Livear PRrocRAMMING PROBLEM : Smptex METHOD 5.1. INTRODUCTION It has not been possible to obtain the graphical soluti i analytic solution is also not possible because the tools ‘ofanalys arenes eka irepeckens insch cases, a simple and most widely used simplex method is adopted which was developed by Gopi ai The simplex methodt provides an algorithm (a rule of procedure usually involving repetitive applicatio of aprescribed operation) which is based on the fundamental theorem of linear programming, aig Itis clear from Fig. 3.4 (page 78) that feasible solutions may be infinite in number (oecause there are infin number of points in the feasible region, ABCD). So, itis rather impossible to search for the optimum soluti amongst all the feasible solutions. But fortunately, the number of basic feasible solutions are rte in numbet (which are corresponding to extreme points O, A, B, C, D, respectively). Even then, a great labour is required in finding al! the basic feasible solutions and to select that one which optimizes the objective function te The simplex method provides a systematic algorithm which consists of moving from one basic feasible solution (one vertex) to ancther in a prescribed manner so that the value of the objective function is improved. This procedure of jumping from vertex to vertex is repeated. Ifthe objective function is improved at each jump, then no basis can ever repeat and there is no need to go back to vertex already covered. Since the numberof vertices is finite, the process must lead to the optimal vertex in a finite number of steps. The procedure is explained in detail through a numerical example (see Example 2, ch. 5, page 70 Unit). The simplex algorithm is an iterative (step-by-step) procedure for solving LP problems. It consists of — (having atrial basic feasible solution to constraint-equations, Gi) testing whether it is an optimal solution, Cid improving the irst ial solution by set of rls, nd repeating the process ln optimal slain obiained. The computational procedure requires at most m [equal to the number of equations in (3:12)] non-zero variables in the solution at any step. In case of less than m non-zero variables at any stage of computations the degeneracy arises in LP problem. The case of degeneracy has also been discussed in detail in this chapter. Further, itis very interesting to note that a feasible solution at any iteration is related to the feasible solution at the successive iteration in the following way. One of the non-basic variables (which are zero now) at one iteration becomes basic (non-zero) at the following iteration, and is called an entering variable. To compensate, one of the basic variables (which are non-zero now) atone iteration becomes non-basic (ero) at the following iteration, and is called a departing variable. ‘The other non-basic variables remain zero, and the other basic variables, in general, remain non-zero (though their values may change). For convenience, re-state the LP problem in standard form : Max. 2= 01x) + en + os + pty + Otpy 1 + Opa t oo + OF om (5.1) subject to the constraints : aux, + a1 + ayn + an + +A 5.2) 1X) + OaXy + ome + Oraekn +inem = bm + For complete development of ‘Simplex Method’ please see Appendix-A (Theory of Simplex Method ) on page 1119.68 / OPERATIONS RESEARCH 4 20,822 05 on B42 0s By 41 20, mon ppm ZO (5.3) and basic feasible solution of m equations (5-2) is usually takes a, For easiness, an — p= a +nem= Om: For this solution, the value of the objective nt: xyo ns ay = =e 70! N11 och equal to zero) are non-basic variables and remaining variableg ($1) is zero. Here 31.120). yariables (some of them may also have the value zero) Chat tna ds Raed ne SOME MORE DEFINITIONS AND NOTATIONS: ie feasible solution is : The first basic feasible ted LP problem : Max CX , subject to AX = b and X20. = by forthe reform Sette fh column cof mx (n+) matrix A by a; (j= 1, 2, 3, ... .m+™m), so that A=[81, 825+) Amend G4) -singular matrix B, called basis matrix, whose column vectors are m lineat! c sn olumns selected Mog matrix A and renamed as B; . Bz » By, ... » Bm - Therefore, B= (By. Bas -+- + Bd =[@n+1s Qns2 9-0 > Ant m) (5.5) in ible solution, Forint OEE DOs 2.0), (0 1,050 ove 0) one »(0 0, e+ I] =U (identity matrix) ‘The matrix B is evidently a basis matrix because column vectors in B form a basis set of m-dimensional lidean space (E”). ec econd, denote ihe basic variables Xq +1» %n+2»--+ »%n+m DY Xp1 »XB2, --- » 48m Tespectively, to give the basic feasible solution in the form : Xp = (epi s XB2 + XB3y --- Bm) = n+ A429 X43 nt m) (5.6) For initial basic feasible solution, Xp= (by by bs, «.s bm) = right side constants of (5.2). Next, the coefficients of basic variables xp, , Xp2 . ... » 4pm in the objective function z will be denoted by 1 + Cx2 +--+ » Cam Fespectively, so that Ca = (Cais Cas +++» CBm)- For initial basic feasible solution, C= (0,0, ...,0)=0 (null vector) Consequently, the objective function ZH CX + Oak, + C3hy + 2 + Cyn + Ogg | + Op 4 2 +». + Oy 4 m bECOMES. 2=C).0+ 62.0 +... +690 +cBiXB1 + --- + CBB [since x, =x, = x3 Xn = 0} or Z=CpXp- (5.7) Because Cy = 0 (null vector) for initial solution, therefore z=0,X,=b. Since B is an m x m non-singular basis matrix, any vector in E” can be expressed as a linear combination of vectors in B (by definition of basis for vector space). In particular, each vector a (j= 1, 2, ...,n+m) of matrix A can be expressed as a linear combination of vectors B; (i= 1,2, ..,m) in B. ‘The notation for such linear combination is given by ay 9 = xB) +24jB2 +... +.B = 61-B Ba 2 ems +A5.8) nj where xy (i= 1,2,3,..,m) are scal i i= : combination of basis vectors By «Bs re ea aes Ce eee ‘Therefore, xj = B'ay and hence matrix For initial solution, a)= 14, =x; . Next define anew variable, say z), as (%) will change if the columns of (A) forming (B) change. YAH CB) + 4) CB +... + Spy Cag = Bi Xij = Cy Xj. A5.9)UNIT2: LINEAR PROGRAMMING PROBLEM: SIMPLEX METHOD / 69 4;denotes the net evaluation which is computed by the formula : A= 4-6 =CHX)— 6, ww(5.10) Lastly, these notations can be summerized in the following Starting Simplex Table 5.1. Table 5.1 : Starting Simplex Table ga a a on o 0 «wo BASIC Cn Xe Xia) Xa (=a) Xa(=m) Xen Xena = Xoo MIN VARIABLES Bb 5) RATIO: Serre [ome O amb) [an ean) xaeag Ana) 1 . o reerten) [emt O rm (by | (an) an lean) alee) 0 1. 0 Sasa (= Sn) | Chm (= 0) Xm (= bm) | Xmt (= ct) Xm (= dg) Jn (= ne) oO oO at 1 r=CpXy a ra be 0 0 a 0 | a= CeX)~6 Note. Basic variables inthe first column are always sequenced in he order of ‘columns forming the unit matrix. ‘Above definitions and notations can be clearly understood by the following numerical example. 5.2-1. An Example to Explain Above Definitions and Notations Example 1. Illustrate definitions and notations by the linear Programming problem : Maximize 2= x, + 2xy + 345 + Ong + Oxs, subject to 4xy + 2p +35 + 44 = 4, 21 +2xy + 315-5 =8, Solution. First ofall, constraint equations in matrix form may be written as x A i Moma om ows fal oy [4 2411 oyeh[e] 1203 0 -1J\x Xs, or AX=b, A basis matrix B = (B,, B2) is formed using columns ay and ay, so that 1 (4 =(5)-B=a=(})- The rank of matrix Ais 2, and hence ay, a, column vectors are linearly independent, and thus forms a basis for R?. a a ‘Thus, basis matrix is B=(B,.B)=(1 4 } 301 Using (5-4) and (5.8), the basic feasible solution is Xg=B"! b= Ria @ } 28/11)_[ a1 or x0 wre Therefore, basic variables are xg, = 28/11 = x5, x99 = 4/11 =.,, and remaining variables are non-basic (which are always zero) i.e.,.x2 =x4= x5 = 0. Also, coefficient of xg, = coeff. of x5 =, =370 / OPERATIONS RESEARCH 1 =e= can = coetticie Hence =D nis “rome ee 8 ) 88 nt of Xs2 coeff. of 1 reXe= {gsi ft pination of vectors Bi (= 1,2), as linear com| . == ‘an be express Also, any vector a= (j= 1.2.3. 4.5) can 0 GPT Therefore, to express a; as linear combination of Br ay = 128) + 222B2 = 4123 ‘use the result (5.3) to get 1-4/2), (971! Jo *) The anh ni(-3 WG} (an xn fore xy7 = 6/11, x2 = 4/11. : Similar treatment can be adopted for expressing other a's a linear combs Now, using (5.60, the variable = coresponding to vector a can be Obe3 5 o/l ee ana) anon -0.0(S!e(sxi¢+ ir) 2 can also be computed. +272 To compute values of scalars x12 and 22, 1 ‘combinations of B, and Bo- Simi ‘53. COMPUTATIONAL PROCEDURE OF SIMPLEX METHOD ‘The computational aspect of the simplex procedure is first explained by the followi Example 2. Consider the linear programming problem : Maximize z = 3x, + 2x, , subject to the constraints : &, +2) $4.2; — 2) S2,and x, 20. [Kanpur 2000, 96; IAS (Maths.) 92) ‘Solution. Step 1. First, observe whether all the right side constants ofthe constraints are non-negative. If not, it can be changed into positive value on multiplying both sides of the constraints by — 1. In this example, all the 6; °s (right side constants) are already positive. ‘Step 2. Next convert the inequality constraints to equations by introducing the non-negative slack or surplus variables. The coefficients of slack or surplus variables are always taken zero in the objective function. In this example, all inequality constraints being ‘<’, only slack variables s; and s2 are needed. ‘Therefore, given problem now becomes : ‘Maximize z= 3x, + 2x, + 0s; + 0sp, subject to the constraints : tts, ing simple example. -m +8 11255145220. ‘Step 3. Now, present the constraint equations in matrix form : a [i 11 Oj} 2] _f4@ 1-10 1] 3 =(}]. $2. ‘Step 4, Construct the starting simplex table using the notations alread it ghould be remembered that the values of non-basic varinbren ns) ©*Plained in See a 52, zyaxy=0 bere. Column Xp gives the values of basic varialos art al@YS 2er0 at each iterati 7 Fa and s2= 2here. The complete starting basic feasible solution sa nent M4 in the first column So 74, = 4, 522 2.41 = 0.42= 0 and the value ofthe objective fy migeeimmediately read f Tables. in tis ste, the variables sand are corespendng to te coum is eae Note. asic variables. Other variables, x, “ee rarbaevaraerwichaey rans S08) mate 50 vl be catag zero,LUNIT2; LINEAR PROGRAMMING PROBLEM: SIMPLEX METHOD / 71 Table § 2: Starting Simplex Table i : qo 9 2 0 0 BASIC CG x = caged Si ) ” oy M My ‘Xy(S)) Xa ($2) MIN. RATIO VARIABLES | O_o. Xq/Xx for Xe> " 0 4 t 1 ‘TOBE COMPUTED IN NEXT STEP. Se CaXp aiged dened dnt Ano Bay 9= 8X)- 4) Shep 5. Now, proceed to test the Basic feasible solution for optimality by the rules given below. This is donctar computing the 'net evaluation’ A foreach variable x (column vector X) by the formula y= 4-9 = CaX)~ 6 [from (5-10)} Thus, we get ‘ a X= C1 Ay = OX) — C2 Ay = CpX3- 3 bg=0 0.90 ,)-3 =(0,0)(1-1)-2 (0,0) (1 ,0)-0 Eoxisonii =(0x1-0x1)-2 (0x 1+0x0)-0 =- =-2 =0 ‘Remark. Note that in the starting simplex table 4/s are same as (- ¢)'s. ‘so, 4/s corresponding to the colurnns of unit matrix (basis matrix) are always zero. So there is no need to calculate them. (Teall, ) 20 the solution under test will be optimal. Alternative optimal solutions will exist if any non-basic A; is also zero. (ii) If at least one A, is negative, the solution under test is not optimal, then proceed to improve the solution in the next step. ii) Ifcorresponding to any negative Ay, all elements of the column X, are negative oF 2670 ($0) ,then the solution under test will be unbounded. Applying these rules for testing the optimality of starting basic feasible solution, it is observed that ‘A, and A; both are negative. Hence, we have to proceed to improve this solution in Step 6. ‘Step 6. In order to improve this basic feasible solution, the vector entering the basis matrix and the vector oe ved from the basis matrix are determined by the following rules. Such vectors are usually named as ‘incoming vector’ and ‘oulgoing vector’ respectively. “Incoming vector’. The incoming vector X, is always selected corresponding tothe most negative value of Ay (say, ,)-Here Ay = min [Ay , Aa] = min [- 3 ,~ 2} =~3= A Therefore, k= | and hence column vector X; must center the basis matrix. The column X; is marked by an upward arrow (h. “Outgoing vector’. The outgoing vector Bis selected corresponding to the minimum ratio of elements of Xp by the corresponding positive elements of predetermined incoming vector Xx This rule is called the Minimunt Ratio Rule, In mathematical form, this rule can be written as oe _ min} Br >0 Xk - aya i Fork=1, rin 3 2] iF] Ay" X21 wi or ate, aq da Comparing both sides of this equation, we get'r=2. So the vector Bp, ie., Xq marked with downward ae (4) should be removed from thebasis matrix. The: Starting Table 5:2 is now modified to Table 5-3 given low.72. OPERATIONS RESEARCH Table 5:3 key element a E 5) BASIC Ca Xa Xi, % 7X35) 2 MIN. RATIO VARIABLES (B,)__ (Bs). (Xyix,) m 0 4 1 1 1 0 “a a 0 2 H =3 2 0 0 eke | 2=CpXp=0 | (min. A) A 5-4" C6 —Tentering vector Tieaving vector aca Tas Step 7. In order to bring Bz =| : in place of incoming vector X; -[ 1 } unity must occupy in the m faces of X; . If the number in the marked ‘C1’ position is other es ‘key element’. (The element at the intersection of minimum than unin, the key element or pivotelement). ratio arrow the other (remaining) rows, so as to ob be fortified by simple matrix aes fa as ‘D position and zero at all other pl divide all elements of that row by the (€) and incoming vector arrow (1) is called Then, subtract appropriate multiplies of this new row from the remaining positions of the column X;. Thus, the process can follows: The intermediate coefficient matrix is Xp Xt Xt Xs Xu a 4 1 Hi 1 ° Rp 2 1 ~1 o 1 Ry z=0 =3 =2 0 e Apply Ri > Ry — Rp , Ry > Ry + 3Re to obtain na as: Xt Xs X X 2 0 7 ; = 2 1 ia a ; Nowons . ‘ =. = o 3 lH iow, construct the improved simplex table as follows : y Table 54 gq 3 2 0 ° BAS! vaniapies | 2B | MH HS XUlSD MIN-RATIO 7 7 Ba £ a (%0/%2, X2>0) +(2)-----1--=-=1 1--|-2 2 key row 4 = te | wd a 7 : 2=CaXp=6 0 se 7 ; (negative ratio is not counted) t ar From this table, the improved basis amy , ed : : imprcet sei able the improved base feasible solution is read as :x,=2,.)=0,41=2,4=0. The Itis of particular i method. However, ‘iecorelod at ere that 4's are also computed while transforming the table by matrix formula A= CX), 4's can be verified by computing them independently by using the Step 8. Noy Table 5.5. Ww repeat Steps 5 through 7 as and when needed until an optimum solution is obtained in Ae = most negative Aj=~5= A,UNIT 2: LINEAR PROGRAMMING PROBLEM : SIMPLEX METHOD / 73 ‘Therefore, k = 2 and hence X, should be the entering vector (key column), By minimum ratio rule : Minimum Ratio( 8, x)>0)=Min[2,] (ince negative ratio is not counted, so the second X 2 ratio is not considered) Since first ratio is minimum, remove the first vector B, from the basis matrix. Hence the key element is 2. Dividing the first row by key element 2, the intermediate coefficient matrix is obtained as : a 2 Xw x Xs Xs Xa Ri v 0 oO 2 “12 Ok 2 1 =I 0 1 ay 226 0 =3 0 3 Jeay Applying Ry > Ry+ Ry , Ry > Ry + 5R 1 0 7 2 “12 3 1 o 2 12 0 0 5/2 12. Ay Now construct the next improved simplex table as follows : Final Simplex Table 5-5 97 3 2 ° ° BASIC Ce Xe X (By) % (6) Ss St VARIABLES a 2 1 0 2 12 x 3 3 1 o vA 12. z=CpXp=Hl ° 0 2 wa “4 ‘The solution as read from this table is : x =3,.x2=1, 5) =0,s=0, and max. z= 11. Also, using the formula Ay = Cy X; — ¢) Verify that all A,’s are non-negative. Hence the optimum solution is x) =3,a;=1,/maxz=11. Note. If at the optimal stage, itis desired to bring s; n the solution, the total profit will be reduced trom 11 (the optimal value) to 5/2 times of 2 units of 5; in Table 34, Le. 2=11-5/2x2=6. This explains the economic interpretation of nat-evaluations 4, . 54. SIMPLE WAY FOR SIMPLEX METHOD COMPUTATIONS: | Complete solution with its different computational steps can be more conveniently represented by the following single table (see Table 5.6). Table 56 a> 3 2 o o [BASIC VARIABLES| Cy Xe x oc St Ss MIN RATIO ax) n . 4 4/1 re o 2 === 2/1 eMin mend oa an Reno S122 =0 z=CyXpel | 0 0 2 vA Calla, 30 ‘Thus, the optimal solut n is obtained as : x; = 3, x2 = 1, max z=74 | OPERATIONS RESEARCH a1. {ar programming prob! 2. Writo tho stops used in the simplox method. 2. Describe a computational procedure ofthe simplex method forthe solution of a maximization Lp.p, Tips for Quick Solution : 1. In the first iteration only, since Ay: separately by using the formula A; = CpX, - ¢- Mark min (A) by ‘T' which at once indicates the column X; needed for computing the minimum rat io Kanpur (8, ge, " are the same as~ g's, s0 there is no need of calculating i 8 them (Xy/Xp found atthe place where the upward directed arrow ‘T" of min Ayand the left direct d 3. ‘Key element arrow (€) of minimum ratio (Xp/X,) intersect each other in the simplex table. 4. ‘Key element" indicates thatthe current table must be transformed in such a way that the key eleme nt becomes | and all other elements in that. column become 0. 5, Since d,'s corresponding to unit column vectors are always zero, there is no need of calculating then % While transforming the table by row operations, the value of z and corresponding A,’s are also computed atthe same time, Thus alot of time and labour can be saved in adopting this technique, Example 3. Min z=.) ~ 3x2 + 2ts , subject to: xj wg + 3xy S Te— 2k + 4xy S$ 12,— 4x, +319 + Bay $10, and. 22,2920. [Kanpur (B.Sc.) 95, 93, (B.A.) 99) is the problem of minimization. Converting the objective function from minimizationio Solution. Th maximization, we have Max. z= — 2) + 3x) — 2x5 = Max. z’ where-2=2', Here we give only tables of solution. The students are advised to verify them. Table 5:7. Simplex Table je. sil y -2 0 0 BASIC 7], MIN. RATIO variapies) | oH Xe | ™ mH mM Me Xi) 7 ° 3 7 rr 7 ens o ev | 2 4 == =~ 12/4 ein 0 10 al 3 3 oO oO i 10/3 3 t ° 420 0 0 1 o " o o " 1-12 0 0 1351/58/10 The optimal solution is: x; =4 ,x.=5,x3=0,Minz= Example4, Max. z = 3x; + 2x) + 5x3, subject to the constraints : xy + xy +x $ 430 , 3x) + 2xy $460 ,xy +44, $420, and x} , x2,4320- {AS (Main 94))UNIT2; LINEAR PROGRAMMING PROBLEM : SIMPLEX METHOD / 75 Solution. Table 5.8. Simplex Table oo a 2 5 ° 0 o [Basic : ‘ MIN variapies | Ce Xe f XM Xe Me ar x 0 a0 7 2 43071 eas o 60 3 ~ = 40072 aS o 420 L 4 = i z=0 3 2 ey ae 2 too | - 174 T 0 2-140 y 5 230 2 0 ' 0 2 0 36 0 20 a 0 0 2 1 1 ever 251350 4 0 0 1 2 0 420 L_ =o Since all A, 2 0, the solution is : 0, x2 = 100 , x3 = 230, max z= 1350. ExampleS. Solve the LP problem :Max. z = 3x; + 5x2 + 4x3, subject to the constraints : Day + By SB, Bey + Sxq S10, 3xy + 2xy + 4x3 S15, and x, 42,4520. [Tamilnadu (ERODE) 97; Rewa 93; Kanpur (B.Sc.) 92; (B.A.) 90, Meerut (M.Sc. Stat. & B.Sc. Math.) 90] Solution. Aficr introducing slack variables, the constraint equations become : 2x, + 3x +X = 8 Dry + 5x5 +45 3x, + 2g + 445 +%6 Table 5-9. Starting Simplex Table Pe 3 5 4 ° o BASIC Cy Xn x MIN RATIO. (Xw/X2) VARIABLES ra ° 3 2 7 as o 10 o 102 % 3 2 4 ° 1572 nag = > 4 0 0 “4 =0 7 J Incoming outgoing ° Gj wan voce op WH Now apply short-cut method for minimum ratio rule (min Xp/X,), and find the key element 3. This key clement indicates that unity should be at first place of X; , so the vector to be removed from the basis matrix is Xq. Now, in order to get the second simplex table, calculate the intermediate coefficient matrices as follows : First, divide the first row by 3 to get R 73 ya) T 0 1 0 0 R 10 0 2 3 ° 1 0 Re 1s 3 fa 6. 0 0 1 Rs 0 =3 -4 0 0 0 Applying Ry —> Ry —2R, . Ry > Ry— 2R, , Re > Re + Ri \76 | OPERATIONS RESEARCH — me CU 0 77) : . R ws 7 : e hs ws a ; ; : Ry 29/3 ~ : + - x é 1 . k wi 3 [ — 1 i .10) is constructed as below : Now the sevond simplex table (Table 5.10) is construct o> 3 5 4 0 0 i : tt iz e ms MINRATIO “se To! Um | OC 2 TS VARIABLES cae : : = Tar a ze] 4g ua] wn 0 Xe 0 29/3 53 0 4 = cy 3 0 4 mene ae Incoming Outgoing Now verify that y= CX, 61 = 3 + (5 0, 0) 2/3, - 4/3, 5/3)= 173 Aya CyXy— 69 = 4 + (50,0) 05,4) =- 4 Ag=CyXq—€4=0+(5 0,0) (1/3 2/3, - 2/3) =5/3. lement is found to be 5. Hence the vector to be removed from the basis matrix is Xs. Thus The key-el u ‘i proceeding exactly in the same manner, the remaining simplex tables are obtained (Tables 5.11 and 5.12). Table 5-1 BASIC Cy Xe Xt x Xs Xe Xs Xs MINRATIO RIABLES. 2 5 7 v3 1 0 v3 0 0 22 2 3 a 4 14/15 4/15 0 1 -2S vis 0 - om 0 89/15 AI/IS. o oO -2AS -4/5 1 9a 15/15 jem=m=0) 2=25671I5 | —Ti7IS* 0 0 WAS a5 0 ary 1 4 Tncoming ‘Outgoing Table 5.12. Final Simplex Table [_ Basic GX x % % Xs Xs Xe MIN VARIABLES RATIO. % soft 0 H 0 15741 eal 10/41 4 4 pv | oo ° 1 -w4l sl al ¥ ‘goat || I oO 0 -wal —1A4t 15/4 2 = CyXy= 765/41 0 oO o 45/41 24/41 NAl 420 Since all 4, 2 0 , the solution given by x, = 89/41 , x) = 50/41 ,x;= 62/41 , max z= 765/41 , is optimal. Example6. Minimize z= x, — 3x3 + 2x5 ,subject to the constraints : We 3t2— 45 + Des $7, ~ Dey + 4x5 S12 dey + 3x + 8x5 $10 and x2,23,4520. [NTU (Mech.) 99; Kanpur 96; Madural B.Sc. (Comp. Sc.) 82, (Appl. Math) 85; Kerala B.Sc. (Math.) 90] Solution. Equivalently, max z’ =~ x5 + 3x. i emax 2 =~ = 2x = variables, the constraint equations become? S NeT® # ——% Introducing 21» x4 and.x a8 slack 1+ 34p x5 + Oxy + 2x +036 = 7 Oxy ~ Dey + dics +24 + Oxs + Oxg = 12 Oxy ~ 49 + 3x5 + Og + 8rs +26 = 10,UNIT2: LINEAR Pi Now proceeding as in above example the simplex computations are performed as follows : Table 613, ~ BASIC VARIABLES, Cn Xe Sir ar eee ee MIN RATIO (%p/Xs) T zo “148 10/3 4 a) 0 - x 5 Ww 0 1 wo ws oO Heed o " Io fo zt CE UWorz=-1 Ws 00 3 ~1S 0 820 Thus, optimal solution is :x2 = 4 ,x3=5,x5=0, min, 2=- 11. Alternative forms of Example 6 : (i) Min, z= x, — 3x + 2xy, subject to 3x) — x2 + 2xy $7, — 2x1 + 4x S12, — 4x) + 3xy + By S10 and ay ay 320. (i) Min, 2 = x3 ~ 3xy + 2s , subject to the constraints : Ay + Bay xy + Deg = 7,— Dey + 4acy xy = 12, — Any + Bay + Bis +46 = land ny, 29, ... 44620. Example 7 (Bounded Variables). A manufacturer of three products tries to follow a policy of producing those which continue most to fixed cost and profit. However, there is also a policy of recognising certain minimum sales requirements currently, these are : Product x, % Xs Units per week : 20 30 60 There are three producing departments. The product times in hour per unit in each department and the total times available for each week in each department are : [ Departments Time required per productin hours Total hours available i Xi x X 1 025 020 015 20 030 040 050 toss L 3 025 030 025 529 The contribution per unit of product X, , Xa, Xz is Rs. 10-50, Rs. 9-00 and Rs. 8-00 respectively. The company has scheduled 20 units of Xy , 30 units of X, and 60 units of Xs for production in the following week, you are required to state: (0) Whether the present schedule isan optimum one from a profit point of view and if ts not, wha it should be; (ii) The recommendations that should be made to the firm about their production facilities (following the answer to (i) above). Solution. The formulation of the problem is as follows : Maximize z = 10-5X, + 9X + 8X, Subject to the constraints : 0-25X, + 0-20X, +0-15X,5 420 0:30X, + 0-40X, +.0:50X3 $ 104878 | OPERATIONS RESEARCH 0-25X, + 0:30X2 + 0:25X3S 529 0s X,220, 05X22 30,05 X32 60. ready producting minimum of X2 and X3 it should, at least, produce maximum of i Lower bounds are specified inthis problem, i¢.,X; 2 20, X22 30, X260, introducing the new variables x; , x, and x3 such that 20 + x3, Xp = 30-+¥2X3 = 60+ 25. i it of x, .X2 »%3 the problem now becomes : eels Mosh von i Be conta, abject to the constraints : 0-25x; + 0°20x, + 0-15x3 $400, Maximise dy + 050%; $1000 025%) + 030%: + 0.2513 $ 500 ,and x, 20x) 20,1320 Th vadents may ant proceed to find the optimal solution by simplex method in the usual manner. _ Example 8. Fora company engaged in he manufacture of three products, viz. X, Yand Z, the available data are given below: ones Product: X 7 Zz Min. sales requirement permonth: 10 20 30 Since the company is a X; limited by the first constr i ‘Thiscan be handled quite ewsily by Operation, Required Processing Times and Capacity [Operations Time (hrs.) required per item of Total available hours per month x Y Zz 1 ' 2 200 2 2 1 1 220 3 3 1 2 180 Profit (Rs.) per unit Product: =X Z Profit (Rs.)/unit 10 15 8 IC.A. (Nov.) 89] Find out the product-mix to maximize profit. units produced per month for the products X, Y and Z, Solution. Let x, yandz denote the number of respectively. ‘Minimum sales requirements give the constraints :x 2 10, y 220,22 30, wherex,y,z20- Operations, processing times and capacity lead to the following constraints : Peay 22200.) Wx+y+2$220.1ii) — 3x+y+ 22S 180. o(it) ‘The objective function is: Max. P= 10x + 15y + 8c . Thus we have to solve the following problem: Max P= l0x+I5y+82, subject tox++ 2y + 2z $ 200, 2x + y+2zS 220, 3x+y +228 180, and O0 and no artificial variable appears in the basi ~15/2x, + 3xq + Oxy + Ox4 + Oxs Now apply simplex method in the usual manner. Phase 2: Table 5:23 an optimum solution to the auxiliary sider the actual costs associated with the original variables, the objective g> = 1572 3 ° 0 0 BASIC MIN RATIO VARIABLES Xs X Xs Xs Xs (Ky/X) ” 1 172 0 = = = o =12 1 “4 -uA 0 3/4 0 15/8 15/8 <4 Since all Aj 2 0 , an optimum basic feasible solution has been attained. Hence optimum solution is : x; /4 42 =0,x5= 3/4, minz=75/8. (7 EXAMINATION PROBLEMS Solve the following LP problems by two-phase method 1. Max. 2= 3% — a 2 subjectto the constraints : 2x4 22 X14+3%S2 esd and x1, x20. TAns. = 2, x2=0Max z= 6) Max, 2 = 8x1 + 8x2 a. subject tothe constraints : xy + 2x23 Xt dey xy +285 and x; ,x220. [JNTU (Mech. & Prod.) 2004) [Ans. x)= 0, %=5 , max. z= 40] Max z= x; + 1.5x2 + 2x3 + 5x4 with the conditions : 3x; + 2xp + 45 + x4 5 6 2x +a + HG + BKG 4 2xy + 6x2 ~ BX + 4x4 =0 y+ Bx ~ Axg + 3% = 0 +2,3,4)20 [Ans. x; = 1.2, 2=0, %5=0.9 x4=0, max. 2= 19.8) 4. Minimize z= x; 21-93%, subjectto 5. Max.2=3% 42% +%5+4% 6. Max z= Sxy~ 2x0 + 3x — 2xy + Xo + 3xy=2 ‘subject to ‘subject to 2x +314 4y=1, xy + 5g + 45 - 9% = 5 2x, + xp - 22 420,J=1,2.3, 2x — Bx, — 49 + 5X4 Bx-4y $3 [Ans. Here all 4/20, but at the Xy + 4% + 2.5% - 44 = 6 2+ 9%qS5 ‘same time artificial variable a; Xe ZO Hy Xe Xe MZO~ appears in the basis. Hence the [Ans. No solution} [AIMS (BE Ind.) Bang. 2002] given LPP has no feasible solution) [Ans. x, = 23/3, 2 =5,%=0, max. z= 85/3) 1, Max. 2= 2x; + 3x + 5x3, B. Max. 500 x; + 1400 x2 + 900X3,, ‘subject to the constraints : ‘subject to, Bxy + 109 + 5x95 15, Xi + Xe + X= 100 X + Be tmz 4 12x) + 95x2 + 15% 225 38x ~ 10% + 9x5 93, Bx + 3x + 4H 26; Hs 1m20 2X1, %, %92 0. [Meerut (MA) $3] [Ans. There does not exist any feasible solution, because artifical vartable Is not removed in the problem] 9. _Afirmhas an advertising budget of Rs. 7,20,000. It wishes to allocate this to two media : magazines and televisions, $0 that total exposure is maximized. Each page of ‘each spot on television i each spot on t {[Hint. The problem is Max. 2 9.000%; + 12,000% < 7,20,000, 2X; = 10. of pages of magazine 2 = No.of spots on television} where vision costs Rs. 12,000. An addi magazine advertising be used and atieast3 spots. ‘magazine advertising is ‘on television, Determine the optimum: 160,000x; + 12,000%28.t A 2 2x2 2 3,m He 2 0, fans. x, =2, %¢= 585 and max. z = 7,14,000] to result in 60,000 exposures, whereas stimated to resull in 1,20,000 exposures. Each page of magazine advertising costs Rs. 9,000 and tonal condition that the firm has specified is that at least two pages of ‘Media-mix for this firm. [Delhi (MBA) 97)