OPERATIONS f, : 1 pe B, and ifall bots were of type B, the company coud i! Sa neat g coher egrsen for ey 800 belts per day (Both A and 1 Combined Det ie fancy buckle and only 400 buckles per day ae available. There are only 700 bulla anata Belt B. Determine the optimal produet mix. (Marat MBA, 2006; Delt M.Com. 2065 (oa ‘Solution. ‘ . / ‘Step 1. The appropriate mathematical formulation of the given \near Programming Problem ig. ‘ g Maximize z= 4x, 43x) subject tothe constraints + : 2uy + xy $1,000 4 (Time Constrain 0 SoA + ay S 800 Avaitabiiy of Leaieg 1 $400. and “xs 700 _ (Availabiity of Bucky x20 and x) 20, : : where x; = number of belts of type A, and x, = number of belts of type B: Step 2. Next we construct .the graph by considering the cartesian rectangular axis OX,X, in the plane. As each point has the coordinates of the type (x, %2)5 any point satisfying. the conditions %, 2 0 and x, > 0 lies in the first quadrant. : Now, the inequalities are graphed taking them as equations, eg. the first constraint ES 2x1 + 42 $ 1000 will be graphed as 2x, +, = 1000. The equation is re-written as 4 a 500 * To09 = = This equation indicates that when it is plotted on the graph, it cuts an x-intercept ‘of $0 and Zzinereept of 1000..These two points are then connected by a straight line which, is shown m Fig. 3.(@) as line AB. Any point (representing’a combination of x, and x,) that falls on tis lie cece tr ee erg 8 ceptable in so far as this constraint is concerned. ‘The region OAB forme by two axes and the Tine re Presenting the equation 2x, + x, = 1000 is the region containing acceptable values of x, and x, in respect of this constraint, Similarly, the constraint x; + x < 800 cain-be equation x, + x = 800. The region OCD, which any point would satis and x) < 700 are also lines 2) = 400 and x, X plotted. The. line CD-in Fig. 3.1(b) represents the , formed by the two axes and this line represents the area in fy this ‘constraint of leather: availability. Further, the constraints x, < 400 Plotted on the graph which tepresents the area between the two axes and the = 700 as shown in Fig. 3.1(b). 5 i x, 2x + x = 1000 = 800 600 400 200 Fig. 3.1(a) Now all the constraints feasible region or solution n Fig.3.10)° | casi 2 i I: these: constiaints. om in Fig. 3.1(6) by ya : : th RST. the feasible region, The coordi aes oeetivg functic y the shaded area OPO. have bee: olution space, ig ge fetPhed: The area bounded Step 3. The optimum yal 2 éalled ina n Occurs gt one of net) Poi v= G0) P= 0.8 ca tone Po are O° OF the extreme (comet) 200, 600),.5 = ts of (100, 700), and 7 = (0,700)errs cs r . UNEAR PROGRAMING PROBLEM GRAPHICAL SOLUTION AND EXTENSION 67 Step 4. We now compute the z-vattes Corresponding to the extreme points : Extreme point tym) dearer 0,0) (400, 0) (400, 200) (200, 600) (100, 700) 700). Step S. The optimum solution is that extreme point for which the bjective function has the wm solution occurs at the potnt 2, /.e., 4, = 200 and:x, = 600 with the objective function value of Rs, 2600, ‘ 7 Hence, to maximize profit, the company should Produce 200 belts of type A and 600 belts of ype B per day. ? , Alternative Method (Iso-profit approach) ‘The feasible region (solution space) obtained in step 2 is as shown in Fig. 3.1(c) by the shaded area OPORST. Ur Let the profit to the company (arbitrary) is Rs. 1200. The objective function then becomes : 4xy + 3x2 = 1200, We draw this equation as a straight line in the feasible region shown in Fig, 3:1(¢): This Ithe i known as iso-profit line. é 2x, + xy = 1000 Direction’ of 1000/2400 iran 2 TES ey “Wo X + % = 800 © 200° 400-600 800 1000 X Fig. 3.1(c) fit function (objective: value function) is a straight line on which It may be noted that the iso-prot “very point has the same total profit the NO: WE move the iso-profit line parallel to itself farther from the origin. We observe that one of temed PEt line touches only point R before leaving the feasible region. This iso-profit line is med as highest possible iso-profit line and point R gives the extreme Point of the solution space. Hence, the optimum feasible solution is : 1 = 200 and x2 = 600 with Maximum z = Rs, 2600, iio Letus assilme that you have inherited Rs, 1,00,000 from. your father-in-law that can be. Pontolig 2 Cambination of only two stock portfolios, with the. maximum investment allowed in either ‘Soot Set-at Rs. 75,000. The first portfolio'has an. av rage rate of Heturn.of 10%, whereas t Sion HS 20%, In terms of risk factors associated with slese Portfolios, the first leas.a risk rating of 'eieale from 0 16 10), and tt second\has 9. Since You wish to maximize,your return, you will”; Ee"OPERATIONS gue Peden, ww 12% oF a Wisk factor above 6: Hencé, fst in each portfolio? Problem and solve it by Graphic Method, Ic. Fla ayy ne appropriate mathematical formulation of the linear programming, problem Maximize z = 0.10, + 0,20xy subject to the constraints : ss ny +x $ 1000000, 1 $ 75,000, 42 S 75,000 40.20 2 0.12) + 4) OF 70.024); +-0.08% 2 0 +m) or -2r, + 3x2 $0 8 , ‘ belo' You tf * return hen foes ‘ e vate t not accept an avert Och should you inv it nym ‘mapostant question. HOW ™ i eeeormntare mis as @ Linea” Programming 0,10xy Ary + 9x2 S OC x) 20 and x2 20 «in portfolio Z,.and x = amount invested in portfolio 2. where x, = amount investe < 1,00,000 can be graphed by plotting the straight fie Step 2. The first constraint x, + 2 a %2___ 1, This cuts a x-intercept and x,-intercept of 1,00,000 each. The area below 1 7,00,000 * 1,00,000 thie line represents the feasible area in respect of this constraint. Similarly, the other constrains are depicted by. plotting the straight lines corresponding to the equations x = 75,000, - = 75000 2x, + 3x = 0, and -0,02r, + 0.08x, = 0. Here, the area below the first three lines and beyond the - fourth line gives the feasible region in respect of these: four constraints. ae Thus, the feasible region in respect of the given problem is as shown in Fig. 3.2. x= 75,000 Tesponi Buree pois ‘ponding to the ‘extreme points are ; io Oy x, A nt 2 = 0.10x, + 0.20, - (75,000, 18,750) 2 Eu, (75,000, 25,000) 11,250 Hence, the optimum solution jg . (60,000, 40,000) 12,500 41 = 60.000, 100, and xy, x9 2 0. where x; = number of units of product A, and x, = number of units of product-B. Step 2. Consider now a set of cartesian rectangular axis OX,X, in the plane. As each point has the coordinates of the type (x), x2), any point satisfying the conditions x, > 0 and x, > 0 lies in the first quadrant only. The constraints of the given problem are plotied as described earlier by treating them as equations : : 36x) + Gxy = 108, 3x; + 12x) = 36 and 20x + 10x; = 100 set se Gea © Fy Buty ye Brandl tag res ygutg rand The area beyond these lines represents the feasible region in respect of these constraints; any Point on the straight lines or in the region above these lines would satisfy the constraints. The feasible gion of the problem is as shown in Fig. 3.3. oe 4 6 8 0 12 W 16 18 x, at Fig. 3.3 Si , *P 3. The coordinates of the extreme points of the-feasible region are : * A = (0, 18), B = (2, 6), .C = (4, 2) and D = (12, 0)-es NEAR PROGRAMMING PROBLEM—GRAPHIGAL, SOLUTION AND EXTENSION ms at least 3 programmes ought to be on TV and the number rahe tse be ; SE of radio prog ited 10.5. Surveys indicate that a single TV tomers in target inte ‘and 130,000 te nan si Programme reaches 7,50,000 custom lence B. O; ‘hes 40,000 in target aafience A and 2,60,000 in target audience p, | ON AHO Programme reaches [Dethi M.B.A. (Oct.) 2010] Solution. Let x, and x represent the numbet of Programmes to be released on TV and radio, respectively. The mathematical formulation of the LPP is: Maximize z= TV programme viewers + Radio Programme viewers 50,000 + 1,50,000). +. (40,000 + 260,000) x, = 900,000.21 + 3,00,000.x, subject to the constraints : 50,000, + 20,000. < 2,00,000 * 23 and x <5 % 20 and x20 For solving this L.P.P. graphically, we fi suaight lines of these equations. Now, satisfying all the inequalities, vic., the regi X 10 (Budget constraint) (Programme constraints) (Non-negative restriction) ist consider the inequalities as equations and draw the consider the inequalities and shade the feasible region ion ABC consisting of the extreme points A, B and C. Fig. 3.7 The coordinates of the extreme points of feasible region are : A = (3, 0), B = (4, 0) and C = (3, 5/2). The z-value Corresponding to extreme points are ;- Extreme point ry x2) 2 = 9,00,000x, + 3,00,000.r, ‘A @, 0) 27,00,000 a (4, 0) 36,00,000 maximum 5; £ @, 5/2) 34,50,000 “ $ rages the “aximum value of z occurs at the extreme point B(4, 0); the agency must release iencoe mes on TV and no Programme on radio to achieve the maximum reach of 36,00,000 my | PROBLEMS Sects oe Manufactures two products A and B on which the %e on 4g 2° Product is processed.on two machines My and Mp, {Wo minutes on Mz while B requires one minute or profits eamed per unit are Rs. 3/aiid’Rs. 4° ‘Product A requires one minute of processing nM, and one minute on Ma. Machine Mis“OPERATIONS n oe on SEARCH The z-value corresponding (° exteine points ars ) eae $2 (xy. X) see ai Fe saree point 0.0 2 (2. 0) i 2 minim : 2) : 9 = 4 @.4) 2 > (0, 2) a oint A (2, 0). Hence, the optimum solution of the ‘The minimum vale of = occurs at the extrerne P Lpp is: = agejy = 0 and minimum. 2 = 2. 4306, Use the graphical method 10 solve the following LPP : é Maximize = 2) + 3824 subject 10 the constraints; ay +42 £30, 4-20 22H 0s x, $20 and osx 12 the given inequalities, We first treat them nym ORME 0 and xp = 12 and plot each of these equ the inequality condition of each constraint to plot the corfesponding fe region ABCDE satisfying all the inequalities (feasible region) is shown in XA aoe Solution. To graph Fig. 3.6 The coordi in inates of the extreme points of the feasible region are : as equations x + % = 30, tations as straight lines. We use asible region. The common shaded Fig: 3.6. A= G, 3) B= (20, ‘The 2 : ), 3), C = (20, 10), D= mis corresponding to extreme aie ay fa, Uy ad = 02 treme point : a Gry 9) =u, +3 4 en z Ht 3 c 20, 3) 75 : D 20, 10) i ; oo (18, 12) 2 { ‘The maximumValue of z occurs at th (12, 12) iB res extreme poi x= is 1= 18, y= point D (18,- 12), Hence, the optimum solution 5° gor, The adverttin asency wish 5 10 12 and maximum z = 72. : rhan Rs. 40,00 incomes greater U 000 (tar reach two ty ‘ incom ars andre B, fret audience A) an [pPes of audiences, customers with annual ce A.000 one a tPeising budget ts. 2 ith annual incomes of ess ay 5. 2,00,000. One 4 ,000, One programme advertising 8 nine oy °f radio advertising costs Rs. 20,000. For contract7 NEAR PROGRAMMING PROBLEM—GRAPHICAL SOLUTION AND EXTENSION . x Step 3. The coordinates of the extreme p Step 4, The value correspondin; Extreme point sare: A= (8, 5/3), B = (8, 10), and C = (3, 10). ig to the extreme points are : Cy 22), 2 = 400r, + 360x, 3 A Bs (8, 5/3) 3,800 ———minimum B (8, 10) 6,800 Cc (3, 10) 4,800 Henee,-the optimum solution is %1 = 8 and x) = 1.7 6¢ 2 (approximately), minimum z= 3,800 4 jus, 8 grade 7 inspectors and 2 grade 2 ingpectors should be assigned to have Ré. 3,800 as the ‘oul minimum inspection cost. 305. Use the graphical method to solve the following LPP : Minimize 2 =x; +.2x7; subject to the constraints : ay + 3xy S10, x, + <6, 4-2 S 2, and x20, 20. re-written in the intercept form. Thus, we write oo iy mo % : Tho * joa S| ete sh and aris! s-straight lines. Now, considering common region OABCD gives the Solution. The constraints are First, we treat these inequalities as equations and graph them a ts inequalities, shade the feasible region for each constraint, ‘The flition space of the LPP. It may be noted that we have conside first quadrant of th red the feasible’ region only in the fe cartesian axis, as x, > O and x, > 0. 1 2 2 0. X 210 : T72. Wgr ag OY The Fig. 3.5 ant: 8 “fordinates of the extreme Points are : " erase O. ©, 0), A= 20), B= (4,2, ¢ Nes = 2/4) and D———_ * OPERA TIONG Fete Step 4. The vata of objective function at each of the extreme points can be Cvalated ag fa Fey E nae (xy 9) 2 = 20r + dog, 000" Molloyy int ; iy See oy (0, 18) 70 § B (2, 6) 280 c; (4, 2) 160 minim D . (12, 0) * 2140 Hence, the optimum solution is to purchase 4 units of Product A and 2 units of prods p waig to maintain a minimum cost of Rs. 160," can check at the rate of 15 pieces per hour with an accuracy of 95%. The wage rate inspector is Rs. 40 per hour while that of grade.2'is Rs. 30 per hour. Each time an error is caused by the ingpector the cost 10 the company is Rs: 20. The company has ei inspectors. The company inspection cost. Formulate Solution. Step I. The data of the given problem can be summarised as follows ; Grade 1 inspector Grade 2 inspector Number of inspectors 8 10 Rate of checking per hour 25 pieces 15 pieces Inaccuraty in checking 1 = 098 = 0.02 : 1-095 = 0.05: - Cost of inaccuracy in checking Rs, 20 Rs. 20 Wage rate per hour Rs. 40 . Rs. 30 Let x, and x, designate the number of Grade 1 and Grade 2 inspectors, respectively. Hourly costs of each Grade J and Grade 2 inspectors are given by : Grade J inspector : Rs..(40 + 20 x 0.02 x 25) = Rs. 50. ~ ye OTe 2 inspector = Rs. (30 + 20 x 0.05 x 15) = Rs. 45, Using the above information; nimi ‘he appropriate linear programming problem is : Minimize z= 8 x 50x, +8 x aSiy = 400s; + 360% subject to the constraints : 8% 251 + 8 x 15% 2 1,800. oF Sey + ax 2 45 (Number of pieces) shal a S 10 (Number of inspectors) (Non-negative restriction) i in the plane. As each point has the coordinates of the type (x), x4), Miesian rectangular axis OX,X, any point satisfying the conditions x, > 0 and x, > 0 lies in the first quadrant only. The first construint 5x + 32, > 45 can be an 4 ! . This gives an x-intercept of 9 and an Azimercept of 15. The area away from the origin, 4¢-+ above this line represents the feasible are igh lines comesngt tis CORStTaint, Simiarly, the other two constraints 8 depicted by plotting the straight lines Corresponding to the equations es = 8 and x, = 10, Here, the ae ae eed WRAL) BM feats arta bo aoe whieh satisfies all the-three constraints isthe solution gon are x hed by plotting the straight line a +GAMES AND.STRATEGIES nee ‘The expected value of the game (corresponding to the above strategies) is given by B39) (69) 00 Jul YS Pel ae ay Toss Thus, the optimum mixed strategies for matching player and non-matching player are given by H T u T mye i oi i wis | Sammi Tas. WAS Clearly, we would like to be the non-matching player. PROBLEMS T716. Solve the following game and determine the value of the game : @ B (b) z; Siecle Mis 0 ey } . [ a3 (Madras BE (Mec) 1999) [Wllhatod 0.8.4. 1985) 1717. Tn a game of matching coins with two players, suppose A wins one unit of value, when there are two heads, wins nothing when there are two tails and loses 4 unit of value when there are'one head and one tail Determine the payoff matrix, the best strategies for each player and the value of the game to A. TAmravathi BLE. (Rut) 1994) ATE. Two players A and B match coins. If the coins match, then A wins two units of value. Ifthe coins do ganesh then B wins 2 units of value. Determine the optimum strategies forthe players and the vale op the game. ‘(Madras M.B.A. (Nov.) 2006; Delhi M.Com. 2008) A719. A and B each take out one or two matches and guess how many matches opponent has taken, If one op, the players guesses correctly then the loser has to pay him a5 many rupees as the sum of the number held by both Plvers. Otherwise, the payout is zero, Write down the payoff matrix and obtain the optimal stategies eg boty players. Wodhpur M.Sc. (Bath.) 1994) 17:6. GRAPHIC SOLUTION OF 2 xn AND m x2 GAMES he procedure deséribed int the last section will ‘generally be applicable for any game with 2x2 Payoff matrix unless it possesses a saddle point. Moreover, the procedure’ ean be extended. to any fata PayotT matrix of any order. But it will not work for the game whose payoff matric happens to be & rectangular one, say mn. In such cases a very simple graphical method is avallabre ne either m or m is two. The graphic short-cut enables us to reduce the original 2% or moe game to a much simpler 2x2 game. Consider the following 2x n game : Player B PiksuliBige an . AL ay a Player A i “i A ( ay It is assumed that the game does not have a saddle point, Let the optimum mixed strategy for A be given by 5, = [ iltstteps ] where py +p,=1 ‘The average (expected) Payoff for A when he plays S, against B's pure moves By, Bay ..., By is given by : pan B's pure move A's expected payoff E(p) > 8, : . Z zi F,Oy = ayy, + 21 Pr = ay, py + a4 (1'=p)) < Ee 4121 + a9 p, = aap, + a2 (1-1) . ¥ Bn é = MnPl + doy P2 = Min Py + Arq (1 py). i esteERATIONS RESEARCH 4sa a Hence, we have =a jan ayy dn = O13 Oot n : a i ot a Paton een ar ey ada we ay aaa at ithout saddle wit “Note : The above formulae for py pas gy a2 and v are vad only for 2x2 games points, ‘ SAMPLE PROBLEMS ; ind the 1714, For the game with the following payoff matrs, determine the optimum sraegies valu of fhe game : 2 Sel o [ ge Nig ] i — [ICSI (June) 1996; Madurai M.Com. (Nov.) * ii: ies of P, and Solution, Clearly, the given matrix is without a saddle point. So, the mixed strategies of Py Py are: ees 1 2 [ } = 2 op = 3 PytPe Ss [ mn Pe ] web % and qi+92= 1 If E(p, g) denotes the expected payoff function, then E@, @ = Spa + 3(1-pdg Hpi (l-av + 4p) 9) = Spin - 31 — 91 + 42 Sr 1/5) (gi - 3/5) + 17/5. If P, chooses p; = 1/5, he ensures that his expectation iat least 17/5. He cannot be sure of more than 17/5, because by choosing q, = 3/5, P, can keep E(p;, q;) down to 17/5. So: P, might as well settle for 17/5 and P, reconcile to 17/5. Hence, the optimum strategies for P, and P, are t cj 1 2 Sp = = "1 [ 5 4/5 } oe [ 3/5 OS ] and the value of the game is v = 17/5. } | 1715. Consider a “modified” form of “matching biased coins” | Player is paid Rs. 8.00 if the two coins turn both heads and Re. 1.00 a ee rar Rey game problem. The matching if the coins turn both tails. The, h. Given the choice of being the» what would be your strategy? (Delhi M.B.A, 1999, 2007] non-matching player is paid Rs. 3.00 when the two coins do not mate matching or non-matching player, which one would you choose and Solution. The payoff matrix for the matching player is given by Be Non mashing Player i rT | wacimtine [8 8 / T 3 1 Cleary, the payoff matrix does not possess any saddl le strategies. The opti i . i strategies. The optimum mixed strategy for ‘matching player is deter ate players will use mixed 6 = 3 ys : Pi a1- Goa 4 Pail and for the non-matching player, by 15 mi 1- (2 ty CT ore) Tairerye te ae i, 15GAMES AND STRATEGIES 451 the optinnum mixed strategies Selah 8 | sand gyn [BB | mr Ns are determined by a= a, Nd = a m= 812 2 My = ay and q\+4,=1. The value v of the game to'A’is given by ua = dy dia ayy + ayy ~ (ayn + ay)" where p, +P: A Proof. Let a mixed strategy for player A be given by 5, [ : i i es |: where p, +p)=1. Thus, i te if player B moves By the net expected gain of A will be Ev) = ay Pr + a2 Py and if B moves By, the net expected gain of A will be 2x) = api + apr By 3 Similarly, if B plays his mixed strategy s, -[ tee 1 }: where gq, +q)=1, .then B’s net expected loss will be FL@) = aun + a9 if A plays A,, and 2.) = ain + an ay if A plays Ay. ‘The expected gain of player A, when B mixes his moves with probabil given by i€S-qy and q, is, therefore, Ea) = 41 lap + an po) + 42 [1271 + x22). Player A would always try to mix his moves with such probabilities so as to maximize his expected gain. Now, Ep. a) = alan py + ay (Lp) + (9) fazpy + a9 (1-py) = Lan + aan ~ (ay + an) pygr + (aye ~ a2) Py + (@21 ~ 32) 4) 4 G9 2 _ m= ay f= an) | a1 a9 ~ ayn an1 alos x las ¥ js 7 where 2 = ay +a - (ayy + a1), We see that if A chooses p, = %2—% * he: ensures an expected gain of at least (241 259 ~ a2 @)/2. Similarly, if B chooses g, = most (ay, dz2 ~ A @y,)/2. These choices of py and q, cy + then B will limit his expected: loss to at will thus be optimal to the two players. Thus, we get o2 fxn = May Y aya or? = and [eps ae an + ay ~ (ay + ay) Pe a + an = (ain + a a= oe ian ay = 31 ain + dan ~ (ayy + ay) and gn Day 5 = M1122 = 43 a9) ' M1 + 32 - (ayy + ay))* a hae a Pay andOPERATIONS RESEARCH 17:5. GAMES WITHOUT SADDLE POINTS—MIXED STRATEGIES As determining the minimum of column maxima and the maxima of row minima are wo different operations, there is no reason to expect that they. should always lead to unique payoff position—the saddle point. In all such cases to solve games, both the players must determine an optimal mixture of strategies to find a saddle (equilibrium) point. The optimal strategy mixture ‘or each player may be determined by assigning to each strategy its probability of being chosen. ‘The strategies 80 determined are called mixed strategies because they are probabilistic combination of available choices of strategy. ‘The value of game obtained by the use of mixed strategies Teprescers which least player A can expect to win and the least which player B can lose. The expected payoff to a player in a game with arbitrary payoff matrix (q,) of order m x is defined as : Ew. 9 =,2,, 2,709 where p and q denote the mixed strategies for players A and 3 respectively. xn game (aj) without any saddle point, ie. Maximin-Minimax Criterion. Consider an m be the probabilities with which player A will play his moves 450 = pTAq strategies are mixed. Let Py, Pa» +++» Pm ‘Ay Apy «ns Ay Fespectively: and let qys day --» dn be the probabilities with which player B will play hig moves By. By, ..-. By respectively. Obviously, py = 0 (i = 1,2, seg), G20 G = boat and py + Po to 4Pm = 1s NtR t+ t%0 = 1 ‘The expected payoff function for player A, therefore, will be given by a E@. d= 2, 2 Pd 7 i Making use of maximin-minimax criterion, we have For Player A. y= max. min, E(p, a) = max.| min, | zane ACO, es [ bi (Ze) Sm] Here, min. {3 > ay} denotes the expected gain to player A, when player B uses his jth pure strategy. mat |' main’ | pags ° [» {Ee 2 Ped For player B. min, | max. | z [ol Sata fibre dad Here max. { az au] denotes the expected loss to player B when player A uses his ith strategy. The relationship » S ¥ holds good in general and when p, and q, correspond to the optimal 1) nd to the optimal strategies the relation holds in ‘equality’ sense and sail 2 Pia apiionumn prpetied vahie,atdaeibane, the expected value for both the players becomes Definition, A pair of strategies (p, q) for which Tpysofem 17-2. For any 2x2 two. payoff matrix for player A so hel vey & called a saddle point of E (p. " game without any saddle point having the yh 4 [au A: Loan aul i we‘ OPERATIONS RESEARCH aa . mrookey Woy haN may, lay) 2 ay for all f= 1 2 sen Sa iin, tay) Say for all = 1,2, am ‘Let the above maxiniuin be attained at # = i uid the thinimum be attained at =f, te., myays, fay) = ey, and min Cay) = aye Veisin ‘Then, we must have 3 " My 2 aye for all = 1,2, .6m3 9 = 42.08 From this, we get : j min, {ay} 2 ay 2 max. {ay} for all f= 1,2, 0.45 f = 1,2, cm VSjsn isis ‘ i > in. a4) * VES EPS y COUR mae, Cie, aud as or Fey Remarks 1. A game is said to be fair, if ¥ to be strictly determinable, Rule Yor determitting a Saddle Point - ‘how/Summarize the procedure of locating the saddle point of a payoff matrix as follows : P 1/Select the minimum element of each row of the ‘payoff’ matrix and’ mark them [*]. Step/2. Select the greatest element of each column of the payoff matrix and mark them [+]. 3. If there appears an element in the payoff matrix marked [*] and [#] both, the position of that clement is a saddle point of the payoff matrix. SAMPLE PROBLEMS 1701. Determine which of the following two-person zero-sum games are\ strictly determinable and fair. Give optimum strategies for each player in the case of strictly determinable games : Player B B 3 0 (@) Player A [ ] os perianal iy Bo. afi dos 2d ‘ (Madurai M.Com. 1997] Solution. (a) The payoff matrix for player A is oe Plow Row minima By By A’ st oF 3 Ar oF ot > Column maxima 5 : The payoffs marked with [*) represent th : {f] represent the maximum payoff in eash cars enum Payot in each row and those marked with umn of tI H a represents y (maximin value) and th he payoff matrix. The largest component of row (minimax value). ‘Thus obviously, we have ¥=0 and Since y # ¥, the game is not strictly determinable,eg AND STRATEGIES on phyer A, minimum value in each row represents the least gain (payoff) to him if he chooses, é particular strategy. ‘These nl in the matrix by row minima, He will then select the strategy tpt maximizes his Erne athe This choice of player A is called the maximin principle, and the ‘responding gain is called the maximin value of the game, “ For player B, on the other hand, likes to minimize his losses. ‘The maximum value in each column sents the maximum loss to him if he chooses his particular strategy. These are written in the patix by column maxima, He will then select the strategy that minimizes his maximum losses. This ahoice of player B is called the minimax principle, and the corresponding loss is the minimax value of . the game. i If the maximin value equals the minimax value, then the game is said to have, a saddle (equilibrium) point and the corresponding strategies ate called optimum strategies, The amount of peyott at an equilibrium point is known as the value of the game, ! To illustrate the maximin-minimax principle, let us consider a two-person zero-sum game with the following 3X2 payoff matrix for player A : Player B BB A (9 2 Player A 42 | 8 ot a As (6 4 Let the pure strategies of the two. players be designated by Sa = {Ay Aa, As} and Sp = {By By}. Suppose that player A starts the game knowing fully well that whatever strategy he adopts, B will select that particular counter strategy which will minimize the payoff to A. Thus, if A selects the strategy Aj, then B will reply by selecting B,, as this corresponds to the minimum payoff to A in the first row corresponding to Aj. Similarly, if A chooses the strategy Ay, he may gain 8 or 6 depending upon the strategy chosen by B. However, A can guarantee a gain of at least min. {8, 6} = 6 regardless of the strategy chosen by B. In other words, whatever strategy A may adopt he can guarantee only the ninimum of the corresponding row payoffs. Naturally, A would like to maximise his minimum assured gain. In this example the selection of strategy A, gives the maximum of the minimum gains to A. We shall call this gain as the maximin value of the game and the corresponding strategy as the maximin strategy. The maximin value is indicated in bold type with a star. On the other hand, player B wishes to minimize his losses. If he plays strategy B,, his loss is at the most max.{9, 8, 6} = 9 regardless of what strategy A has selected. He can lose no more than max. {2, 6, 4} = 6 if he plays By. This minimum of the maximum losses will be called the minimax value of the game and the corresponding strategy the minimax strategy. The minimax value is indicated in bold type marked with [*]. We observe that in the present example the maximum of row ‘minima is equal to the minimum of the column maxima. In symbols, max (q) = 6 = min {6} - aan in (a3) in. an Ca | where i= 1,2,3 and j= 1,2. | TheoremA7-1. Let (ay) be the mxn payoff matrix for a two-person zero-sum game: If v. denotes | the maxivin value and ¥ the minimax value of the game, then ¥ > v. That is, eg | omit ig ol 2 BR, Ki, ful oe] |