Fiebabilites with which the player A uses ie pure B, mutesies. then py + py = 1 (py. Pa 2 0) 90 that pa = I~ Ph , 42 Pi + Gant ~ pid = (axa ~ Ona) r+ Op, objective 1s to determine the ‘optimal value of p, and Pe ° " to pO" each of the pure strategies By, Bayo» By availabe taphlwver B. the respected payoff for player yy Pa + an (1 ~ Pid = (Qin a) Pr + On ‘ulated as ‘shown : a n e4 A should select the value cer its ike may be dove by plotng the following u,Atcording to the maximin or ata. ‘expected payoffs. This may wean’ Pr So as to. maximize ‘ stags) mem, lines : du + oan Be = (O12 ~ Gas + ane By = (ayy ~ Qa1)% + Gare ‘Scanned with CamScanner OPERATIONS RESEARCH : QUANTITATIVE TECHNIQUES FOR MANAGEnEt \d mark a scale on ea yo . we draw {wo parallel lines one unit apart an‘ ch of ‘To plot the expected Pees the tee Strategies aWallable (o player A. Then we draw lines to represent each then, These to repwesent B's st strategy (B,). we Join a, on scale 1 to daa, on scale 2. This line a ae rcent the expected payoft ofthe line By = (a4, ~ day) % + cq With x, a8 axis and Ey a8 y -axion Similarly repres : y off ines can be drawn. Pa et point an the ler doundary of these lines will glve maximum expected payoff among the mut expected pavotis on the lower boundary and the opttinal value of probataty py. any Pa. Now the two strategies of player B corresponding to these lines which passes through the maximin point can he determined. It helps tn redueing the size of the game lo (2 x 2) whieh can be easily solved by any of the et ' thod discussed carlier. mo (mx 2) games are also treated in the same way except that the upper boundary of the straight lines sponding to 8 expected pavolt will give the maxiinum expected payalf. to player Bard the wae poet on this boundary wil then give the minimum expected payoff (minimax value) and the optimal value of probability q, and q2. Example 11-11. Solve the following game graphically and find the value of the game : Player 8 5 BB aA [8 4 -2 A [2 -1 3 Solution. STEPS 1. This 2 x 3 game has no saddle point as maximum of row minimums is not equal to minimum of column maximum. Therefore graphic method s to be used to reduce this game to 2x2 game, 2.1 p, and p2 are the probabilities with which the player A uses his pure strategies, then Pi+P=1 or p,=1-pylp,, p22 0) Player A B's pure strategies E(u): A's expected payoff | peas _ E(u): A's expected payoff | Player A's expected payoff corresponding to player B, Ey, =8p,-20.—p) B's pure strategies is as shown in the adjoining B =4p-(-p) table: Py - 3 (1 - p) ‘These three expected payoff lines are plotted on the graph to solve the game. 3. Graph for player A. Draw two parallel lines one unit apart and mark a ‘scale on each. These two lines represent the two strategies available to player A. Player A determines the expected payoff for each alternative strategy available to him. If player B selects strategy B,, player A will gain 8 by selecting strategy A, and -2 by selecting strategy Ay. ‘The value 8 plotted along the vertical axis under strategy A, and the value -2 ts plotted along the vertical axis under strategy Ay. A straight line Joining the two points ts then drawn. ‘The line represents the maximum possible payoff to player A, Proceeding in the same manner, other two lines may be drawn, Assuming that player B will always select the alternative strategies yielding the worst result to player 4, the payolls {aaine) to A are represented by the lower boundary (shown by thick line in the figure) for any probabilistic value of A, and Ay between 9 and 1, According to the maximum eriterion, player A’ will always er select a combination of strategies A; and A, such. that “Ye re. of maximizes his minimum gains, In this case the optimal solution occurs at the intersection of the two pay lines. he 4. The point of optimal solution, (Le., highest or maximum point on the lower boundary) occurs at th intersection of two lines : E,=4p,-(1-p)) and By=-2p,+3(1-p) ‘Scanned with CamScanner THEORY OF Games (a3) THEORY OF Games te, AINE Glagram indicates that the player A's expected payoff depends on whieh strategy player B selects. At the point where the aco les Ey. Ald By Intersect, the pavolt ie the wane the player Ano watt. which counter strategy player B uses. This unique payoft Is obtained by setting Es equal tO Ey and solving for pr. Le., 4p, ~(1 =p) =-2p, +301 ~ p). Therefore py = 0-4 and 1 p, = 0-6. Then substituting the value of py in equation E; (or Ey), U= 4% 04-06 = 1-0 SIs Ure optimal value of the game when the optimal mixed strategy forthe player A ts (0-4, 6). Fxample 11-12. Obtain the optimal strategies for both players and value of the Game Jor two-person zero- sum game whose payoff matrix is as follows : B By BR Arr -6 7 Alo 4-5 A As] tO =2 43| 72 5 al 7-6 Solution. Steps 1. Since maximum value of maximin A's strategies B's Expected payoff of the rows called ‘maximin’ (= -2) is not equal to the re ere ~a) minimum value of the maximum of the columns called A 4q, +5020) | minimax (= 7), there is no saddle point. 2 Ea= 4g, +5 (1-q) | 2. There is no dominating row or column in the matrix As Es= -m -21-q) | and it is 5 x 2 matrix which will now be solved by graphic Ay Ey=-2q, +5 (1-q) | method. As | Es= 7m -6 (1-q) {__Es= 7m -6 (1-a) | 3. Let the strategies of player B be in the proportion of q: and 1 - q, for Strategies B, and Bz. B's expected payoffs Corresponding to A’s pure strategies are given in the adjoining table : These five pay-off lines can now be plotted in the usual manner as shown in adjoining Fig. 11-2: | ts obvious from the above diagram that the minimax fceurs at the point P (here take the lowest point on upper most boundary as it is 5 x 2 game), which is the point of Intersection of pay-off lines, Ey 9, 2q.+5(L-q) and Es=7q, -6(1-q) ig. 12 [ matrix of size (2 x 2) The solution to the original (5 x 2) game reduces to that of the game with payolf matrix of size (2 x 2) as shown below : / ayer By By Ay [-2 5 Payera G5 a2x2 game. “hich can be solved by the usual algebraic method of solution for a rf The optimal solution of the given game is: wz Optimal strategy for player Ais : (0. 0.0.3.5 23 (4. ) and value of game, v=39° nd optimal strategy for player Bis ¢ ( 3b+20 ‘Scanned with CamScanner OPERATIONS RESEARCH ; QUANTITATIVE TECHNIQUES FOR MANAGEMENT OPERATIONS RESEARCH i eww Example 11-19, Solve the game for the pay-off matrtx Player B BR By By A 19 6 7 5 Ad 7 3 4 6 PiauerA ay | 12 8 14 A Le 7 19-1 ‘Solution. Steps 1. Since maximin # minimax, there {s no saddle point. 3 dominates row 4, the game Is finall 2. Further, columns 1 and 3. are Again row " m ty column 2, the reduced matrix reduced to 3 x 2 matrix and can now be solved by dominated by coun reduced 1 3 : B B bh Om B By ny 6 5 AL 6 5 a @) 3 § a4 | 3 6 as| 8 4 ale 3 & aL 7-4 3. If ga. qq are the probabilities with which the player B uses his pure strategies, then qz + q = 1 oF qe= 1-42 (20, a2 0)- Bs expected payoffs corresponding to A's pure strategies are given below : A's pure strategies B's expected payoffs 1 6m +5(1-@)=m+5 2 3q2+6(1-a)=-3n +6 3 8m +4 (1-q)=4a+4 ‘These three payoff lines can be plotted as functions of q as follows : Payot Payott Draw two lines Bz and B, parallel to cach other one unit apart and mark a scale on each of them. These two lines 5 represent the two strategies available to player B. To represent A’s first strategy, join mark 6 on By with mark 5 on B, : to represent A's second strategy, Join mark 3 on B, with mark Gon 3 By ; and so on and bound the figure from above as shown in the adjoining Fig. 11-3: a a=0 ae! Fig. 113 : Graphic solution to the game. Since player B wishes to minimize his maximum expected losses, the two lines which intersect at the lowest point of the upper bound show the two courses of action player A should choose in his best strates: Le., Ay and Az. We can, thus, immediately reduce the 3 x2 game to the following 2 x 2 game : B BR Bs Ay 6 5 A hs [ 3 6 ‘The adjoining 2 x 2 game can be solved by algebraic method. ‘Scanned with CamScanner igoBY OF GAMES e—_ pence an optimum solution of the given game I Optimal strategy for player A( 3,1. 0,0) Optimal strategy for player B(0, 4,0, 3) L and Value of the game to Als 4 ‘Scanned with CamScanner 11-7. LIMITATIONS OF GAME THEORY Game theory, which was initially received in literature with great enthusiasm as holding promise, has been found to have a lot of limitations. The important limitations are mentioned below = 1. As the number of players increases in the game, the analysis of the gaming strategies becomes increasingly complex and difficult. In practice. there are many firms in an oligopoly situation and game theory cannot be very helpfull in such situations. 2. The assumptions of maximin and minimax show that the knowledge of the strategies. These do not seem practical. 3. Rather than each player in an oligopoly situation working under uncertain conditions, the players will allow each other to share the secrets of business in order to work out a collusion. Thus, the mixed strategy are also not very useful. Players are risk-averse and have complete 4. Game theory assumes that each firm has knowledge of the strategies of the other as against its own strategies and {s able to construct the pay off matrix for possible solutions. This is a highly unrealistic assumption and has its little practicability. An entrepreneur is not fully aware of the strategies available to him, much less those available to his rival. He can only have a guess of his and rival's strategies. 5. The theory of games assumes that both the duopolists are sensible men. Each rival moves on this Presumption that his opponent will always make a wise move and he adopts a counter move. This is an unrealistic assumption because entrepreneurs do not always act rationally. 6. It is easy to understand a two-person constant sum game. But as the analysis ts elaborated to three or four person games, it becomes complex and difficult. Most economic problems involve many players. For Instance, the member of sellers and buyers is quit large in monopolistic competition and the game theory does ‘Not provide any solution to it. 7 7. Even in it ication to duopoly, game theory with Its assumption of a constant sum game ts unrealistic. The ‘mininax principle which: provides ‘a solution to the constant sum game assumes that each Player makes the best of the worst possible solution. ‘Scanned with CamScanner

You might also like