Q2. What is the procedure for generating the valid inequality?
If the algorithm terminates without closing the duality gap because it is no longer possible to generate a valid inequality, a branch and bound approach can be used to carry on the solution procedure.
Q3. What is the advantage of the heuristic?
The advantage of the heuristic is that itcan be used in a reduction algorithm that attempts to reduce the problem size by fixing variables to one or to zero.
Q4. What is the second procedure of arnsten and Varbrand?
The second procedure of arnsten and Varbrand (1987) generates bounds through the solution of a surrogate relaxation of the knapsack constraints (2) and valid inequalities of the knapsack type (17).
Q5. How many jobs are split between agents?
In the practical problems (with number of jobs >> number of agents) that were solved the heuristic led to solutions within 0.1% of the lower bound and consequently within 0.1% of the optimal solution.
Q6. What is the strength of the relaxation given by Wilcox (1989)?
The strengthened relaxation is given by :(GR) min 1 22 (ci - wj + 8) xij (2), (4) } (16)15Their branching scheme is a combination of depth-first and breadth-first branching.
Q7. What is the difference between the two methods?
TheLagrangean procedure of Mrnsten and Varbrand discussed in the previoussection appears to be more efficient (stronger bound at root node, less valid inequalities added) than the surrogate one.