Assignment Problems
Assignment problem is a special case of general linear programming problem,
just as the transportation problem. For example a departmental head may have
four machines to be assigned to different mechanics (which differ in efficiency
while working on different machines) available for assignment. He is willing to
know that which machine should be assigned to which mechanic so that total
work can be completed in minimum time and minimum cost.
Solution methods of an Assignment Problem
ENUMERATIVE SIMPLEX METHOD TRANSPORTATION METHOD HUNGARIAN METHOD
METHOD
Most preferred and popular
method.
Structure
The Hungarian Method(Minimization)
Maximizing Case
Unbalanced Assignment Problems
Prohibited Routes(A Problem with Impossible Assignments)
1 Page
The Hungarian Method (Minimization)
Step 1. Develop the cost table for the given problem.
Step 2. If the no. of rows is not equal to no. of columns or vice-versa, dummy
rows or dummy columns must be added. The assignment cost of dummy rows
or columns is always zero.
Step 3. Select smallest element from each row and subtract it from each
element of corresponding row.
Repeat step 3 for each column.
Step 4. Optimality test: - to check the optimality of given cost matrix draw
minimum number of straight lines to cover all the zeros.
Step 5.
No. of covering lines = No. of rows (or columns)
Optimum assignment can be obtained.
If number of straight lines drawn is exactly equal to the number of rows(or
columns) in the matrix, an optimal assignment can be obtained.
If the no of covering lines is less than no of rows (or columns) optimal
assignment cannot be made. In such cases, least element out of uncovered
elements is to be subtracted from the uncovered elements and added to the
elements which are at intersection. And all the entries covered by one straight
line remain unchanged.
Step [Link] step 4 and step 5 until no of covering lines become equal to no
of rows or no of columns.
Step7. If No. of covering lines = No. of rows (or columns) assignment can be
made by following procedure-
Examine those rows which have exactly single zero and make assignment( )
around it and cross all other zeros of corresponding column.
Examine row having a single unmarked zero, make assignment ( ) around it
and cross all other zeros of corresponding column.
2
Page
After rows repeat the same steps for columns.
Maximizing Case
To transform maximizing problem into minimizing problem, we subtract all the
entries of the original matrix from the maximum entry of that matrix.
Maximization problem Minimization Problem
(Pay-off Matrix) (Loss Matrix)
Maximum element – each element
Unbalanced Assignment Problem
Balanced No of persons (no of rows) = No of jobs (no of columns)
Assignment
Problem
Unbalanced
Assignment
No of persons ≠No of job
problem
3
Page
For solving the problem, unbalanced assignment problem should be converted
into balanced one with the help of dummy jobs or dummy persons.
Convert
With the help of ‘dummy rows’ or
‘dummy columns’.
Unbalanced Balanced
Further procedure would be the same.
Prohibited Routes ( a problem with impossible assignments)-
In some cases a certain worker cannot be assigned a particular job. The
reasons for impossible assignments are numerous. Lack of required skills,
deficiency in technical know-how improper training and physical inability are
only a few among many reasons. For solving such type of assignment problems
we put infinite cost (∞) in the cell where no assignment is possible. The
remaining procedure is exactly the same as the ordinary assignment problem.
4 Page
Page 5