Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd
Upload
64 views

Linear Programming II

The simplex method is an algorithm for solving linear programming problems by moving from one feasible solution to another, eventually arriving at the optimal solution. It involves setting up a simplex tableau with constraints expressed as equations using slack and surplus variables. The method proceeds by obtaining an initial solution, testing for optimality, and improving any non-optimal solutions by selecting incoming and outgoing variables to generate a new tableau, until an optimal solution is reached.

Uploaded by

KAMAL PATI
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
64 views

Linear Programming II

The simplex method is an algorithm for solving linear programming problems by moving from one feasible solution to another, eventually arriving at the optimal solution. It involves setting up a simplex tableau with constraints expressed as equations using slack and surplus variables. The method proceeds by obtaining an initial solution, testing for optimality, and improving any non-optimal solutions by selecting incoming and outgoing variables to generate a new tableau, until an optimal solution is reached.

Uploaded by

KAMAL PATI
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

3

Linear Programming II:


Simplex Method
________________________________________________________________________

The simplex method is a step-by-step procedure for finding the optimal solution to a
linear programming problem. To apply simplex method to solving a linear programming
problem, it has to be ensured that
(i) all the variables in the problem are non-negative, and
(ii) all the values on the RHS of constraints are non-negative.

For a constraint involving a negative bi value, it is multiplied on each side by –1, and
the direction of inequality is reversed. Thus a constraint 2x1 – 7x2 ≥ –66 would be
replaced as – 2x1 + 7x2 ≤ 66. In case a variable is unrestricted in sign, it is replaced by the
difference of two non-negative variables. Thus, if x3 is given to be unrestricted in sign, it
may be replaced in the problem by x4 – x5, where x4, x5 ≥ 0.
To apply simplex method, the constraints are first expressed as equations. The slack
variables convert “≤” type of inequalities into equations while surplus variables are used
to convert “≥” type of inequalities into equations. Once all equations are obtained,
writing variable names on the top row and the aij values below it draws up a simplex
tableau. The bi values are written on the RHS of the tableau and the cj values in a bottom
row. The three steps involved in a simplex algorithm are:

1. Obtain an initial solution.


2. Test if the solution is optimal. If it is optimal, then stop. Otherwise, proceed to
step 3.
3. Obtain an improved solution. Repeat step 2.

Obtaining Initial Solution To obtain a solution in a simplex tableau, it is necessary to


look for an identity matrix within the matrix of aij values. If such a matrix cannot be
located by any combination of variables, then artificial variables (each with a large co-
efficient equal to M in the objective function of minimisation type and equal to –M in the
objective function of maximisation type) are added. In any event, the variables
constituting the identity matrix are then entered in the basis in the same order in which
they yield the matrix. Along with them, their co-efficients in the objective function are
entered. Finally, the solution is written in a row below the cj values. Each of the variables
in the basis is assigned the corresponding value in the bi column while the remaining
variables are all set equal to zero.

Testing the Optimality Once an initial solution is obtained with the aid of identity
matrix, it is tested for optimality in terms of Δj = cj - zj. The zj value for each variable is
the summation of the products of the values under the variable and the corresponding
values in the basis. For a maximisation problem, if all Δj ≤ 0, the solution is optimal and
for a minimisation problem, the condition is all Δj ≥ 0.

Improving a Non-optimal Solution A non-optimal solution is improved by


(i) selecting the variable with largest Δj (for a maximisation problem) and with most
negative Δj (for a minimisation problem) as the incoming variable (called the key
column);
(ii) selecting the variable with least non-negative replacement ratio as the outgoing
variable (known as the key row); and
(iii) obtaining a new simplex tableau. For obtaining new tableau, first derive the row
corresponding to the key row by dividing each element by the key element
(which lies at the intersection of key column and key row). For every other
element of the matrix, use the rule:
Corresponding element × Corresponding element
in key row in key column
New element = Old element – ___________________________________________
Key element

The revised solution obtained from the new tableau is also tested for optimality. Similar
improvements in the solutions are done until an optimal solution is reached.

Two-phase Method Where artificial variables are involved, the two-phase method can
also be used. In Phase-I, all artificial variables are eliminated while in the other phase the
optimal solution is obtained.

Some points are notable:


(i) Basic variables in any solution have a Δj value equal to zero. In an optimal
solution, if some non-basic variable(s) has Δj = 0, the problem has multiple
optimal solutions, while if none of the non-basic variables have Δj = 0, the
solution is unique optimal solution.
(ii) Infeasibility is present when a problem has no feasible solution. In solving an
LPP by simplex method, if a final solution is reached in terms of the Δj values
and an artificial variable is present in the basis, then infeasibility is indicated.
(iii) In improving a non-optimal solution, the outgoing variable is chosen as the one
with the least non-negative replacement ratio. If all such ratios are negative, or a
mix of negative and infinity type, then the problem has unbounded solution.
(iv) A solution is said to be degenerate if some basic variable(s) has a solution value
equal to zero. A degenerate solution may be optimal or non-optimal. When a
degenerate solution is non-optimal and the outgoing variable happens to be a
degenerate variable, there is no change in the objective function value for the
solution obtained in the next tableau.

You might also like