A projection method for \(l_ p\) norm location-allocation problems (Q1337136)

The authors present a method for solving \(l_p\)-norm location- allocation problems involving bivalent allocations. The \((0, 1)\) constraints on the allocations are relaxed and the proposed method relies upon necessary conditions for local minima of the relaxed problem. The algorithm finds a local minimum by using active set methods and projections onto successive subspaces in which the current activities remain active. It is shown that all local minima of the relaxed problem can be achieved at points with bivalent allocations. The implementation of the algorithm exploits the special structure of the problems for reducing the computational cost. Finally, numerical results are given for a set of thirty-six test problems.