Eastman Algorithm

(Concepts of Hungarian Method for solving Assignment Problem and Branch &
Bound Methods are required).

Step-1. Let CLUB represent the current least upper bound on the optimal solution
of the Traveling Salesman Problem.
Step-2. Solve the problem as if it is an assignment problem, where the distances
D(I,J) are the elements of the effectiveness matrix. The solution provides
the lower bound on the optimal solution, go to step-3, otherwise the
optimal solution of the assignment problem is also an optimal solution of
the traveling salesman problem, so stop.
Step-3. Select a sub-tour and let k be the number of arcs (links) in the selected sub
tour. Eastman selects the sub tour with the smallest number of arcs. All
other sub tours at this node can be ignored.
Step-4. Branch into k sub problems. If the sub tour is I 1I2IkI1 then for
the sub problem 1 let D(I1,I2)= , for sub problem 2 let D(I2,I3)= , etc,
and for sub problem k D(Ik, I1)= .
Step-5. Solve the k new assignment problems. Each solution distance is a
lower
bound for the corresponding sub problem.

Step-6. If there are one or more feasible solutions from step-5 and if the
smallest total distance for these feasible solutions, say STD, is smaller
than CLUB, set CLUB= STD and save the corresponding feasible
solution. Otherwise CLUB remains unchanged.

Step-7. If CLUB is less than the lower bounds on all other unexplored sub
problems, then the solution corresponding to CLUB is an optimal
solution of the Traveling Salesman Problem, so stop; otherwise, go to
step-8. By unexplored sub problems, we mean sub problems that have
not been divided into further sub problems.

Step-8. From the set of all unexplored non feasible (sub tours present) sub
problems with a bound less than CLUB, select the sub problem with the
smallest lower bound for further branching. Go To Step-3.