Resource Allocation with Tree Constraints

We extend Zipkin's variant of the Luss and Gupta algorithm for the allocation of a single resource to activities described by concave return functions, to handle additional constraints imposed upon the total allocations to certain subsets of activities, and to subsets of these subsets, and so on. We determine computational times for certain return functions and extensions to problems with an objective function defined by the difference or the ratio of return and cost, and to problems with several resources.