Linear Programming: Duality and Sensitivity Analysis

Linear programming (LP) is a mathematical method used to determine the best possible outcome (such
as maximum profit or minimum cost) in a mathematical model whose requirements are represented by
linear relationships. Two important aspects of linear programming are duality and sensitivity analysis.

5.1. Dual Formulation and Interpretation

Duality in linear programming refers to a relationship between every linear programming problem
(called the primal problem) and another linear programming problem (called the dual problem).

1. Dual Formulation: The dual problem is derived from the primal problem by associating a variable with
each constraint. If the primal problem is a minimization problem, then the dual problem will be a
maximization problem, and vice versa. Here’s how you can formulate the dual from the primal:

 Primal Problem (Standard Form):

Minimize: cTxsubject to:Ax≥b,x≥0\text{Minimize: } c^T x \quad \text{subject to:} \quad Ax \geq b, \quad
x \geq 0

where cc is a vector of coefficients for the objective function, AA is the matrix of coefficients for the
constraints, and bb is the right-hand side vector of the constraints.

 Dual Problem: The dual problem is formed by introducing a new variable for each constraint of
the primal problem. If the primal problem is a minimization problem with constraints of the
form Ax≥bAx \geq b, then the dual is a maximization problem with constraints of the form
ATy≤cA^T y \leq c, where yy is the vector of dual variables (also called shadow prices).

Dual Formulation:

Maximize: bTysubject to:ATy≤c,y≥0\text{Maximize: } b^T y \quad \text{subject to:} \quad A^T y \leq c, \
quad y \geq 0

Here:

o yy represents the dual variables associated with the primal constraints.

o bTyb^T y is the objective function of the dual.

o ATy≤cA^T y \leq c represents the constraints of the dual.

2. Interpretation of Dual Variables:

 In economic terms, the dual variables yy represent the shadow prices or marginal values of the
constraints in the primal problem. A dual variable provides information on how much the
objective of the primal problem would change if the right-hand side of the corresponding
constraint were increased by one unit.
  For example, if the dual variable associated with a resource constraint is large, it means the
resource is highly valuable (scarce), and increasing its availability would improve the objective of
the primal problem significantly.

3. Duality Theorems:

 Weak Duality Theorem: The objective value of the dual problem is always a bound on the
objective value of the primal problem. Specifically, if the primal is a minimization problem, then
the objective value of the dual is always greater than or equal to the objective value of the
primal.

 Strong Duality Theorem: If both the primal and dual problems have feasible solutions, then
both problems have the same optimal objective value.

 Complementary Slackness: If the primal constraint is satisfied with equality, the corresponding
dual variable is positive. Similarly, if the dual constraint is satisfied with equality, the
corresponding primal variable is positive.

5.2. Primal-Dual Relationship

The primal-dual relationship is the fundamental connection between the primal and dual problems in
linear programming. This relationship is crucial for understanding how changes in the primal problem
affect the dual problem and vice versa.

1. Complementary Slackness Conditions:

o Let x∗x^* be the optimal solution to the primal and y∗y^* be the optimal solution to
the dual.

o For each ii, if xi∗>0x_i^* > 0, then the corresponding dual constraint is tight (i.e., the
dual inequality becomes an equality).

o For each jj, if yj∗>0y_j^* > 0, then the corresponding primal constraint is tight (i.e., the
primal inequality becomes an equality).

2. Interpretation of Optimal Solutions:

o At optimality, the primal and dual problems have the same objective value, and the
complementary slackness conditions describe how the primal and dual solutions are
related. In other words, if a primal constraint is not active (it has a positive slack), then
the corresponding dual variable is zero, and vice versa.

3. Weak and Strong Duality Theorems:

o Weak Duality implies that the optimal value of the dual objective provides a bound on
the optimal value of the primal objective.

o Strong Duality asserts that if both the primal and dual problems are feasible, they have
the same optimal value.
Simplex Method and Duality

The Simplex method is a popular algorithm used to solve linear programming problems. It works by
moving along the edges of the feasible region from one vertex (basic solution) to another in search of
the optimal solution.

1. Simplex Algorithm and Duality:

o The Simplex method can be applied to both the primal and dual problems.

o The dual Simplex method is a variant of the Simplex algorithm used for solving the dual
problem, and it proceeds in a similar way to the primal Simplex method.

o In practice, the Simplex method solves the primal, and the dual variables are updated as
part of the algorithm. The complementary slackness conditions help ensure the dual
feasibility is maintained.

2. Dual Simplex Method:

o The dual Simplex method is a version of the Simplex method used when the primal
solution is feasible but the dual is not. It works by iterating to restore dual feasibility
while maintaining primal feasibility.

o This method is particularly useful when solving LP problems where changes in the
constraints cause an initial infeasible dual solution.

Summary of Key Concepts:

 Duality establishes a relationship between the primal and dual linear programs, with each
problem having a corresponding counterpart.

 Primal Problem: Minimization with inequality constraints.

 Dual Problem: Maximization with corresponding dual variables and inequality constraints.

 Weak Duality: The optimal value of the dual provides a lower bound to the primal (in
minimization).

 Strong Duality: The optimal values of the primal and dual are equal when both are feasible.

 Complementary Slackness: Ties the primal and dual solutions together by indicating that if a
primal constraint is non-binding, its corresponding dual variable must be zero, and vice versa.

 Simplex Method: Can be used to solve both the primal and dual problems. The dual Simplex
method handles situations where the primal is feasible, but the dual is not.

Sensitivity Analysis
Sensitivity analysis in linear programming involves studying how the optimal solution changes when
there are variations in the coefficients of the objective function or the right-hand side of the constraints.
Key areas include:

 Sensitivity to Objective Coefficients: How changes in the coefficients of the objective function
affect the optimal solution.

 Sensitivity to Right-Hand Side: How changes in the right-hand side vector (the constraints)
influence the solution.

 Range of Optimality: The range of values for an objective coefficient over which the current
optimal solution remains optimal.

 Range of Feasibility: The range of values for a right-hand side parameter for which the current
feasible region remains feasible.

In duality, sensitivity analysis can also reveal how the dual variables (shadow prices) change when the
constraint coefficients or right-hand side values change.

These notes provide a concise overview of duality and sensitivity analysis in linear programming,
including dual formulation, interpretation, the primal-dual relationship, and the connection to the
Simplex method.
Sure! Here are study notes for Special Types of Linear Programming (LP) Models:

6. Special Types of LP Models

Linear programming is a powerful tool for optimization, and there are several specialized models
designed to solve specific types of problems. Below are three important types:

6.1. Transportation Model

Purpose: The transportation model is used to determine the most cost-effective way to transport goods
from several suppliers to several consumers while satisfying demand and supply constraints.

Key Elements:

 Suppliers: Where the goods are produced (each has a supply capacity).

 Consumers: Where the goods are needed (each has a demand).

 Transportation cost: Cost of transporting one unit of goods from a supplier to a consumer.

General Structure:

 Let xijx_{ij} be the quantity transported from supplier ii to consumer jj.

 Objective: Minimize transportation cost, typically of the form: Minimize Z=∑i=1m∑j=1ncijxij\

text{Minimize} \, Z = \sum_{i=1}^{m} \sum_{j=1}^{n} c_{ij} x_{ij} where cijc_{ij} is the cost per
unit of transportation from supplier ii to consumer jj, and xijx_{ij} is the amount transported.

Constraints:

1. Supply constraints: Total supply from each supplier cannot exceed its available supply.

∑j=1nxij=Si∀i\sum_{j=1}^{n} x_{ij} = S_i \quad \forall i

where SiS_i is the supply at supplier ii.

2. Demand constraints: Total demand from each consumer must be met.

∑i=1mxij=Dj∀j\sum_{i=1}^{m} x_{ij} = D_j \quad \forall j

where DjD_j is the demand at consumer jj.

3. Non-negativity: xij≥0x_{ij} \geq 0.

6.2. Transshipment Model

Purpose: The transshipment model is an extension of the transportation model, where goods are not
only transported from suppliers to consumers, but can also be transferred between intermediate points
(called transshipment nodes).

Key Elements:

 Suppliers, consumers, and transshipment nodes: These are the various locations involved in the
movement of goods.

 Transshipment cost: The cost of transferring goods between intermediate nodes.

General Structure:

 Let xijx_{ij} be the amount of goods transported from location ii to location jj.

 The objective is to minimize the total transportation cost: Minimize Z=∑i=1m∑j=1ncijxij\

text{Minimize} \, Z = \sum_{i=1}^{m} \sum_{j=1}^{n} c_{ij} x_{ij} where cijc_{ij} is the cost of
transporting goods from location ii to location jj.

Constraints:

1. Supply constraints: As in the transportation model, the supply from each origin location must
not exceed its available supply.

2. Demand constraints: Each destination or transshipment node must meet its demand or outflow.

3. Flow conservation at transshipment nodes: For each intermediate node, the total inflow must
equal the total outflow.

Applications:

 Supply chains with intermediate processing or distribution points.

 Transportation of goods through distribution centers.

6.3. Assignment Model

Purpose: The assignment model deals with assigning resources (such as workers or machines) to tasks in
such a way that the total cost (or time) is minimized or profit is maximized.

Key Elements:

 Resources (e.g., workers): These are the entities performing tasks.

 Tasks (e.g., jobs or assignments): These are the tasks that need to be assigned to resources.

 Cost or time: The cost (or time) to assign a particular resource to a particular task.

General Structure:
  Let xij=1x_{ij} = 1 if resource ii is assigned to task jj, and 00 otherwise.

 The objective is to minimize the total cost: Minimize Z=∑i=1m∑j=1ncijxij\text{Minimize} \, Z = \

sum_{i=1}^{m} \sum_{j=1}^{n} c_{ij} x_{ij} where cijc_{ij} is the cost of assigning resource ii to
task jj.

Constraints:

1. Assignment constraints: Each resource is assigned to exactly one task, and each task is assigned
to exactly one resource.

∑j=1nxij=1∀i(Each resource assigned to one task)\sum_{j=1}^{n} x_{ij} = 1 \quad \forall i \quad \

text{(Each resource assigned to one task)} ∑i=1mxij=1∀j(Each task assigned to one resource)\
sum_{i=1}^{m} x_{ij} = 1 \quad \forall j \quad \text{(Each task assigned to one resource)}

2. Non-negativity: xij∈{0,1}x_{ij} \in \{0, 1\}, where xij=1x_{ij} = 1 means the assignment is made,
and xij=0x_{ij} = 0 means no assignment.

Applications:

 Assigning workers to tasks (e.g., scheduling, personnel assignments).

 Assigning machines to jobs in a factory.

 Vehicle routing or transportation task assignment.

Summary of Differences:

Model Objective Typical Application Key Features

Transportation Minimize Shipping goods from Direct supply to demand

Model transportation cost suppliers to consumers transportation

Flow can go through

Transshipment Minimize total More complex logistics with
intermediate transshipment
Model transportation cost intermediate nodes
nodes

Assignment Minimize cost of Assigning resources to tasks Each resource is assigned to

Model assignments (workers, machines) exactly one task

Key Takeaways for Study:

1. Transportation Model focuses on finding the least-cost way to transport goods between
suppliers and consumers.

2. Transshipment Model is an extension where goods can be transferred through intermediate

nodes.
 3. Assignment Model optimizes the assignment of resources to tasks, minimizing costs or
maximizing efficiency.

Understanding the unique characteristics of each model will help you apply the right model to real-
world problems, like supply chain logistics or job scheduling.