Chapter Two Linear programing (LP) infotesfish@gmail.com History of LP LP was developed by the Russian mathematician L. V. Kantorovich in 1939 and extended by the American mathematician G. B. Dantzig in 1947 at air force. The original name for this technique, "programming in a linear structure," which was later shortened to "linear programming." Meaning of LP LP is a mathematical technique for choosing the best alternative from the a set of feasible alternatives. Linear programming (LP) problems are optimization problems where the objective function and the constraints of the problem are all linear. Linear programming is the subject of studying and solving linear programs. DEFINITION Linear Programming(LP) is a field of management science or OR that finds most efficient way of using limited resources to achieve the objectives of a business. Optimization Characteristics of Optimization Problems One or more decisions Restrictions or constraints e.g. Determining the number of products to manufacture a limited amount of raw materials a limited amount of labor Objective – The production manager will choose the mix of products that maximizes profits – Minimizing the total cost Expressing optimization problems mathematically Decision variables X1 , X2 , X3 , … , Xn e.g. the quantities of different products Index n = the number of product types Constraints – a less than or equal to constraint : – a greater than or equal to constraint : – an equal to constraint : Objective – MAX(or MIN) : f(X1 , X2 , X3, …, Xn) f(X1 , X2 , X3 , … , Xn) < b f(X1 , X2 , X3 , … , Xn) > b f(X1 , X2 , X3 , … , Xn) = b Mathematical formulation of an optimization problem MAX(or MIN) : f(X1 , X2 , X3 , … , Xn) Subject to: f(X1 , X2 , X3 , … , Xn) < bm f(X1 , X2 , X3 , … , Xn) > bm f(X1 , X2 , X3 , … , Xn) = bm note : n variables , m constraints Requirements For Application Of Linear Programming 1. The objective should be clearly identifiable in mathematical terms. 2. The activities involved should be represented in quantitative terms. 3. Limited availability/ constraints should be clearly spelt out. 4. The relationships between the objective function and the resource limitation consideration must be linear in nature. 5. Feasible alternative courses of action should be available to the decision makers that are determined by the resources constraints. ASSUMPTIONS UNDERLYING LINEAR PROGRAMMING 1. Proportionality 2. Additivity 3. Continuity 4. Certainty 5. Finite choices 6. Non-negativity ADVANTAGES OF LINEAR PROGRAMMING 1. LP helps a decision maker to ensure effective use of scarce resources 2. 3. 4. LP techniques improve the quality of decision making. It generates large number of alternate solutions This technique can also cater for changing situations. The changed conditions can be used to readjust the plan decided for execution and so on. Model formulation A linear programming model consists of certain common components and characteristics. Components of LPM: Decision variables, Objective function Model constraints, which consists d/t DV and parameters Decision variables are mathematical symbols that represent levels of activity by the firm. Contd The objective function is a linear relationship that reflects the objective of an operation. The objective function always consists of either maximizing or minimizing some value (e.g., maximize the profit or minimize the cost of producing items). A constraint is a linear relationship that represents a restriction on decision making. Parameters are numerical values that are included in the objective functions and constraints. STEPS IN FORMULATION OF LP PROBLEMS There are three basic steps in formulation of LPM Step 1 : define the decision variables how many x1, x2, x3……xn to produce Step 2 : define the objective function maximize profit or minimize a cost Step 3 : define the constraints resources available to produce something A maximization model example SYNERGY Company is a small crafts operation run by an American natives. The company employs skilled artisans to produce clay bowls and mugs with authentic Native America designs and colors. The two primary resources used by the company are special pottery clay and skilled labor. Given these limited resources, the company desires to know how many bowls and mugs to produce each day in order to maximize profit. The two products have the following resource requirements for production and profit per item produced (i.e., the model parameters): Resource Requirements product Bowl Mug Labor (hr/unit) Clay (lb/unit) 1 4 2 3 Profit/unit 40 50 Required: formulate linear model There are 40 hours of labor and 120 pounds of clay available each day for production. Summary of LP Model Formulation Steps Step 1. Define the decision variables How many bowls and mugs to produce Step 2. Define the objective function Maximize profit Step 3. Define the constraints The resources (clay and labor) available Solution o Decision Variables how many bowls and mugs to produce X1: numbers of bowls to produce X2: numbers of mugs to produce o The Objective Function maximize total profit maximize Z = $40x1 + 50x2 Where, Z= total profit per day $40X1= profit from bowls $50X2= profit from mugs Contd Class work DeReal wood co., produces wooden soldiers and trains. Each soldier sells for $27, uses $10 of raw materials and takes $14 of labor& overhead costs. Each train sells for $21, uses $9 of raw materials, and takes $10 of overhead costs. Each soldier needs 2 hours finishing and 1 hour carpentry; each train needs 1 hour finishing and 1 hour carpentry. Raw materials are unlimited, but only 100 hours of finishing and 80 hours of carpentry are available each week. Demand for trains is unlimited; but at most 40 soldiers can be sold each week. How many of each toy should be made each week to maximize profits. Answer Decision variables completely describe the decisions to be made (in this case, by DeReal). DeReal must decide how many soldiers and trains should be manufactured each week. With this in mind, we define: x1= the number of soldiers produced per week, x2= the number of trains produced per week, Contd Objective function: maximizing the total weekly profit (z). Here, profit equals to (weekly revenues) – (raw material purchase cost) – (other variable costs). Weekly profit from soldiers toys: 27-(10+14)= 3 Weekly profit from Train: 21-(9+10)= 2 Hence DeReal’s objective function is: Max z = 3x1+ 2x2 Cont Constraints: Here there are three constraints: – Finishing time per week – Carpentry time per week – Weekly demand for soldiers non-negative values: (DeReal can not manufacture negative number of soldiers or trains!) Contd The complete Linear Programming (LP) problem Max z = 3x1+ 2x2 (The Objective function) st: 2x1+ x2 ≤100 (Finishing constraint) x1+ x2 ≤80 (Carpentry constraint) x1 ≤40 (Constraint on demand for soldiers) x1, x2 >0 (Sign restrictions) A minimization model example A farmer is preparing to plant a crop in the spring and needs to fertilize a field. There are two brands of fertilizer to choose from, Super-gro and Crop-quick. Each brand yields a specific amount of nitrogen and phosphate per bag, as follows: Chemical contribution Brand Nitrogen (lb/bag) Phosphate (lb/bag) super-gro 2 4 Crop-quick 4 3 The farmer's field requires at least 16 pounds of nitrogen and 24 pounds of phosphate. Super-gro costs $6 per bag, and Crop-quick costs $3. The farmer wants to know how many bags of each brand to purchase in order to minimize the total cost of fertilizing. Summary of LP Model Formulation Steps Step 1. Define the decision variables How many bags of Super-gro and Crop-quick to buy Step 2. Define the objective function Minimize cost Step 3. Define the constraints The field requirements for nitrogen and phosphate Contd Decision Variables x1 = bags of Super-gro x2 = bags of Crop-quick The Objective Function minimize Z = $6x1 + 3x2 where $6x1 = cost of bags of Super-gro $3x2 = cost of bags of Crop-quick Contd Class work DeReal makes luxury cars and jeeps for high-income men and women. It wishes to advertise with 1 minute spots in comedy shows and football games. Each comedy spot costs 50birr and is seen by 7M highincome women and 2M high-income men. Each football spot costs 100birr and is seen by 2M high- income women and 12M high-income men. How can DeReal reach 28M high-income women and 24M high-income men at the least cost. Answer: The decision variables are x1 = the number of comedy spots x2 = the number of football spots. Giving the problem min z = 50x1 + 100x2 St: 7x1 + 2x2 ≥ 28 2x1 + 12x2 ≥ 24 x1, x2 ≥ 0 LP: Graphical method Solution of Linear Programming Problems The linear programming problems can be solved to determine optimum strategy by two methods- Graphical and Simplex method. GRAPHICAL METHOD Graphical method is suitable when there are only two decision variables. Models with three decision variables can be graphed in three dimensions, but the process is quite cumbersome, and models of four or more decision variables cannot be graphed at all. STEPS IN GRAPHICAL METHOD Step I. Formulate the LP Problem as explained in previous class. Step II. Convert the inequalities in to equalities to obtain graphical form of the constraints. (Draw the line of each constraint, first putting x1=0 to find the value of x2 and then putting x2=0 to find the value of x1. Then draw the line for the values of x1 and x2 which represents the particular constraint. Once the lines are drawn for all the constraints, identify the feasible polygon (area) by shading the area below the line for the constraint < and shading above the line for the constraint > type). Contd Step III. Identify the extreme points of the feasible polygon and name the Corners. Step IV. Evaluate the objective function Z or C for all points of feasible region. Step V. In case of maximizing objective function Z, the corner point of feasible region giving the maximum value of Z becomes the value of decision variables. Similarly in minimizing case, the point of minimum value of C gives the answer. Example 1 (problem of Maximizing Z) Two commodities P1 and P2 are to be produced. The profit Margin on P1 is $ 8 and on P2 is $ 6. Both the commodities are required to be processed through two different machines. Sixty hours of time are available on I machine and forty eight hours of time are available on II machine. One unit of P1 requires 4 hours of time in machine I and 2 hours of time on machine II. Similarly, one unit of P2 requires 2 hours of time on machine I and 4 hours of time on machine II. Determine the number of units of P1 and P2 to be produced in order to maximize the profits using graphical method? solution Step I. Formulate LP Problem. The information available can be put into structural matrix form as follow. Commodities Requirement P1 4 2 8 Machine I Machine II Profit $ per unit Total P2 2 4 6 Let x1 be number of units to be produced for P1 Let x2 be number of units to be produced for P2 60 48 - DV X1 and X2 are unknown decision variables. Max Z= 8x1 + 6x2 st: 4x1 + 2x2 < 60 2x1 + 4x2 < 48 x1, x2 > 0 Objective Function Resource constraints non-negativity condition. contd Step II. Convert constraint inequalities in to equalities, draw respective lines and determine feasible polygon (area). Taking constraint (i) ∴ 4x1= 60 or x1 =15 X1 X2 0 30 ∴ 2x2 = 60 or x2 = 30 15 0 By these coordinates (15,30) we get line BD in graph. Similarly, taking constraint (ii). 4x1 + 2x2 = 60 2x1 + 4x2 = 48 X1 X2 0 12 24 0 ∴ 2x1= 48 or x1 =24 ∴ 4x2 = 48 or x2 = 12 By these coordinates (24,12) we get line AE in graph. contd Now, any point on line BD satisfies (i) constraint and any point on line AE satisfies (ii) constraint. The constraints cannot be violated, they must be satisfied. Any solution which satisfies all the know constraints is called optimal solution. Since both the constraints are of the type < hence any point on the right hand side (RHS) of BD or AE becomes infeasible area/solution for which we are not concerned. Contd Feasible region contd Step III. Both the constraints are to be satisfied simultaneously, therefore, OACD becomes the region of feasible solution. This is also known as feasible polygon. – On line OA, point A give maximum profit, on line OD, point D gives maximum profit. contd Step IV. Evaluate the objective function Z= 8x1 + 6x2 for all points of feasible region i.e. O,A,C,D. At point O profit is zero ∴ Z=O At point A x2=12, x1=0 ∴ Z=12x6=72 At point D x1=15, x2=0 ∴ Z=15x8=120 At point C x1=12, x2=6 ∴ Z=12x8+6x6=132 (from graph) Step V. Z is maximizing objective function, hence the point with maximum value of Z is the optimal solution point. – Therefore at point C (Z=132) with x1=12 and x2=6 is the optimal point. A maximization model (class work) SNERGY Company is a small crafts operation run by an American natives. The company employs skilled artisans to produce clay bowls and mugs with authentic Native America designs and colors. The two primary resources used by the company are special pottery clay and skilled labor. Given these limited resources, the company desires to know how many bowls and mugs to produce each day in order to maximize profit. The two products have the following resource requirements for production and profit per item produced (i.e., the model parameters): Resource Requirements product Bowl Mug Labor (hr/unit) Clay (lb/unit) 1 4 2 3 Profit/unit 40 50 Required: formulate linear model, solve graphically, find the optimum point There are 40 hours of labor and 120 pounds of clay available each day for production. answer Contd X2 The labor constraint area 60 50 Letting X1 and solving for X2 40 Let’s first consider labor constraint line first 30 1(0)+2X1=40 20 X2= 20 10 1X1+2(0)=40 0 10 20 30 40 X1=40 (40,20) X2 The constraint area for clay Then, let’s consider clay constraint 60 50 4(0)+3X2=120 40 X2=40 30 4X1+3(0)= 120 20 X1=30 10 (30, 40) 0 10 20 30 40 50 60 X1 50 60 X1 Contd The feasible solution area is an area on the graph that is bounded by the constraint equations. X2 60 50 40 Common area to both constraints 30 20 10 X1 0 10 20 30 40 50 60 The Optimal Solution Point After plotting the graph, the next step in graphical solution method is locate the point in the feasible solution are that will result in the greatest total profit. Contd Let’s assign letters to each corner 0ABC are the feasible region X2 60 50 40 30 4X1+3X2=120 A 20 B 10 1X1+2X2=40 C 0 10 20 30 X1 40 50 60 Calculate the value of each corner (O, A, B and C) to get the optimal solution Max Z= $40X1+$50X2 At point 0, Profit is 0 (substitute both X1 and X2 by 0 in the OF) At point A, profit is 1000 (substitute X1 by 0 and X2 by 20) At point B, profit is 1360 (substitute X1 by 24 and X2 by 8) At point C, profit is 1200 (substitute X1 by 30 and X2 by 0) Thus the optimal solution is point be (the firm should produce 24 bowls and 8 mugs so as to meet its objective). Contd Let’s assign letters to each corner 0ABC are the feasible region X2 60 50 40 30 4X1+3X2=120 Optimum solution A 20 B 10 The optimal solution is the best feasible solution. 1X1+2X2=40 C 0 10 20 30 X1 40 50 60 LP: Cost Minimization A minimization problem minimizes the value of the objective function rather than maximizing it. Minimization problems generally involve finding the least- cost way to meet a set of requirements. Problem of Minimizing C Minimize C= 50x1 + 20x2 Subject to 2x1 – x2> 0 x1 + 4x2 > 80 0.9x1 + 0.8x2 >40 Where x1 ,x2 > 0 non- negativity condition. Solution: The first step is skipped as LP problem is already formulated. We will follow other steps simultaneously. In constraint (i) 2x1 –x2 > 0 there is no constant value, hence it must pass through the origin. First convert it into equality. 2x1 –x2 > 0 . Now give x1 any arbitrary value. When x1 =0, x2=0 x1 =1, x2=2 x1 =2, x2=4 and so on. contd We draw the line with these coordinates and get line I drawn in the graph passing through origin. Now, convert constraint (ii) in equality x1 + 4x2 = 80 When x1 =0, x2=20 X2 =0, x1=80 We draw the line II (80, 20) as shown in graph. Now, convert constraint (iii) in equality 0.9x1 + 0.8x2 =40 When x1 =0, x2=50 X2 =0, x1=44.4 We draw line III (44.4, 50) as shown in graph. contd contd For feasible area we need to examine all the there constraints equations (Note, all are > type) In equation (i) if we move vertically upward, meaning x1=o and x2 increasing, the equation becomes negative or less than, which is not permitted. Hence feasible area should be on RHS. In equation (ii), the feasible area should be above the line because it is greater than the sum of x1 and x2. Similarly in equation (iii) it is on the RHS therefore feasible area (region) is indicated by three rows or shading and extends upto infinity. Contd Now we have to find out different values of Z at different corner points, B,C,E by finding out their coordinates (x1, x2) then putting them in objective function Z. The point which gives the minimum value is the answer. At corner B x1=16,x2=32 therefore Z= 1440 At corner C x1=34.4,x2=11.4 therefore Z= 1948 At corner E x1=80,x2=0 therefore Z= 4000 From the above we can see that minimum value of Z is at point B where x1=16 and x2 =32 and hence it is the answer. Class exercise feeding farm animals. Animals need: 14 units of nutrient A, 12 units of nutrient B, and 18 units of nutrient C. Two feed grains are available, X and Y. A bag of X has 2 units of A, 1 unit of B, and 1 unit of C. A bag of Y has 1 unit of A, 1 unit of B, and 3 units of C. A bag of X costs $2. A bag of Y costs $4. Required: 1. Minimize the cost of meeting the nutrient requirements. 2. Solve the problem graphically 3. Find the optimal solution in the graph Answer Feed grains Nutrients cost X A 2 B 1 C 1 2 Y 1 1 3 4 Total needs 14 12 18 Change inequalities in to equalities X1 X2 0 14 7 0 (7, 14) X1 X2 0 12 12 0 (12, 12) X1 X2 0 6 18 0 (18, 6) Plot the graph X2 14 A 2X1+X2= 14 12 10 2X1+X2= 14 B 8 6 C 4 2X1+X2= 14 2 D 0 2 4 6 8 7 10 12 14 16 18 X1 Optimum solution Simplex method Simplex method of solving LP When a large number of variables (more than 2) are involved in a problem, the solution by graphical method is difficult/ not possible. – The simplex method provides an efficient technique which can be applied for solving LPPs of any magnitude, involving two or more decision variables. – In this method, the objective function is used to control the development and evaluation of each feasible solution of the problem. The simplex Algorithm is an iterative procedure for finding, in a systematic manner, the optimal solution that comes from the corner points of the feasible region. – Simplex algorithm considers only those feasible solutions which are provided by the corner points and that too not all of them. – It is very efficient algorithm. – The technique also has the merit to indicate whether a given solution is optimal or not. – Was formulated by G.B. Dantzig in 1947. For application of simplex method, following conditions must be satisfied. • Right Hand Side (RHS) of each constraint should be non-negative. In case of negative RHS, the whole solution (inequality) to be multiplied by-1. • Each of the decision variables of the problem should be nonnegative. In case of ‘unrestricted’ variables it is treated as the difference of two non-negative variables-such as xl, x2 > 0, x3 unrestricted can be written as xl, x2, x4, x5 > 0, where x3 = x4 –x5 , After the solution is reached, we substitute difference of x4 and x5 as x3. Basic Definitions Before using the simplex method, let us examine and understand certain basic terms involved in the procedure. 1. Standard Form: represents the linear relationships of objective function and constraints, making RHS of constraints as equal produces standard form, involves converting inequality situation (canonical form) into standard form 2. Slack and Artificial Variables: are designated as S1, S2 . . . . etc. and A1, A2 etc. respectively. Whereas the slack variables indicate spare capacity of the constraints, artificial variables are imaginary variables added for standard form. 3. Surplus Variable: A variable subtracted from the left hand side of a greater than or equal to constraint to convert the constraint into equality. Physical sense or interpretation of the surplus variable is that it is amount of resource over and above the minimum required level. In case the constraint inequality is of the type "less than or equal to", then it is called slack variable. 4. Basic Solution: There may be n variables and m constraints in a linear programming problem. When we evaluate the solution of this problem by setting (n - m) of the variables to zero and solve the other m variable equations, we obtain a unique solution. It is called "Basic Solution". 5. Basic Feasible Solution: When a basic solution satisfies even the nonnegativity requirement is called Basic Feasible Solution. 6. Simplex Table: A table used for calculations during various iterations of the simplex procedure, is called Simplex table. 7. Variable Mix: The values of the column that contains all the variables in the solution. 8. Basis: The set of variables which are not set to zero and figure in the column of "Product Mix" are said to be in the 'Basis'. 9. Iteration: steps of moving from one solution to another to reach optimal solution are called Iterations. 10. Cj Row: It is the row containing the coefficients of all the variables (decision variables, slack or artificial variables) in the objective function. 11. Constraints: Restrictions on the problem solution arising from limited resources. 12. Cj - Zj = ∆j or Index Row: The row containing net profit or loss resulting from introducing one unit of the variable in that column in the solution. A positive number in the ∆j row would indicate an algebraic reduction or increment in the objective function if one unit of the variable of that column is introduced in the basis. 13. Pivot -Column: The column with the largest positive number in Cj - Zj row in a maximization problem or the smallest number in a minimization problem is called Pivot column. This indicates the variable entering the solution in the next iteration by replacing an appropriate variable. 14. Pivot Row: When we work out the ratio of quantities bi's and the elements of the Pivot column, we get the last column of the simplex table. The outgoing variable to be replaced by the entering variable (decided by the key row) would be the one with the smallest positive value of the ratio column. 15. Pivot Element: The element at the point of intersection of the key column and the key row is called the Pivot element. 16: Optimal Solution: The best of all feasible solutions. 17: Linear Function: A mathematical expression in which a linear relationship exists amongst various variables. Standard form of LP Problem: In order to develop a general procedure for solving any linear programming (LP) problem, we first introduce the standard form. Let us assume the decision variables as x1, x2, x3 . . . xn such that the objective function (Linear) of these variables assumes an optimum value, when operated under the given constraint of resources. Thus, the standard form of LPP can be written as follows. Objective Function Optimise (Maximise or minimise) Z = C1 x1 + C2 x2 + . . .+ Cn xn where Cj (j = 1,2,. . . . . . . . . n) are called cost coefficients. Constraints (linear) St: a11 x1 + a12 x2 + a1n xn = b1 a21 x1 + a22 x2 + a2n xn = b2 ......................... aml x1+ am2 x2 + amm xn = bm. Where bi (i = 1, 2…m) are resources constraints and constants aij (i = l,2,….m; j = 1,2,…..n ) are called the input output coefficients. Slack and Artificial variables : Normally constraints are in the form of inequalities or equalities, when constraints are in the inequality form, we use imaginary variables to remove these inequalities and convert the constraint to equation form to bring in deterministic nature of resources. When the constraints are of the type < bi, then to convert the it into equality we need adding some variable (not constant) this is normally done by adding variables such as S1, S2. . . . . Sn, which are called slack variables. In physical sense, these slack variables represent unused resources, the slack variables contribute nothing towards the objective function and hence their coefficients in the objective function are to be zeros. Thus, to illustrate the above concept, Constraints ailxl + ai2x2. . . . .ainxn < bi ; i = 1,2,. .m(Canonical form) Can be written as ail x1….ain xn + Si = bi ; i = 1,2,. . m(Standard form) And the objective function can be written as Max. or Min. Z = c1 x1 + c2 x2 + . . . . . cn xn + 0S1 + 0S2 + ...... Similarly for the constraints of the type ≥, the addition of slack variables has to be in the form of subtraction. Thus, equation of constraints can be written as ai1 x1 + ai2 x2 +….ain xn - Si = b I ; i = 1,2,…….m To bring it to the standard form, we add another variable called artificial variable (Ai), as follows: ai1 x1 +ai2 x2 +. . . . . . . ainxn- Si + Ai = bi ; i = 1, 2, 3,. . . . . . . . m This is done to achieve unit matrix for the constraints. But artificial variables can not figure in the solution as there are artificially added variables and have no significance for the objective function. These variables, therefore, are to be removed from the solution. Standardization/Tableau Form/ Types of constraint Standard form ≤ Add a slack variable = Add an artificial variable ≥ subtract a surplus variable and add an artificial variable Steps in Simplex Method 1. Write the problem in standard form: Characteristics: All constraints are expressed in the form of equalities or equations. All right hand sides are non-negative All variables are non-negative 2. Develop an initial simplex tableau Steps in developing initial simplex tableau: i. List the variables in the model across the top of the tableau ii. Next fill-in the parameters of the model in the appropriate rows and columns iii. Add two columns to the left side of the tableau. The first column is a list of variables called ‘Basis’. iv. The Cj at the top second column indicates that the values in that column and the values in the top row are objective function coefficients. v. The last column at the right is called the quantity column. It refers to the right hand side values (RHS) of the constraints. vi. There are two more rows at the bottom of the tableau. The first raw is a Zj-row. For each column the Zj – value is obtained by multiplying each of the number of the column by their respective row coefficient in column C. The last row is Cj-Zj row. The values in this row are also calculated column by column. For each Column, the value in row Zj is subtracted form the Cj value in the top row. 3. Interpreting the initial simplex tableau 4. Determining the entering variable: For a maximization problem; the entering variable is identified as the one which has the largest positive value in Cj-Z row. The column which corresponds to the entering variable in the simplex tableau is called pivot column. In a minimization problem, the entering variable is identified as the one which has the largest negative Cj-Z row value in the simplex tableau. 5. Determining the leaving variable: the leaving variable is identified as the one with the smallest non-negativity ratio for quantity divided by respective positive pivot columnar entries. The row of the leaving variable is pivot row. 6. Make the entering variable basic and the leaving nonbasic by applying elementary row operations of matrix algebra. 7. Iteration for improved solution: (a) Replace outgoing variable with the entering variable and enter relevant coefficients in Zj column. (b) Compute the, Pivot row with reference to the newly entered variable by dividing the old row quantities by the key element. (c) new values for the other rows. In the revised simplex table, all the other rows are recalculated as follows. New row elements= Elements in the old row - [corresponding key column element multiplied by the corresponding new element of the revised row at (b) above.] (d) The same procedure is followed for modification to bi column also. (e) Having obtained the revised simplex table, evaluate ∆j = Cj -Zj and test for optimality as per step 3 above. 8. Check for optimality Remark: A simplex solution for a maximization problem is optimal if and only if cj – z row contains only zeros and negative value (i.e. if there are no positive values in the cj – z row). The simplex solution for a minimization problem is optimal if Cj-Z row contains only zero and positive values (Cj-Z ≥ 0). (a) Obtaining Optimal Solution- If the table indicates optimality level by examining ∆j or index row, the iteration stops at this point and values of bi's for corresponding variables in the product mix column will indicate the values of the variables contributing towards the objective function. The value of the objective function can be then worked out by substituting these Values for corresponding decision variables. (b) If the solution is not optimal, proceed to Step 9. 9. Revise or improve the Solution-For this purpose, we repeat Step 4 to Step 7 till optimality conditions are fulfilled and solution is obtained. Rule for Ties. Whenever two similar values are encountered in index row or ratio column, we select any column or ratio, but to reduce computation effort, following can be helpful. (a) For key column, select the left most tie element. (b) For ratio, select nearest to the top. The artificial variables in a minimization problem will be expressed in the objective function with a large positive coefficient so that they are quickly eliminated as we proceed with the solution. Note that: if the solution is not optimal the steps will be repeated again and again until the optimal solution is obtained! Simplex AlgorithmMaximization Problem The structure of most algorithms has the following steps as a main framework Formulate Problem as LP Put In Standard Form Put In Tableau Form Execute Simplex Method A. Simplex Algorithm-Maximization Problem Solve the following problem by simplex method. Max. Z = 8x 1 + 16x2 Subject to, x1 + x2 < 200 x2 < 125 3x1 + 6x2 < 900 x1, x2 > 0 Solution We convert the inequality into equations by adding slack variables. Above statements can thus be written as follows. x1 + x2 + S1 = 200 x2 + S2 = 125 3x1 + 6x2 + S3 = 900 and x1, x2, S1, S2, S3 > 0. where S1,S2, S3 are slack variables and objective function is re-written as: Max. Z = 8x1 + 16x2 + 0S1 + 0S2 + 0S3 Now there are five variables and three equations and hence to obtain the solution, any two variables will have to be assigned zero value. Moreover, to get a feasible solution, all the constraints must be satisfied. To start with, let us assign x1 = 0; x2 = 0 (Both decision variables are assigned zero values) Hence, S1 = 200, S2 = 125, S3 = 900 and Z=0 This can be written as initial simplex table 1 Unit Cj profit BV (Zj) Q 8 16 0 0 0 Ratio X1 X2 S1 S2 S3 Q/aij 0 S1 200 1 1 1 0 0 200/1=200 0 S2 125 0 1 0 1 0 125/1=125 0 S3 900 3 6 0 0 1 900/6= 150 Zj 0 0 0 0 0 0 8 16 0 0 0 Cj-Zj Key No. EV Where: EV= entering variable (Key column) LV= leaving variable (key row) LV Entering variable Since Cj - Zj is maximum at 16, i.e., profit is more for each unit for x2 variable, we introduce x2 into the solution. It is the marked as key column and x2 becomes the entering variable. Dividing Quantities (bi's) by the corresponding key elements of each row, we obtain the ratio (Q/aij) column such as for row S1, it is 200 ÷ 1 = 200, S2= 125 and S3=150. Leaving variable The leaving variable is, the row which has least ration (Q/aij), here, S2 has 125 ratio which small compare to other BV, it will be replaced by X2. Now each of the elements of the Key row is divided by Key element to get x2 row in the new table. Thus we get the key row as follows: Unit profit 16 Q 125/1 125 X1 x2 0/1 0 S1 1/1 1 S2 0/1 0 1/1 1 S3 0/1 0 In order to obtain the corresponding values of the table, we follow the relationship as follows: New row= old row – corresponding coefficient in pivot column X new tableau row value Q X1 X2 S1 S2 S3 For S1, row 200(1x125)=75 1-(1x0)=1 1-(1x1)=0 1-(1x0)=1 0-(1x1)=-1 0-(1x0)=0 For S3 row 900(6x125)=150 3-(6x0)=3 6-(6x1)=0 0-(6x0)=0 0-(6x1)=-6 1-(6x0)=0 Arranging these values into the simplex table, we obtain: revised simplex table II Zj Cj 8 16 0 0 0 variable Q x1 x2 s1 s2 S3 0 S1 75 1 0 1 -1 0 16 X2 125 0 1 0 1 0 0 S3 150 3 0 0 -6 1 Zj 2000 0 16 0 16 0 0 0 -16 0 Cj-Zj 8 Since the Cj-Zj row is contains positive value it is not optimal, as a result we have to revise the tableau to reach the optimal solution Lets identify the Ev and Lv Cj 8 16 0 0 0 Q/aij variable Q x1 x2 s1 s2 S3 0 S1 75 1 0 1 -1 0 75/1=75 16 X2 125 0 1 0 1 0 125/0=∞ 0 S3 150 3 0 0 -6 1 150/3=50 Zj 200 0 0 8 16 0 16 0 0 0 -16 0 Cj-Zj Key column Key row Zj Now each of the elements of the Key row is divided by Key element to get x2 row in the new table. Thus we get the key row as follows: Unit profit 8 Q X1 150/3 50 x2 3/3 1 S1 0/3 0 S2 0/3 0 -6/3 -2 S3 1/3 1/3 In order to obtain the corresponding values of the table, we follow the relationship as follows: New row= old row – corresponding coefficient in pivot column X new tableau row value Q X1 X2 S1 S2 S3 For S1, row 75-(1x50)=25 1-(1x1)=0 0-(1x0)=0 1-(1x0)=1 -1-(1x-2)=1 0-(1x1/3)= -1/3 For X2 row 125-(0x50)= 150 0-(0x0)=0 1-(0x1)=1 0-(0x0)=0 0-(0x-2)=0 0-(0x1/3) =0 Arranging these values into the simplex table, we obtain: revised simplex table III Zj Cj 8 16 0 0 0 variable Q x1 x2 s1 s2 S3 0 S1 25 0 0 1 1 -1/3 16 X2 125 0 1 0 1 0 8 X1 50 0 0 -2 1/3 Zj 2400 8 16 0 0 8/3 0 0 0 -8/3 Cj-Zj 1 0 Now, all the values of ∆j being zero or negative, suggesting that the solution is optimal and Z = 2,400 for x1 = 50 and x2 = 125. S1 indicates surplus. Home work Maximize Z = 30x1 + 40x2 Subject to, 60x1 + 120x2 < 12,000 8x1 + 5x2 < 600 3x1 + 4x2 < 500 x1, x2 > 0 Answer X = 200/11 X =1000/11 Profit= 46,000/11 1 2 Class work Maximise Z = 10x1 + 15 x2 + 20 x3 S.T. 10 x1+ 5x2 + 2x3 ≤ 2,700 5x1 + 10x2 + 4x3 ≤ 2,200 1x1 + 1x2 + 2x3 ≤ 500 and All 1x,x2 andx3 are ≥ 0 Maximise Z = 10x1 + 15 x2 + 20 x3 +0S1+0S2+0S3 S.T Solution ; S1 1600 10 x1+ 5x2 + 2x3 +S1 = 2700 5x1 + 10x2 + 4x3 + S2 = 2200 X2= 150 1x1 + 1x2 + 2x3+S3 = 500 X2= 174.4 x1, x2 and x3 all ≥ 0 Profit= 5738 Simplex AlgorithmMinimization problem B. Simplex Algorithm- Minimization problem 1. • Some of the important aspects of minimization problem Artificial variables have no economic significance Introduced only to bring in the standard form of simplex method. • Need be removed from the solution as soon as they become non-basic. 2. Since these variables are added for computation purpose only, ensure their zero value in the optional solution. This can be done by assigning very large penalty (+M) for a minimisation problem, so that these do not enter the solution. 3. If artificial variables cannot be removed from the solution, then the solution so obtained is said to be Non-Feasible. This would indicate that the resources of the system are not sufficient to meet the expected demand. 4. Equality Constraints also can be handled by using artificial variables to obtain initial solution. Big M-Method In this method, we assign the coefficients of the artificial variables, as a very large positive penalty i.e., +M therefore called Big M-method. The Big M-method for solving LP problem can be adopted as follows: Step 1 : The standard simplex table can be obtained by adding surplus and artificial variables. Surplus variables are assigned zero coefficients and artificial variables assigned +M coefficients in the objective function. Step 2: We obtain initial basic feasible solution by assigning zero value to the decision and surplus variables. Step 3: Initial basic feasible solution is obtained in the form of the simplex table as above and then values of ∆j = Cj - Zj are calculated. If ∆j ≥0, then the optimal solution has been obtained. If ∆j< 0, then we select the largest negative value of ∆j and this column becomes the key column indicating the entering variable. Step 4: Determine the key row as in case of maximisation problem i.e., selecting the lowest positive value of the ratio Q or bi/aij, obtained by dividing the value of quantity bi by corresponding element of the key column. Step 5: Repeat steps 3 and 4 to ensure optimal solution with no artificial variable in the solution. If at least one artificial variable is present in the basis with zero value and coefficient of M in each Cj - Zj values is negative, the LP problem has no solution. This basic solution will be treated as degenerate. A tie for the pivot row is broken arbitrarily and can lead to degeneracy. If at least one artificial variable is present in the basis with positive value, and coefficient of M in each Cj - Zj values is non-negative, then LP problem has no optimal basic feasible solution. It is called pseudo-optimum solution. Example Food A contains 20 units of vitamin X and 40 units of vitamin Y per gram. Food B contains 30 units each of vitamin X and Y. The daily minimum human requirements of vitamin X and Y are 900 and 1200units respectively. How many grams of each type of food should be consumed so as to minimise the cost, if food A costs 60 cents per gram and food B costs 80 cents per gram. Solution: LPP formulation is as follows Min. Z = 60x1+ 80x2 (Total Cost) Subject to, 20x1 + 30x2 > 900 40x1 + 30x2 > 1,200 and x1, x2 > 0 (Vitamin X Constraint) (Vitamin Y Constraint) Adding slack and artificial variables, we get Min. Z = 60x1 + 80x2 + 0S1 + 0S2 + MA1 + MA2 Subject to, 20x1 + 30x2 – S1 + A1 = 900 40x1 + 30x2 - S2 + A2 = 1,200 and x1, x2, S1, S2, A1, A2 > 0 Initial non-optimal solution is written as follows: simplex table I Zj Cj 60 80 0 0 M M Ratio BV Q x1 x2 s1 s2 A1 A2 M A1 900 20 30 -1 0 1 0 45 M A2 1200 40 30 0 -1 0 1 30 Zj 2100M 60M 60M -M -M M M M 0 0 Cj-Zj 60-60M 80-60M M Since ∆j = 60 – 60M is the lowest, x1 becomes the entering variable, similarly Ratio bi/aij = 30 is lowest positive value, hence it goes out. Simplex table II Zj Cj 60 80 0 0 M Ratio BV Q x1 x2 s1 s2 A1 Q/aij M A1 300 0 15 -1 1/2 1 20 60 X1 30 1 3/4 0 -1/40 0 40 Zj 1800+300M 60 Cj-Zj 0 45+15M -M -3/2+1/2M M 35-15M (3-M)/2 M 0 Since ∆j = 35 – 15M is the lowest, x1 becomes the entering variable, similarly Ratio bi/aij = 20 is lowest positive value, hence it goes out. Initial non-optimal solution is written as follows: Simplex table III Zj Cj 60 80 0 0 BV Q x1 x2 s1 s2 80 X2 20 0 1 -1/15 1/30 60 X1 15 1 0 1/20 -1/20 Zj 2500 60 80 -7/3 -1/3 0 0 7/3 1/3 Cj-Zj Since ∆j = zero and positive value, hence this the solution. Class work Mixed constraints Mixed constraints Initial non-optimal solution is written as follows: simplex table I Zj Cj 4 2 0 0 M M Ratio BV Q x1 x2 s1 s2 A1 A2 M A1 3 3 1 0 0 1 0 1 M 0 A2 S2 Zj 6 3 9M 4 1 7M 3 2 4M -1 0 -M 0 1 0 0 0 M 1 0 M 6/4 3 4-7M 2-4M M 0 0 0 Cj-Zj X1= Q X1 X2 S1 S2 3 3 1 0 0 3/3 3/3 1/3 0/3 0/3 new row= 1 1 1/3 0 0 New row= old row – corresponding coefficient in pivot column row A2, Q= 6-(4x1) = 2 X1= 4-(4x1)=0 X2= 3-(4x1/3)=5/3 S1= -1-(4x0)= -1 S2= 0-(4x0)= 0 A2= 1-(4x0)= 1 A2 0 0/3 0 new tableau X row value S2, Q= 3-(1x1)= 2 X1= 1-(1x1)=0 X2= 2-(1x1/3)=5/3 S1= 0-(1x0)=0 S2= 1-(1x0)=1 A2= 0-(1x0)=0 Revised simplex table II Zj Cj 4 2 0 0 M Ratio BV Q x1 x2 s1 s2 A2 4 X1 1 1 1/3 0 0 0 3 M 0 A2 S2 Zj 2 2 4+2M 0 0 4 5/3 -1 5/3 0 4/3+5/3M -M 0 1 0 1 0 M 6/5 6/5 0 2-5M/3 0 0 Cj-Zj M Select near to the top X2= Q X1 X2 S1 S2 2 0 5/3 -1 0 2/5/3 0/5/3 5/3/5/3 -1/5/3 0/5/3 new row= 6/5 0 1 -3/5 0 New row= old row – corresponding coefficient new tableau in pivot column X row value row X1, Q= 1-(1/3x6/5) = 9/15 X1= 1-(1/3x0)= 1 X2= 1/3-(1/3x1)= 0 S1= 0-(1/3x-3/5)= 1/5 S2= 0-(1/3x0)= 0 S2, Q= 2-(5/3x6/5)= 0 X1= 0-(5/3x0)= 0 X2= 5/3-(5/3x1)= 0 S1= 0-(5/3x-3/5)=1 S2= 1-(5/3x0)= 1 Optimal solution simplex table III Zj Cj 4 2 0 0 BV Q x1 x2 s1 s2 4 X1 9/15 1 0 1/5 0 2 0 X2 S2 Zj 6/5 0 72/15 0 0 4 1 0 2 -3/5 1 -2/5 0 0 0 0 0 0 0 Cj-Zj This is the optimal solution, with X1= 3/5 X2= 6/5 S2= 0, and total cost= 24/5 Irregular types of LPP The basic simplex solution of typical maximization and minimization problems has been shown in this chapter. However, there are several special types of atypical linear programming problems. For irregular problems the general simplex procedure does not always apply. These special types include problems with more than one optimal solution, infeasible problems, problems with unbounded solutions, problems with ties for the pivot column or ties for the pivot row, and problems with constraints with negative quantity values. Multiple optimal solution 40 A 10 20 30 Profit @ corner B and C is equal (1200) B FR C 10 20 30 40 50 An infeasible solution Multiple optimal solution The three constraints do not overlap to form a feasible solution area. Because no point satisfies all three constraints simultaneously, there is no solution to the problem. 8 X1= 4 C 6 X2=6 4 B 2 4X1+2X2=8 A C 2 4 6 8 10 An unbounded problem In some problems the feasible solution area formed by the model constraints is not closed. In these cases it is possible for the objective function to increase indefinitely without ever reaching a maximum value because it never reaches the boundary of the feasible solution area. In an unbounded problem the objective function can increase indefinitely because the solution space is not closed. An unbounded solution But unlimited profits are not possible in the real world; unbounded solution, like an infeasible solution, atypically reflects 10 The objective function is shown to increase without 2 6 the model 4 problem or in formulating 8 an error in defining the FR 2 4 6 8 10 bound; thus, the solution is never reached Sensitivity analysis (post optimality) Carried out after the optimal solution is found Is begins with the final simplex tableau Its purpose is to explore how change in any of the parameters of a problem i.e. coefficient of the constraints, coefficient of the objective function, quantity or RHS values Example: a change in the RHS of a constraints Change in RHS or Q of one constraint is considered at a time Consider shadow price Shadow price: is a marginal value; it indicates the impact that a one unit change in the value of the constraint would have on the value of the objective function. Shadow prices are the values in the Z-row of slack columns The LPM of the micro computer problem above is: Max Z: 60x1+50x2 Subject to: Assembly time: 4x1+10x2≤100 Inspection time: 2x1+x2≤22 Storage space: 3x1+3x2≤39 x1, x2≥0 Basis Cj 60 X1 S1 0 0 X1 60 1 X2 50 0 Z 60 50 X2 0 0 1 50 0 S1 1 0 0 0 0 S2 6 1 -1 10 0 S3 -16/3 -1/3 2/3 40/3 Cj-Z 0 0 10 -40/3 0 Quantit y 24 9 4 740 Shadow price From the above tableau; the shadow prices are $ 0 for S1, $10 for S2 and $40/3 for S3. This tells us that if the amount of assembly time was increased by one hour, there would be no effect on profit; if the inspection time was increased by one hour, the effect would be to increase profit by $ 10, and if storage space was increased by one cubic foot, profit would increase by $40/3. In the above problem, the optimal solution was given by the intersection of inspection time and storage space constraints. Thus, the shadow prices in the final simplex tableau will remain the same so long as the same constraints give the intersection for the optimal solution. For what range of changes in the RHS value of those constraints the current shadow prices remain valid? This is answered by range of feasibility. Range of Feasibility (Right hand side range) The range of feasibility is the range over which the RHS value of a constraint can be changed and still have the same shadow prices. Range of feasibility The range within which resources/constraints can changed having the proportionate change in objective value Steps Step 1. compute the ratio (feasibility ratio) quantity/respective slack value = Q/S both –ve and +ve ratio are considered Step 2. identify the smallest +ve ratio and –ve ratio closest to zero Step 3. find the upper limit or allowable increase and lower limit or allowable decrease (range of feasibility) Determine the range of feasibility for each of the constraints in the ff LPP, whose final tableau Zj 0 60 50 Bv S1 X1 X2 Zj Cj-Zj Cj Q 60 X1 50 X2 0 S1 0 S2 0 S3 24 0 0 1 6 -16/3 9 1 0 0 -1 -1/3 4 740 0 60 0 1 50 0 0 0 0 -1 10 -10 2/3 40/3 -40/3 Solution 1. Recall the original value of the resources Original value constraints S1 100 S1 1 22 S2 0 39 S3 0 2. ratio = Q/respective slack values S1= 24/1= 24 S2= 24/6= 4 9/0= undefined 9/-1= -9 4/0= undefined 4/-1= -4 S2 6 -1 -1 S3 -16/3 -1/3 2/3 S3= 24/-16/3= -4.5 9/-1/3= -27 4/2/3= 6 3. Find the range of feasibility Constrai nts S1 S2 S3 Origina l value 100 22 39 Lower limit Upper limit 100-24= 76 100+∞ 22-4= 18 22+4= 26 39-6 = 39+4.5= 43.5 33 Range of feasibility 76-∞ 18-26 33-43.5 Therefore: Constraint one (assembly line): 100-24 up to 100+∞= 76-∞ Constraint two (inspection time): 22-4 up to 22-4= 18-26 Constraint three (storage space): 39-6 up to 39+4.5= 33-43.5 Interpretation First constraint: Each hour decrease in assembly time will decrease the current profit by Birr 0 (i.e no effectindicated by shadow price) as long as the decrease is up to 24 hours. But if the assembly time decreases by more than 24 hours (or if the total available assembly time is lower than 76 hours), the current shadow price will no longer be valid. That is, the profit will be affected. But available assembly time can increase indefinitely (=allowable increase is ∞ ) without affecting the current profit level. Second constraint: Similarly, Each hour increase or decrease in inspection time will increase or decrease the current profit by $10, respectively as long as the total inspection time is between 18 and 26 hours. Out side the range of feasibility, the current shadow price ($10) will not be valid. Third constraint: Each cubic feet increase or decrease in storage space results in an increase or decrease, respectively, of profit by $13.33 (i.e 40/3) as long as the total storage space is between 33 and 43.5 cubic feet. Range of insignificance S1= below zero S2= (0-10) S3= (0-40/3) Duality The mirror image of LPP A given LPP has two forms 1. The Primal: the original LP Model 2. The Dual: alternative How to convert the primal to its dual and vice versa? Maximization objective objective of the Dual. of the primal= minimization The primal dual relationship 2 Example: The doctor advises a patient visited him that the patient is weak in his health due to shortage of two vitamins, i.e., vitamin X and vitamin Y. He advises him to take at least 40 units of vitamin X and 50 units of Vitamin Y everyday. He also advises that these vitamins are available in two tonics A and B. Each unit of tonic A consists of 2 units of vitamin X and 3 units of vitamin Y. Each unit of tonic B consists of 4 units of vitamin X and 2 units of vitamin Y. Tonic A and Bare available in the medical shop at a cost of ETB 3 per unit of A and ETB 2.50 per unit of B. The patient has to fulfill the need of vitamin by consuming A and B at a minimum cost. If we solve and get the solution of the primal problem, we can read the answer of dual problem from the primal solution. Primal problem: Min C= 3X1+ 2.5X2 st: 2X1+ 4X2 ≥40 Dual Problem: Max Z= 40Y1+ 50Y2 St: 2x+ 3y ≤3 3X1+ 2X2 ≥50 4x+ 2y ≤2.50 X1, X2≥0 Y1, Y2 ≥0. Solution to primal (minimization) CJ 3 2.5 0 0 M M Zj Bv Q X1 X2 S1 S2 A1 A2 2.5 X2 5/2 0 1 -3/8 1/4 3/8 -1/4 3 X1 15 1 0 1/4 -1/2 -1/4 1/2 Zj 51.25 3 2.5 -3/16 -7/8 3/16 7/8 0 0 3/16 7/8 M-3/16 M-7/8 Cj-Zj Answer: X1= 15 X2= 2.5 cost= 51.25 Solution to dual (maximization) CJ 40 50 0 0 Zj Bv Q Y1 Y2 S1 S2 50 Y2 7/8 0 1 1/2 -1/4 40 Y1 3/16 1 0 -1/4 3/8 Zj 51.25 3 2.5 15 5/8 0 0 -15 -5/2 Cj-Zj Answer: Y1= 3/16 Y2= 7/8 profit= 51.25 The patient has to minimize the cost by purchasing vitamin X and Y and the shopkeeper has to increase his returns by fixing competitive prices for vitamin X and Y. Minimum cost for patient is ETB 51.25 and the maximum returns for the shopkeeper is ETB 51.25. The competitive price for tonics is ETB 3 and ETB 2.50. Here we can understand the concept of shadow price or economic worth of Resources clearly. If we multiply the original elements on the right hand side of the constraints with the net evaluation elements under slack or surplus variables we get the values equal to the minimum cost of minimization problem or maximum profit of the maximization problem. post optimality analysis • Carried out after the optimal solution is found • Is begins with the final simplex • Sensitivity analysis tableau• Duality Sensitivity analysis Examination of the impacts of changes of parameters on the optimal solution. i.e. change of coefficient of the constraints, change of coefficient of the objective function, change of quantity or RHS values Starts with the final tableau of the LPP (simplex tableau) Example: 1. a change in the RHS of a constraints Change in RHS or Q of one constraint is considered at a time Consider shadow price Shadow price: is a marginal value; it indicates the impact that a one unit change in the value of the constraint would have on the value of the objective function. Shadow prices are the values in the Zj-row of slack columns The LPM of the micro computer problem above is: Max Z: 60x1+50x2 Subject to: Assembly time: 4X1+10x2≤100 Inspection time: 2x1+x2≤22 Storage space: 3x1+3x2≤39 x1, x2≥0 Basis Cj 60 X1 S1 0 0 X1 60 1 X2 50 0 Z 60 50 X2 0 0 1 50 0 S1 1 0 0 0 0 S2 6 1 -1 10 0 S3 -16/3 -1/3 2/3 40/3 Cj-Z 0 0 10 -40/3 0 Quantit y 24 9 4 740 Shadow price From the above tableau; the shadow prices are $ 0 for S1, $10 for S2 and $40/3 for S3. for example, an increase of S1 by one unit will resulted increment of objective value by $10. Similarly the opposite is true, i.e. decrease of 1 unit of S1 will be resulted in reduction of objective value by $10. But to what extent this change hold true? Because we can’t increase or decrease the constraint infinitely, there are upper and lower limits, i.e. allowable increase and decrease. Range of Feasibility (Right hand side range) The range of feasibility is the range over which the RHS value of a constraint can be changed and still have the same shadow prices. Range of feasibility The range within which resources/constraints can changed having the proportionate change in objective value Steps Step 1. compute the ratio (feasibility ratio) quantity respective slack value = Q/S both –ve and +ve ratio are considered Step 2. identify the smallest +ve ratio and –ve ratio closest to zero Step 3. find the upper limit or allowable increase and lower limit or allowable decrease (range of feasibility) Upper limit= the original value + negative ratio Lower limit= the original value – positive ratio For both max and Closest to zero Determine the range of feasibility for each of the constraints in the ff LPP, whose final tableau Zj 0 60 50 Bv S1 X1 X2 Zj Cj-Zj Cj Q 60 X1 50 X2 0 S1 0 S2 0 S3 24 0 0 1 6 -16/3 9 1 0 0 -1 -1/3 4 740 0 60 0 1 50 0 0 0 0 -1 10 -10 2/3 40/3 -40/3 Solution 1. Recall the original value of the resources Original value constraints S1 100 S1 1 22 S2 0 39 S3 0 2. ratio = Q/respective slack values S1= 24/1= 24 S2= 24/6= 4 9/0= undefined 9/-1= -9 4/0= undefined 4/-1= -4 S2 6 -1 -1 S3 -16/3 -1/3 2/3 S3= 24/-16/3= -4.5 9/-1/3= -27 4/2/3= 6 3. Find the range of feasibility Constrai Origina Lower limit Upper limit nts l value Range of feasibility S1 S2 100 22 100-24= 76 100+∞ 22-4= 18 22+4= 76-∞ 11-26 S3 39 39-6 = 33 33-43.5 26 39+4.5= 43.5 Therefore: Constraint one (assembly line): 100-24 up to 100+∞= 76-∞ Constraint two (inspection time): 22-4 up to 22-4= 18-26 Constraint three (storage space): 39-6 up to 39+4.5= 33-43.5 Interpretation First constraint: Each hour decrease in assembly time will decrease the current profit by Birr 0 (i.e no effectindicated by shadow price) as long as the decrease is up to 24 hours. But if the assembly time decreases by more than 24 hours (or if the total available assembly time is lower than 76 hours), the current shadow price will no longer be valid. That is, the profit will be affected. But available assembly time can increase indefinitely (=allowable increase is ∞ ) without affecting the current profit level. Second constraint: Similarly, Each hour increase or decrease in inspection time will increase or decrease the current profit by $10, respectively as long as the total inspection time is between 18 and 26 hours. Out side the range of feasibility, the current shadow price ($10) will not be valid. Third constraint: Each cubic feet increase or decrease in storage space results in an increase or decrease, respectively, of profit by $13.33 (i.e 40/3) as long as the total storage space is between 33 and 43.5 cubic feet. Example 2. A change of coefficient of objective function Two cases 1. Range of insignificance the range over which the non basic variables objective function coefficient can change without making these variables entering in the solution 2. Range of optimality the range over which the objective function coefficient of basic variables can change without changing the optimal values i.e. without changing basic and non basic variables but change the optimal function value. Steps for range of optimality For both max and min problems Example Cj Zj 0 60 50 BV S1 X1 X2 Zj Cj-Zj 60 Q 24 9 4 740 50 X1 0 1 0 60 X2 0 0 1 50 S1 1 0 0 0 S2 6 1 -1 10 S3 -16/3 -1/3 2/3 40/3 Solution 0 0 0 -10 -40/3 • S3= ? Determine the range of insignificant for S2 and the range of optimality for decision variables • Range of insignificant for S2 = -∞ up to 10 Solution X1 Cj-Zj 0 0 0 -10 -40/3 X1 values in the tableau 1 0 0 1 ∞ ∞ ∞ -10 -1/3 X2 Cj-Zj 0 0 0 -10 -40/3 0 1 0 -1 2/3 ∞ ∞ ∞ 10 -20 40 • Upper limit= 60+40= 100 • Lower limit= 60-10= 50 • Upper limit= 50+10= 60 • Range of optimality of X1(1st DV)= 50-100 • Lower limit= 50-20= 30 • Range of optimality of X2(2nd DV)= 30-60 Duality The mirror image of LPP A given LPP has two forms 1. The Primal: the original LP Model 2. The Dual: alternative How to convert the primal to its dual and vice versa? Maximization objective of the primal= minimization objective of the Dual. The primal dual relationship Example: The doctor advises a patient visited him that the patient is weak in his health due to shortage of two vitamins, i.e., vitamin X and vitamin Y. He advises him to take at least 40 units of vitamin X and 50 units of Vitamin Y everyday. He also advises that these vitamins are available in two tonics A and B. Each unit of tonic A consists of 2 units of vitamin X and 3 units of vitamin Y. Each unit of tonic B consists of 4 units of vitamin X and 2 units of vitamin Y. Tonic A and Bare available in the medical shop at a cost of ETB 3 per unit of A and ETB 2.50 per unit of B. The patient has to fulfill the need of vitamin by consuming A and B at a minimum cost. If we solve and get the solution of the primal problem, we can read the answer of dual problem from the primal solution. Primal problem: Min C= 3X1+ 2.5X2 st: 2X1+ 4X2 ≥40 Dual Problem: Max Z= 40Y1+ 50Y2 St: 2y1+ 3y2 ≤3 3X1+ 2X2 ≥50 4y2+ 2y2 ≤2.50 X1, X2≥0 Y1, Y2 ≥0. Solution to primal (minimization) CJ 3 2.5 0 0 M M Zj Bv Q X1 X2 S1 S2 A1 A2 2.5 X2 5/2 0 1 -3/8 1/4 3/8 -1/4 3 X1 15 1 0 1/4 -1/2 -1/4 1/2 Zj 51.25 3 2.5 -3/16 -7/8 3/16 7/8 0 0 3/16 7/8 M-3/16 M-7/8 Cj-Zj Answer: X1= 15 X2= 2.5 cost= 51.25 Solution to dual (maximization) CJ 40 50 0 0 Zj Bv Q Y1 Y2 S1 S2 50 Y2 7/8 0 1 1/2 -1/4 40 Y1 3/16 1 0 -1/4 3/8 Zj 51.25 3 2.5 15 5/8 0 0 -15 -5/2 Cj-Zj Answer: Y1= 3/16 Y2= 7/8 profit= 51.25 The patient has to minimize the cost by purchasing vitamin X and Y and the shopkeeper has to increase his returns by fixing competitive prices for vitamin X and Y. Minimum cost for patient is ETB 51.25 and the maximum returns for the shopkeeper is ETB 51.25. The competitive price for tonics is ETB 3 and ETB 2.50. Here we can understand the concept of shadow price or economic worth of Resources clearly. If we multiply the original elements on the right hand side of the constraints with the net evaluation elements under slack or surplus variables we get the values equal to the minimum cost of minimization problem or maximum profit of the maximization problem. The END!