Scribd
Upload
96 views3 pages

Linear Programming III: Duality and Sensitivity Analysis

This document provides an overview of linear programming duality and sensitivity analysis. It discusses: 1) How to derive the dual linear program from the primal by introducing dual variables and transposing matrices while ensuring all variables are non-negative and constraints are of the correct type. 2) How the optimal solutions of the primal and dual programs are identical with dual variables obtained from primal slack variables. 3) How sensitivity analysis examines the effects of changes to objective coefficients, resource availability, and constraint coefficients on the optimal solution using graphical and simplex methods. Shadow prices indicate the marginal value of scarce resources.

Uploaded by

tejashraj93
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
96 views3 pages

Linear Programming III: Duality and Sensitivity Analysis

This document provides an overview of linear programming duality and sensitivity analysis. It discusses: 1) How to derive the dual linear program from the primal by introducing dual variables and transposing matrices while ensuring all variables are non-negative and constraints are of the correct type. 2) How the optimal solutions of the primal and dual programs are identical with dual variables obtained from primal slack variables. 3) How sensitivity analysis examines the effects of changes to objective coefficients, resource availability, and constraint coefficients on the optimal solution using graphical and simplex methods. Shadow prices indicate the marginal value of scarce resources.

Uploaded by

tejashraj93
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

4

Linear Programming III:


Duality and Sensitivity Analysis
________________________________________________________________________

Having discussed about the formulation of linear programming problems and their
solution using graphic and simplex methods, we now consider the subjects of duality and
sensitivity analysis. Both of these have significant managerial implications. Their most
important contribution is a thorough appreciation of economics of the problem. Shadow
prices tell the manager what price would be reasonable for acquiring scarce resources, or
which resources are in abundance and could possibly be diverted to other uses within the
organisation. Sensitivity analysis lets the manager know if more effort should be given to
estimating certain parameters of the problem, and when changes in those estimated
parameters might produce a change in the basic nature of the decision.
Duality

For every LPP, there is another LPP that is related to it and is derived from it. The given
problem is called the primal while the problem derived is called the dual. To write the
dual of an LPP, the following conditions are to be satisfied:
1. The variables should all be non-negative.
2. The constraints should all be type, if the problem is of the maximisation type,
and type if it is of the minimisation type.
If condition (1) is not satisfied so that some variable is given to be unrestricted in
sign, then it is substituted by the difference of two non-negative variables. Thus, if x3 is
given to be unrestricted, it may be replaced in the problem by x4 x5, where x4, x5 0.
In respect of condition (2), if a constraint involves sign while the desired sign is ,
or vice versa, then it is multiplied by 1. If a constraint involves an = sign, then it is
replaced by a pair of inequalities in opposite directions. A constraint 7x1 + 9x2 = 66, for
instance would be replaced by two constraints as 7x1 + 9x2 66 and 7x1 + 9x2 66. One
of these would be reversed again as desired.
Once the two conditions are satisfied, the dual can be written by introducing dual
variables and transposing the three matrices involved in the primal. The dual will have as
many constraints as the number of primal variables, and will have as many variables as
the number of constraints that the primal has. The rows and columns of the aij coefficients will be interchanged. It may be noted that for every unrestricted variable in the
primal problem, a constraint in the dual is of the = form and vice-versa.
Solution to Primal and Dual
If the primal has an optimal solution, then the dual also
would have optimal solution with identical objective function value. Not only this, the
optimal values of the dual variables can be obtained from the j values of the

slack/surplus variables in the optimal tableau. If the primal problem has unbounded
solution, the dual will have infeasibility, and vice versa.
Economic Interpretation of the Dual The j values for the slack/surplus variables
indicate the marginal profitability or shadow prices of the resources which they represent.
The shadow price of a resource is the maximum price a manager would be prepared to
pay for an additional unit of the resource. Note that only scarce resources have positive
shadow prices. The surplus resources have no worth since any addition to them will cause
no change in the profit.
Sensitivity Analysis
This deals with investigating the effect of changes in the objective function co-efficients
(cjs), resource availability (bi values) and the LHS co-efficients (aijs) of the constraints.
The changes in the two-variable problems can be analysed graphically as well as in
context of the simplex solution. However, for problems with three variables or more,
graphic approach is ruled out.
In the graphic approach, to determine whether a change in the co-efficient of a variable
in the objective function would lead to a change in the basis or not, it should be noted that
a change in a co-efficient causes a change in the iso-profit line (for a maximisation
problem). The current basis continues to be optimal as long as the current optimal
solution happens to be the last point in the feasible region to make contact with the isoprofit lines while we move in the direction of increasing values of the objective function.
If the current basis remains optimal, the values of the decision variables remain what they
have been, but the optimal Z-value may change.
To determine whether a change in the right hand side value of a constraint would cause
a change in the optimality of the current basis, we begin by considering the constraints
that are binding for the current optimal solution. For a change in the RHS of a constraint,
the current basis remains optimal as long as the point where the constraints are binding
remains feasible. However, even if the current basis remains optimal, the values of the
decision variables and the optimal value of the objective function might change.
In context of the simplex method solution, sensitivity analysis is performed as follows:
(a) Changes in cj values: For a variable included in the basis, divide the j row values
by the corresponding row values in tableau for that variable. The range of cj for which the
current optimal solution would not change is obtained by adjusting the value by the least
positive and negative quotients. For a variable not included in the basis, any decrease in
the co-efficient (profit) value or any increase by an amount less than the j (absolute)
value shall cause no change in the solution.
(b) Changes in bi values (Right hand side ranging:) Divide the bi values in the optimal
solution tableau by the aij values of the slack variable of interest. The least positive and
the least negative values are then subtracted from the bi value of the original problem. It
gives the range of values for the given resource over which there shall be no change in its

shadow price. It means that any changes in the availability of resources within the limits
established would increase/decrease profit at a rate of the shadow price.
(c) Changes in aij values: They can also be analysed to determine their effect on the
optimal solution. However, the analysis is relatively more complex.
The advisability of introduction of a new product in the existing set-up can also be
determined. For this, we first obtain resource requirements and multiply them with the
respective shadow prices. If the unit profit from the proposed product exceeds the sum of
these multiplications, then it is desirable to add the new product.
Multiple Parameter Changes: 100% Rule When multiple changes take place in the coefficients in the objective function or in the RHS values of the constraints, then the 100
per cent Rule may be used to determine whether they would affect the current solution.

You might also like