7 (=) 4.1 INTRODUCTION In the previous two chapters, we have discussed the nature, formulation and solution to linear programmin problems, using graphic approach and by application of simplex method. Besides yielding solution to, problem, the simplex method provides a host of additional information which is useful from the managemey viewpoint. The present chapter gives an account of this and is divided broadly into two distinct but rlaey parts—duality and sensitivity analysis, First we discuss about the duality which describes that linear programming problems exist in “pairs” so thy corresponding to every given linear programming problem, there is another LPP. The method of deriving the dual to a problem is followed by a discussion of the economic interpretation of the dual problem. Sensitviy analysis deals with the robustness of a solution to an LPP to changes in the inputs. It thus discusses hoy changes in the profit rates of the products, resource availability and resource requirements for the produas would affect the solution to a linear programming problem. 4.2 DUALITY IN LINEAR PROGRAMMING TT ‘As mentioned earlier, for every linear programming problem, there is another linear programming problem which is related to it and, therefore, may be derived from it. We now consider the method of deriving the dua toa given primal problem. Before writing the dual to an LPP, it is necessary to express it in standard formif it is not so. By standard form it is meant that (i) all the variables in the problem should be non-negative; and (ii) all the constraints should be of ‘<’ type if the problem is of maximisation type, and of ‘2’ type ifit ofite minimisation nature. The necessary changes may be made as follows: a minimisation ni When all variables in the LPP are not non-negative If a variable in an LPP is unrestricted in sign so that it is a ‘free’ variable, then it may be replaced by the difference of two non-negative variables, each of which is non-negative. While each of the variables introduced is non-negative, their difference might be positive, negative or zero. When all constraints are not in the ‘right’ direction If a constraint involves an inequality in a direction opposite to the one desired, then it is multiplied by =I throughout and the direction of the inequality is reversed. For example, if a constraint is given ® ‘Tey-5x, 24, and the desired inequality direction is <, then we replace this constraint as -7x, + 5x,$-4 is easy to visuaiize as to why the direction of inequality reverses on multiplication by —1 by considering simple example: as 10 is > 5 but- 10 is <—5. On the other hand, ifa constraint involves an equation, then, to obtain inequalities, itis replaced by 4 inequalities in opposite directions. To understand, if A = 20, then neither can we express itas A $2010" 220, since both are incorrect individually. But, if we consider A < 20 and A > 20 together to satisfy, they both be satisfied for only one value of 4, equal to 20. Accordingly, if the linear programming problem constraint, say 7x, + Sx, = 42, we would replace it by a pair of constraints as follows: Tay + Sxy $42, and Ty + Sx, 242 One of these would be multiplied by —I to bring the inequality in the desired direction. pair! f a aoe A1ap arity ana sensitivity Analy’ 42 writing the Dual ce the problem is known to be, or transformed into, standard form, we ean write its dual problem is, Maximise Ny ppose the primal Subject to arsb x20 c= row matrix containing the coefficients in the objective function, x= column matrix containing the decision variables a= matrix containing the coefficients in the constraints b= column matrix containing the RHS values of the constraints where, The dual corresponding to this problem is defined as Minimise G=by Subject to ayzc y20 where, 6’ = transpose of the b matrix of the primal problem, a’ = transpose of the coefficients matrix of the primal problem, c’= transpose of the matrix of the objective function coefficients of the primal problem, y= column matrix of the dual variables. Tosee as to how the dual of a problem can be formulated, consider the following examples. ExueGke) For the LPP given in Example 3.1 reproduced below, write the dual. Z= 40x, + 38x, Maximise Subject to 2x, +3x,<60 = 4/ AX, + 3x2 $96 = X4, X220 Iaccordance with above, its dual shall be Minimise G=60y, + 96y2 Subject to 2y, + 4y, 240 3y, + 3y, 235 Jvy220 Observing alittle closely, we find that(a) the primal problem is of the maximisation type while the dual is of the minimisation type- : (b) the constraint values 60 and 96 of the primal have become the coefficients of the dual variables y ‘yy in the objective function of the dual in that order; while the coefficients of inciyeninbien ioe “objective function of the primal have become the constraint values in the dual. * (c) the first column of the coefficients in the constraints in the primal has become the first row in th constraints in the dual, and the second column has similarly become the second row. ' (@) the direction of inequalities in the dual is the reverse of that in the primal. Thus, while the inequalig in the primal are of the type S, they are of 2 type in the dual. 7 We can represent them in matrix form as given in Chart 4.1. PRIMAL DUAL ] c x oo y Maximise Z= (40 35] [2] Minimise G = [60 96] (7 % Ye Subject to Chart 4.1 The Primal and Dual Relationship 4.2.2. Dual of LPP with Mixed Constraints Sometimes a given LPP has mixed restrictions so that the inequalities given are not all in the right direct In such a case, we should convert the inequalities in the wrong direction into those in the right direct Similarly, if an equation is given in respect of a certain constraint, it should also be converted into inequality To understand fully, consider the following examples, Write the dual of the following LPP: tion tion Minimise Z= 10x, + 20x, Subject to Bx, + 2x, 218 Xy+3x,2 8 2x,-x%yS 6 Xz 0 it Here, the first two inequalities are in the right direction (being > type with a minimisation tyP° of st function) while the third one is not. Multiplying both sides by — 1, this can be written as 2x, +22" we can write the primal and dual as follows: a“a Primal Dual Minimise, Z= 10x) + 20x, laximise G=I8y, + Subject to prea en 4 ene Subject to 3y, +2 ~ 2y3 S10 Xy+3x,2 8 2y, + 3y2 + 5 $20 2x, +4) 2-6 xpy2 0° Yn W320 Obtain the dual of the LPP given here: Maximise Z= BX, + 10x, + 5X Subject to x =%yS4 2x, +4x S12 By + xy t%yz 2 3x; + 2x) —X = '8. X4, Xo X20 ‘We shall first consider the constraints. Constraints | and 2: Since they are both of the type S, we do not need to modify them. ides by —Ito Constraint 3 This is of type >. Therefore, we can convert it into S type by multiplying both si become x —)-¥3S-2. Itic in the form of an equation. £< type and the other of 2 type. 3x, + 2x,-¥5 $8 3x, + 2x)—¥528 Constraint 4: “Anequation, mathematically, can be represented by a pair of The given constraint can be expressed as inequalities: one o! ‘The second of these can again be converted into type $ by multiplying by —1 on both sides. Thus, it can be written as —3x, — 2x, +3 $-8- Now we can write the primal and the dual Primal Z= Bx, + 10x, + 5X5 as follows: ‘Maximise Subject to x | OBS 4 — ax —m 32, 3x + 2-43 88> 3x, - 2m +H SS, pints 0 ‘ J Dual Minimise G = 4y, + 12y2— 293 + BYs— Bs Subject to NM + 2yy—ys + 3y4— 3528 Ay, Ys + 2g 295210 =yynYs— Yat V5? S Vpn VV Is 0 oeee One point nests mention er, We know hat correspon 9 Hae, conan primal ble re would be m-variable, n-constraint dual problem. For this example involving three variables and f if cer iis the dual should have four variables and three constraints. But we observe hal the dsl thel we ponetbrained contains five variables. The seeming inconsistency can be resolved by expressing (y ae hove able unrestricted in sign, Thus, although, yy and y are both non-negative, thei difference could be to zero, The duai can be rewritten as follow: greater than, less than, or equal Minimise G=4y, + 12y2— 295 + 8%6 Subject to Dit 22 Yat 362 8 Ay, —ys + 2y6210 “ = y3-¥o2 5 VeYo Is NG unrestricted in sign ‘Thus, whenever a constraint in the primal involves an equality sign, i's corresponding dual variable shall be unrestricted in sign. Similarly, an unrestricted variable in the primal would imply that the corresponding constraint shall bear the = sign. FREEAEE Obtain the dual of the following LPP: Maximise Z= 3x, + 5x + 7X3 j Xq + Xp + 3% 510 AX, — Xp + 2x2 15 A 2X4, X>2 0, X3 unrestricted in sign the type <. This results in —4x +x — 2x3 $-15. xc,and xg, This yields Subject to First of all, we should convert the second restriction into by the difference of two non-negative variables, say, Next, we replace the variablex3 ‘to which dual can be written, as shown against it. primal problem corresponding Primal Dual Maximise Z = 3x, + 5xq + 7x4— 75 Minimise G=10y,- 152 Subject to Subject to x, +x, + 3xy-3x5S 10 yy 49223 4x, +.xp —2ty + 2x3S-15 ytd XpXpkyxs2 0 3y,-2227 -3y, +2927 : yw?! ine 8 The fourth constraint ofthe dual can be expressed as 3y; ~2y, <7. Now, combining the third am constraints, we get 3y; — 2y2 = 7. The dual can be expressed as follows: Minimise G=10y,-15y, a —EEE itSubject to The symmetrical relationship between the primal and dual problems, assuming the primal to be a Mi dyy 23 My S 3 -2y,=7 Yy220 maximisation” problem is depicted in Chart 4.2 PRIMAL Maximisation No. of variables No. of constraints «type constraint ype constraint Unrestricted variable Objective function coefficient for th variable RHS constant for ith constraint Coefficient (9, for Ah variable in ith constraint — DUAL Minimisation ’, No. of constraints ( Non-negative variable No. of variables! Unrestricted variable = type constraint RHS constant for the th constraint Objective function coefficient for th variable Coefficient (a,) for th variable in jth constraint Chart 4.2 Symmetrical Relationship between Primal and Dual Problems 4.2.3. Comparing the Optimal Solutions of Primal and Dual Since the dual of a given primal problem is derived from and related to it, it is natural to expect that the (optimal) solutions to the two problems shall be related to each other in the same way. To understand this, let us consider the following primal and dual problems again and compare their optimal solutions. Maximise Subject to Primal Z= 40x, +3 2x, + 3x2 < 4x, +3x) xy 2 Sp 60 96 0 Minimise Subject to Dual G= 60y, + 96y, 2y, + 4y, 240 3y, + 3,235 Yry22 0 The simplex tableau containing optimal solution to the primal problem is reproduced (from Table 3.4) in Table 4.1.re TABLE 4.1. Simplex Tableau: Optimal Solution Basis a % Ss Ss ul 38 ° i 23 -u3 . 40 1 ° 12 2 e 5 40 35 0 ° Solution 18 8 ° ° 4 0 o 1083 = 2513 Now, letus consider the solution to the dual problem which is augmented, by introducing surplus and artificial variables, as follows: Minimise G= 60y, + 96y, + OS, + OS, + MA, + MAy Subject to 2yy + 4yy—S, +A, = 40 3, + 3y)—S, + A, = 35 Ya Yr Sy Sy Ay, AQ= 0 The solution to it is contained in Tables 4.2 through 4.4. Before comparing the solutions, it may be noted that there is a corresp. and the dual problems. The structural variable x, in the primal, corresponds to the surplus variable S; in the dual, while the variable x, corresponds to S,, the other surplus variable in the dual. In a similar way, te Structural variables y, and y, inthe dual correspond to the slack variables S, and S, respectively ofthe pial, ondence between variables of the primal TABLE 4.2 Simplex Tableau 1: Non-optimal Solution Basis wa Ya Ss S A Ay Bi blag 4M 22 oA 0 1 0 40 | We 4 M 3 3 0 -1 0 1 35. | 3583 5 60 96 0 0 M M Solution p50 0 0 0 40 35 Ans 60-SM.° 96-7 My 0 0TABLE 4.3. Simplex Tableau 2: Non-optimal Solution Basis YL da 5, & AA, by Way pile 96 12 1 -1/4 0 “OO 10 | 20 uM 3a 0 3i4 -! =314 1 s | 103 & A 60 96 0 0 MM Solution 0 10 ° 0 0 5 4 123M, Oioa yeni MeiirodM ag ° t 4 TABLE 4.4 Simplex Tableau 3: Optimal Solution Basis i yn Sy Sy Ay Ay by 2 96 0 1 =1/2 1B nm 158 | 258 vn 60 1 0 12 23 -12 23 | 103 5 60 96 0 0 M M Solution 193 + 25/3 0 0 o 0 4. 0 0 Is 8 M-18 M-8 A comparison of the optimal solutions to the primal and the dual, and some observations follow. (a) The objective function values of both the problems are the same. Thus, with x, = 18 andx,=8, Zequals 40 x 18 + 35 x 8 = 1,000. Similarly, with y, = 10/3 and y, = 25/3, the value of G would be 60 x 10/3 ’ +96 x 25/3 = 1,000. (6) The numerical value of each of the variables in the optimal solution to the primal is equal to the value of its corresponding variable in the dual contained in the A, row. Thus, in the primal problem, x, = 18 and , whereas in the dual S; = 18 and S, = 8 (in the A, row). Similarly, the numerical value of each of the variables in the optimal solution to the dual is equal to the \ value of its corresponding variable in the primal, as contained in the A, row of it. Thus, y, = 10/3 and J2= 25/3 in the dual, and S, = 10/3 and S, = 25/3 (note that we consider only the absolute values) in the primal. Of course, we do not consider artificial variables because they do not correspond to any variables in the primal, and are introduced for a specific, limited purpose only. Clearly then, if feasible solutions exist for both the primal and the dual problems then both problems have optimal solutions of which objective function values are equal. A peripheral relationship between them is that ifone problem has an unbounded solution, its dual has no feasible solution. - Further, the optimal solution to the dual can be read from the optimal solution of the primal, and vice versa. The primal and dual need not both be solved, therefore, to obtain the solution. This offers.a big computationalpie some situations. For instance, if the primal problem is a minimisation one involving, Say, 3 on would pose a big problem because a large number of surplus and be introduced, The number of iterations required for obtaining the answer the dual, with 7 variables and 3 constraints can be solved comparatively artificial variables would have to would also be large. On the contrary, much more easily. PSST E Given the following linear programming problem and the simplex tableau (Table 4.5) eettaining the solution tot, (i) formulate the dual to the given problem, and (ji) obtain the solution to the dua from the tableau. Also verify that the objective function values of both the problems are same. Minimise Zexytt % Subject to 6x, + 2x, 2 24 3x, + 2x,2 18 xy #3xp2 12 x %2 0 TABLE 4.5 Simplex Tableau: Optimal Solution Basis oy Berens; Ss SA Ay ‘Ay 4 x 1 1 0 3 1/3. oO V3 -3 0 2 S(O 0 0 76 ~ -83 1 116 83 -1 8 ay 1/2 0 1 V2 -1 0 1/2 1 0 6 5 1 12 0 0 0 M M M Solution 2 6 0 0 8 0 0 0 4 0 OF a 16 0. M-12 M-16 M (i The dual corresponding to the given problem: Maximise G=2Ay, + 18y. + 12y5 Subject to 6y, + 3y2 +y3S1 yy + 2yn* 3938 F Vpn Vs20 ex table! (ii) The values of the dual variables y,, y» and y, can be obtained from the A, row of the simp given. These are the respective vatues of ;, S, and S,. Thus, y, = 1/12, y2= M6 and y; = 0. Now. we can deter™ the objective function values of both the problems to get. For primal, Z=2+4x6=5 For dual G=24x W124 18 x 1/6 + 12x0=5only are the prim : Not only ie enc ically, they arc also related in economic sense. We shall now consider the (fora maximisation and then for a minimisation problem. mple 4.1 again, whose pertinent data are reproduced here. Z=40x, + 35x, Profit The Maximisation Problem Consider E, Maximis Subject to 2 S60 Raw materials constraint 4x; +3x2$96 Labour hours constraint 0 The optimal solution to this problem ind ¢ that producing 18 units of A and 8 units of B per week would yield the maximum profit equal to Rs 1,000. Now. suppose that the firm is approached by an individual who would like to rent the facilities of the firm for one werk. The firm has its assets in the form of 60 kg of raw material and 96 labour hours. If we let y represent the rental rate per kg of raw material and y, the rental rate per labour hour, the firm would receive a total rental equal to 60y, + 96y,. We shall compute the minimum value of the rental so that the firm will know as to what minimum o} Lbe economically accépiable to it. The coristraints can be set up by keeping in mind iat io dana teat eS eee ~The rental rates of the resources should be at least as attractive as producing products A and B. We know that production of one unit of 4 requires 2 kg of raw material and 4 labour hours. Thus, the total rental for these amounts of resources should be greater than, or equal to, the profit obtainable from one unit of the product, i.e. Rs 40. Thus, we should have 2y, + 4y, 2 40 Ina similar way, the resources consumed in producing one unit of product B are 3 kg of raw material and 3 labour hours, The total rental of these resources should equal at least Rs. 35, the unit profit of product B. Thus, we should have, 3y, + 3y2 235 Besides, the rentals cannot be negative. Therefore, y; and y, should both be non-negative. In complete form, the problem can be expressed as: Minimise G= 60y, + 96y2 Subject to 2y, + 4y2 2 40 3y, +3y2 235 Yvy2z 0 This problem is absolutely the same as considered earlier—the dual to the given problem. These rental rates 2 and y, are obtainable from the solution of the dual, as y; = 10/3 and y, = 25/3. Also, the values can be obtained from A, row of the simplex tableau containing optimal solution To tre primmal problem. As observed Previously, the objective function values ofthe primal and the dual would be identical. Thus, it isnot surprising {0 observe that the minimurn total rental acceptable by the firm is equal to the maximum profit that it can earn Producing the output itself using the given resources.“The individual rental les correspond shadow prices or imputed prices, indicating the worth ofthe resoure ric ofthe two resources, Materials and labour hours, are imputed from the Seah obiained ee alaing te Services, and bear no relationship with the original cost of the two. mined From wuliing et Ofthe two products 4 and B, we know that each unit of product 4 contributes Rs 40 to the profit. ‘The imputed price of material and labour being, respectively, Rs 10/3 per kg and Rs 25/3 per hour, we observe that the it imputed cost of the resources used in making a unit of the product would be: : 2kg@Rs 103 perkg = Rs 20/3 4 hours @ Rs 25/3 per hour= Rs 100/3 Total =Rs 120/3 = Rs 40 Thus, the total imputed cost of producing one unit of product 4 equals the unit profit obtainable fom i Similarly, from each ‘unit of product B, the total imputed cost of resources employed would be: 3kg@Rsl03perkg _ =Rs 10 3 hours @ Rs 25/3 per hour=Rs 25 Total =Rs 35 unit of the product. It may again be emphasised that the services ofthe ‘This, obviously equals the profit per hat the total value of resources matches the profit realisable from the resources are valued in such a way tl optimal utilisation of their services. ‘The shadow prices are also called the marginal value produets or marginal profitability of the resoures, Thus, if there were a market for renting materials and labour hours, the firm would be willing to take some ris git the price of the material were less than Rs 10/3 perkg, and labour hours, ifthe price payableisless than Rs 25/3 per hour. ‘This is because an additional kg of raw material would add Rs 10/3 tothe profit while a reduction of onekg ao ddseduce the profit by an equal amount. Similarly, an addition of one about hour would cause the profit to increase by Rs 25/3 and a reduction of one hour would reduce profit by Rs 25/3. If we denote marginal profitability of material as MP4, and the marginal profitability of labour as MP, respectively, the shadow prices of the two resources, we can write the dual as follows: Minimise G=60MP,+ 96MP, Subject to 2MP,+ 4MP, > 40 3MP y+ 3MP;, 2 35 MPy, MP,2 0 Now let us tur to the economic significance of the surplus variables Sand 5, in the ual, whose nue values can be obtained from A, row in the optimal solution of the primal. The value of 5, ae oe solution represents the opportunity cost of the product A while the value of S, represents that of | the product 4, production of an additional unit will get the firma profit of Rs 40 and, ‘atthe sametim®s would use up resources worth 2 x 10/3 +4 25/3 = Rs 40. Thus, the net effect of producing one unitof would be 40 — 40 = 0. Similarly, for product B, the opportunity cost equals zero.Further, the 4, value equal to 18 under 5, indicates that each rupee of additional profit in product A would ease the minimum rent acceptable by Rs 18 while in case of, pa minimum ren product B, the rent shall be up by Rs 8. Similarly, reduction in unit profit in products 4 and 2 shal reduce the rent acceptable by Rs 18 and 8, respectively. ! Minimisation Problem minimisation problem, Example 3.3, is reproduced here. Minimise Z= 0x, + 24x, Total cost Subject to 20x + 50x, 24,800 Phosphate requirement 80x, + 50x, 27,200 Nitrdigen requirement Mm? 0 From the optimal solution given in the Table 4.6, reproduced from Table 3.13, it may be noted that the fertiliser requirement of the farmer can be met with at a minimum cost of Rs 3,456, by buying 144 bags only of mixture B. TABLE 4.6 Simplex Tableau: Optimal Soluition p Basis x % Si Se 4, Te b, xy 24 8/5 1 oO 1/50 0 1/50 144 ag 0 60 0 1 “1 al 1 2,400 5 40 24 0 0 M M Solution: oO 144 2,400 0 0 4 B58 0 0 12/25 M The dual corresponding to the problem can be written as: . Maximise G=4,800y; + 7,200y, Subject to 20y, + 80y, < 40 S0y, + 50y2 $24 Yryr2z 0 The optimal solution to this problem is contained in the Table 4,7. — 1 TABLE 4.7 Simplex Tableau: Optimal Solution va Mn 5 So by, 60 0 1 8/5 |. B/S 1 1 0 1/50 12/25) 4,800 7,200 0 0 O65 egies 8/5 0 2,400 0 0 144WiFeAdy mentioned, ioned, these val it alues e i lies can also be obtait se valtics in the objective func RerisunLieee From this ta ol from the solution to the primal problem), Substitul ing the Now, if we impute a price of 0 to each kg of phosphate and aprice of Rs 12/25 to each ky of nitrogen, then the total value imputed to both of these would be Rs 3,456, which is equivalent to the minimum cost of obtai fentliser by the farmer. Thus, on the basis of their cost OF production, we assign @ vatueof vereramd Rs 1228 to cach kg of phosphate and nitrogen in that order. These, again, are the shadow prices of the two fertilis a that are imputed from the (marginal) cost of producing them. 3 we observe that the fertiliser requirements can be met with by using 144 bags of mic nature of the shadow prices of the two mixtures can be From the primal problem, mixture B and none of the mixture 4, The econot considered in more details as follows. Mixture A Each bag of 100 kg contains 20 kg of phosphate a bag of the mixture can be valued as: 20 kg of Phosphate @ Re O perkg =ReO 80 kg of Nitrogen @ Rs 12/25 per kg = Rs 192/5 nd 80 kg of nitrogen. The contribution of each Total =Rs 192/5 since the cost of each bag is Rs 40 and its contribution to the value is only Rs 192/5, obviously, each bag Now, ered mmeans an increase of Rs 40 — Rs 192/5 = Rs 8/5 in cost. For this reason, this mixture has not been used, “Mixture B. The cost of each bag of this mixture is Rs 24 and each bag contains 50 kg each of the two fertilisers. The total contribution of each bag of mixture B can be obtained as follows: 50 kg of Phosphate @ Re 0 Re 50 kg of Nitrogen @Rs 12/25 =Rs24 Rs 24 Total For this mixture, the value matches the cost. Finally, it may be observed that the fertiliser supply has been obtained by using mixtir® B only, whose 144 bags yield 144 x 50= 7,200 kg. of phosphate and an equal amount: of nitrogen, Now, whereas the nitrogen supply exactly equals the required minimum, the phosphates supplied in excess to the extent of 7,200~4,800 273400 kg, It should not be disturbing, therefore, that this has been assigned a shadow price equal to zero. This is obvious because this resource is in excess supply and economic value is attached to only those resources which are scarce. 4:3.1 A Problem with Unutilised Resources .e economic interpretation of the dual when some resources are not fully utilised inte Now we shall turn to th optimal solution, For this, we reconsider Example 3.2. Maximise Z= 5x, + 10x, + 8x3 Subject to 3x, +5x,+2%3< 60 Fabrication hours 4x, + 4x) +4x5S .72 Finishing hours 2x, + 4x + 5x3 < 100 Packaging hours xp, 0 ———- ‘The final tableau containing the optimal solution is shown in Table 4.8 here: TABLE 4.8 Simplex Tableau: Optimal Solution Basis x 2 4 H Sy Ss by % 10 13 1 0 3 1/6, 0 8 % 8g 23 0 1 -13 Si2 0 10 S 0 83 0 0 30-1712 1 18 , 3 10 8 0 0 0 Solution 0 8 10 0 0 18 4 “1B 0 0 213 5/3 0 From this table, we see that the firm should produce none of product 4, 8 units of product B, and 10 units of product C in order to get the maximum daily profit of Rs 160. The shadow prices of the hours in the three departments can be obtained as the numerical values associated with the slack variables in the A, row of the tableau. These are: Rs 2/3 per hour for the fabrication department, Rs 5/3 per hour for the finishing department and nil for the packaging department. Alternatively, these can be obtained from the values of the dual variables ‘YJ and y, respectively. The dual corresponding to the problem and the final tableau depicting its solution are given here: nim Z 23) Minimise G= 60), + 72y2 + 100y, 36 Subject to 3y, +4. +2y32 5 Sa = ze 3 3 | Sy, + dy. + 4y3 2 10 @® 2y, + 4yy + 5y32 8 % YpYo¥s2 0 3 “TABLE 4.9 Simplex Tableau: Optimal Solution Basis ee eee SSS Ay Ap Ay 8 60 1 0 13 Oo -13 13 V6 13. AW 23 0 0 0 8/3 1) -13 23 716 13 23 11/3 72 0 1 17/12. 0 V6 -S/12: V2 6 SA2 5/3, 60. 2 100 0 0 0 M M M lution 28 ‘5/3 0 ns 0 0 oO i) 0 0 oO 18> 0 8 0 .M+4 -M-8 M-10| TAs we can note from this tableau, the values of the dual variables y,,.» and y, are the shadow f i 7 Pric hours in these three departments. With these imputed values, we can make the following observa tVations: (a) For Product A Here the decision reached is not to produce it. This is because it yields a Profit, per unit, while the value of resources it consumes is: oFRss 3hours@ Rs 2/3=Rs 2 4 hours @ Rs 5/3 = Rs 20/3 2hours@Re0 =Re 0 Total =Rs 26/3 Since the total value of resources required by a unit of A is 26/3 while the profit associated i Rs 5, obviously it should not be produced. Any unit of product A produced would lead to a loss of 26/3 — 5 = Rs 11/3 (observe A, value corresponding to x,). (b) For Product B and C_ The value of resources required for their production is as follows: Produt B Product C Rs Rs Shours @ Rs 2/3 = 10/3 2 hours @ Rs 2/3 = 43 4 hours @ Rs 5/3 = 20/3 4 hours @ Rs 5/3 = 20/3 4hours @ Re 0 — 0 5 hours @ Re 0 = 0 Total =Rs10 =Rs8 Profit per unit =Rs 10 =Rs8 For each of them, the opportunity cost is equal to zero. (©) The hours in the packaging department have a zero opportunity cost because there is excess capaci * this department. 4.4 SENSITIVITY ANALYSIS a Equally important as obtaining the optimal solution to a linear programming problem is the questionsstoh the solution. would be affected if the parameters c, b, or a, of the problem change. This question ee : by the sensitivity analysis, also called the post-optimality. analysis, Of course, we may directly subs ‘oie changed values in a given situation and re-solve the problem to determine the effect on the optimal It will be appreciated, however, that it would be a lot advantageous if we vould obtain this if oy graphically or from the tableau containing the final solution to the problem, without being req¥ir® out the whole exercise again. Let us consider as to how can we get the information of this tyPe-