A Personal Assignment Problem Solved By The Branch-And-Bound Strategy

Problem

    A linearly ordered set of persons P = {P1, P2, ... , Pn}, where P1 < P2 < ... < Pn.

    A set of jobs J = {J1, J2, ... , Jn}, partially ordered.

    Cost Cij is equal to the cost of assigning Pi to Jj. 

    Each Person is assigned to a job, and no two persons are assigned to the same job.

    We require that if f(Pi) <= f(Pj), then Pi <= Pj.

The function is a feasible assignment which maps persons to their appropriate jobs.

Our Problem is to find an optimal feasible assignment which minimizes the total cost of assignments

[NP-hard]

J must be a topologically sorted  sequence with respect to the partial ordering of jobs.

(each sequence represents a feasible assignment.)

[tree searching techniques can be easily used to find all the topologically sorted sequences.]

(1) Take an element which is not preceded by any other element in the partial ordering.

(2) Select this element as an element in a topologically sorted sequence.

(3) Remove this element just selected from the partial ordering set. The resulting set is still partially ordered

[apply the branch-and-bound strategy]

Given a cost matrix, 

->a reduced cost matrix

we can compute a lower bound of our solutions immediately.

Note: if a constant is subtracted from any row or any column of the cost matrix, an optimal solution does not change.

The resulting matrix is a reduced cost matrix in which every row and column contains at least one zero and the remaining entries of the matrix are all non-negative.

--Why did we subtract costs from the cost matrix?

--A higher lower bound will of course lead to an early termination!