Linear programming (LP), also known as linear optimization, is a mathematical

modeling technique used to achieve the best outcome in problems where the objective
and constraints are linear in nature. It's a problem-solving approach for making
decisions in business contexts. LP models are mathematical representations of these
problems.
Components of LP Models
The main components of an LP model include:
 Objective Function: This is the function to be maximized (e.g., profit) or
minimized (e.g., cost).
 Decision Variables: These are the unknown quantities that need to be solved for.
 Constraints: These represent limitations or restrictions, such as the availability of
resources.
 Parameters: These are fixed values that indicate the impact of each decision
variable on the objective function.
Characteristics of LP Models
LP models have several key characteristics:
 The relationship between variables and constraints must be linear.
 There must be an objective function.
 The model must include structural constraints.
 A non-negativity constraint must be present, meaning decision variables cannot
take negative values.
Assumptions of LP Models
The validity of an LP model depends on certain assumptions:
 Linearity: The relationships between variables and the objective
function/constraints are linear.
 Divisibility: Decision variables can take on fractional values (though for some
quantities, integer values might be required).
 Certainty: Parameters are known and constant, and all relevant constraints are
included in the model.
 Non-negativity: Negative values for variables are unrealistic and thus disallowed.
Formulating Linear Programming Models
The process of formulating an LP model involves several steps:
1. Define the Problem: Clearly state the specific objectives, for example, to
determine the number of units of products to produce to maximize profit given
certain restrictions.
2. Determine the Decision Variables: Represent the unknown quantities with letters.
For instance, x1 and x2 could represent the number of units of product 1 and
product 2, respectively.
3. Formulate the Objective Function:
o Ensure all decision variables are included in the objective function.
o The units of measurement for all coefficients in the objective function must
be consistent (e.g., if using kg for x1, use kg for x2).
o All terms in the objective function must include a variable.
 4. Formulate the Constraints: Identify the limitations and express them as
mathematical relationships.
5. Determine the Non-negativity Assumptions: All decision variables must be set as
greater than or equal to zero.
General Mathematical Forms
 Maximization:

MaxZ=C1X1+C2X2+...+CnXn

Subject to:

A11x1+A12x2+...+A1nxn≤R1

A21x1+A22x2+...+A2nxn≤R2

...

Ai1x1+Ai2x2+...+Ainxn≤Rn

Xij≥0 for J=1,2,3,4,...,n

 Minimization:

MinZ=C1X1+C2X2+...+CnXn

Subject to:

A11x1+A12x2+...+A1nxn≥R1

A21x1+A22x2+...+A2nxn≥R2

...

Ai1x1+Ai2x2+...+Ainxn≥Rn

Xij≥0 for J=1,2,3,4,...,n

Application Areas of LP
LP is applied in various fields, including:
 Production Management: Product mix, blending problems, production planning,
assembly line balancing.
 Marketing Management: Media selection, revenue determination, physical
distribution.
 Personnel Management: Staffing problems, determination of equitable salaries.
 Financial Management: Portfolio selection, profit planning.
  Agricultural Applications.
 Military Applications.
Solving LP Problems (LPP)
There are two primary methods for solving LP problems:
1. Graphical Method
2. Simplex Method
Graphical Method
The graphical method is suitable for problems with two decision variables. The steps
are:
1. Formulate the mathematical model of the problem.
2. Plot the constraints on a graph.
3. Identify the feasible region (the area satisfying all constraints) and determine its
corner points.
4. Evaluate the objective function at each corner point. For maximization, the
largest value is the optimum; for minimization, the smallest value is the optimum.
If two corners yield the same optimal value, then any point on the line segment
connecting them is also optimal.
For minimization problems, the feasible solution area is typically away from the origin,
and the optimum solution is the smallest possible value.
Graphical Solutions for Special Cases of LP Problems
 Unbounded Problems: Occur when decision variables can increase indefinitely
without violating constraints, often due to incorrect problem formulation (e.g.,
maximizing instead of minimizing).
 Problems with No Feasible Solutions (Infeasibility): Arise when no solution
satisfies all constraints, often due to conflicting constraints (a mix of ≥ and ≤
inequalities).
 Redundancy in Constraints: A constraint is redundant if its removal doesn't
change the feasible solution space. This doesn't pose difficulty in graphical
solutions.
 Problems with Multiple Optimal Solutions: It's possible for more than one extreme
point to yield the same optimal objective function value.
Simplex Method
The Simplex method is used for LP problems with two or more decision variables,
where the graphical method is not applicable. It's an iterative process that starts with a
basic feasible solution and moves towards the optimal solution by testing each feasible
solution for optimality.
Comparison with Graphical Method:
 Graphical method is for two decision variables; Simplex handles any number.
 Graphical method assumes inequalities as equations; Simplex converts
inequalities to equations by adding slack variables (maximization) or surplus
variables (minimization).
Steps for Maximization Problems using Simplex:
1. Formulate the linear programming model.
 2. Express the model in standard form by adding slack variables to constraints and
assigning them zero coefficients in the objective function.
3. Design the initial feasible solution by setting decision variables to zero.
4. Set up the initial Simplex Tableau.
5. Test for optimality. If all entries in the Cj−Zj row are negative or zero, the solution
is optimal. Otherwise, improve the solution.
6. Iterate towards an optimum solution by replacing a basic variable (leaving
variable) with a non-basic variable (entering variable).
7. Evaluate the new solution by constructing the next Simplex tableau.
8. Repeat steps 6-7 until an optimal solution is obtained (all Cj−Zj values are zero or
negative).
Steps for Minimization Problems using Simplex:
1. Formulate the LP model in standard form by introducing surplus and artificial
variables. Assign a coefficient of 0 to surplus variables and a large positive
number (+M) to artificial variables in the objective function. Artificial variables are
temporary and should not appear in the final solution.
2. Set up an initial solution and complete the initial Simplex tableau with Zj and Cj
−Zj rows.
3. Test for optimality. If all entries in the Cj−Zj row are positive or zero, the solution
is optimal. If any Cj−Zj value is negative, further improvement is needed.
4. Determine the entering variable (largest negative value in Cj−Zj row) and the
departing variable (smallest non-negative ratio).
5. Update the new solution by evaluating entries for the next Simplex tableau.
6. Repeat steps 3-5 until the optimal solution is obtained (all Cj−Zj values are zero
or positive).
Post-Optimality Analysis
Post-optimality analysis involves examining the implications of changes in problem
parameters after an optimal solution has been found.
1. Duality in LPP:
o Every LP problem has a "primal" form (original formulation) and a "dual"
form (mirror image).
o The solution to the primal problem implicitly contains the solution to the
dual, and vice versa.
o Dual LPPs are crucial for economic analysis, helping managers evaluate
the impact of new products or determine the marginal value of resources.
2. Sensitivity Analysis:
o This analysis explores how changes in objective function coefficients or
right-hand side values of constraints affect the optimal solution.
o Since parameters are rarely constant in real life, sensitivity analysis helps
determine how robust a solution is to potential changes.
o It provides decision-makers with greater confidence in using the solution
or warns them if more precise parameter information is needed.