The Budgeted Transportation Problem

S. Kapoor,M. Sarwat

DOI: https://doi.org/10.48550/arXiv.1302.5871

2013-02-24

Data Structures and Algorithms

Abstract:Consider a transportation problem with sets of sources and sinks. There are profits and prices on the edges. The goal is to maximize the profit while meeting the following constraints; the total flow going out of a source must not exceed its capacity and the total price of the incoming flow on a sink must not exceed its budget. This problem is closely related to the generalized flow problem. We propose an auction based primal dual approximation algorithm to solve the problem. The complexity is $O(\epsilon^{-1}(n^2+ n\log{m})m\log U)$ where $n$ is the number of sources, $m$ is the number of sinks, $U$ is the ratio of the maximum profit/price to the minimum profit/price. We also show how to generalize the scheme to solve a more general version of the problem, where there are edge capacities and/or the profit function is concave and piece-wise linear. The complexity of the algorithm depends on the number of linear segments, termed ${\cal L}$, of the profit function.