Abstract
We introduce a variant of the well-known minimum-cost circulation problem in directed networks, where vertex demand values are taken from the integers modulo \(\lambda \), for some integer \(\lambda \ge 2\). More formally, given a directed network \(G = (V,E)\), each of whose edges is associated with a weight and each of whose vertices is associated with a demand taken over the integers modulo \(\lambda \), the objective is to compute a flow of minimum weight that satisfies all the vertex demands modulo \(\lambda \). This problem is motivated by a problem of computing a periodic schedule for traffic lights in an urban transportation network that minimizes the total delay time of vehicles. We show that this modular circulation problem is solvable in polynomial time when \(\lambda = 2\) and that the problem is NP-hard when \(\lambda = 3\). We also present a polynomial time algorithm that achieves a \(4(\lambda - 1)\)-approximation.
Similar content being viewed by others
Notes
We are using the term “space” abstractly. To relate space and time, we would need to introduce a maximum speed limit, call it \(\sigma \). Assuming vehicles move at their maximum speed, the total length of a vehicle platoon that can pass through any intersection in time \(\lambda \) is \(\sigma \lambda \). To move these vehicles to the next intersection in time \(t_{ij}\) at full speed, the road segment between these intersections is of length \(\sigma t_{ij}\). Since \(t_{ij} \ge \lambda \), we have \(\sigma t_{ij} \ge \sigma \lambda \), and, irrespective of the choice of \(\sigma \), it follows that there is sufficient space to park these vehicles, whatever their actual sizes may be.
References
Ahuja, R.K., Magnanti, T.L., Orlin, J.B.: Network Flows: Theory, Algorithms, and Applications. Prentice-Hall, Upper Saddle River (1993)
Boeing, G.: Planarity and street network representation in urban form analysis (2018). arXiv:1806.01805
Cormen, T.H., Stein, C., Rivest, R.L., Leiserson, C.E.: Introduction to Algorithms, vol. 2, 2nd edn. McGraw-Hill Higher Education, Boston (2001)
Dasler, P., Mount, D. M.: On the complexity of an unregulated traffic crossing. In: Proceedings of 14th International Symposium Algorithms Data Structure, volume 9214 of Lecture Notes Computer Science, pp. 224–235. Springer (2015)
Dresner, K.M., Stone, P.: A multiagent approach to autonomous intersection management. J. Artif. Int. Res. 31, 591–656 (2008)
Edmonds, J., Johnson, E.L.: Matching, Euler tours and the Chinese postman. Math. Program. 5, 88–124 (1973)
Edmonds, J., Karp, R.M.: Theoretical improvements in algorithmic efficiency for network flow problems. J. ACM 19, 248–264 (1972)
Eppstein, D., Gupta, S.: Crossing patterns in nonplanar road networks (2017). arXiv:1709.06113
Garey, M.R., Johnson, D.S.: Computers and Intractability: A Guide to the Theory of NP-Completeness. W. H. Freeman & Co., New York (1979)
Goldberg, A.V., Tarjan, R.E.: Finding minimum-cost circulations by canceling negative cycles. J. ACM 36, 873–886 (1989)
Guler, S.I., Menendez, M., Meier, L.: Using connected vehicle technology to improve the efficiency of intersections. Transp. Res. Part C Emerg. Technol. 46, 121–131 (2014)
Kleinberg, J., Tardos, E.: Algorithm Design. Addison-Wesley, Boston (2005)
Lichtenstein, D.: Planar formulae and their uses. SIAM J. Comput. 11, 329–343 (1982)
Orlin, J.B.: A polynomial time primal network simplex algorithm for minimum cost flows. Math. Program. 78, 109–129 (1997)
Tachet, R., Santi, P., Sobolevsky, S., Reyes-Castro, L.I., Frazzoli, E., Helbing, D., Ratti, C.: Revisiting street intersections using slot-based systems. PLoS One 11(3), e0149607 (2016)
Vial, J. J. B., Devanny, W. E., Eppstein, D., Goodrich, M. T.: Scheduling autonomous vehicle platoons through an unregulated intersection. In: Goerigk, M., & Werneck, R. (eds.) 16th Workshop on Algorithmic Approaches for Transportation Modelling, Optimization, and Systems (ATMOS 2016), vol. 54 of OpenAccess Series in Informatics (OASIcs), pp. 1–14. Schloss DagstuhlLeibniz-Zentrum fuer Informatik (2016)
Yu, J., LaValle, S. M.: Multi-agent path planning and network flow. In: Frazzoli, E., Lozano-Perez, T., Roy, N., & Rus, D. (eds.) Algorithmic Foundations of Robotics X: Proceedings of the Tenth Workshop on the Algorithmic Foundations of Robotics, pp. 157–173, Springer, Berlin, Heidelberg (2013)
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Research supported by NSF Grant CCF-1618866.
Appendix A: NP-hardness Generalization
Appendix A: NP-hardness Generalization
Fundamental blocks are constructed in the same manner as described in the \(\lambda = 3\) example in Sect. 5.1, but for completeness, we describe them here in their general form. A fundamental block begins with \(\lambda \) 1-vertices, connected as a star graph (i.e., a central vertex connected to each of the remaining vertices) with edges directed outward. Each 1-vertex is also connected to a \((\lambda -1)\)-vertex via an incoming edge. Finally, these \((\lambda -1)\)-vertices each have an outgoing interface edge (Fig. 8).
The expansion module, too, remains largely unchanged. It consists of a central fundamental block with \(\lambda \) interface edges. A fundamental block is then connected to all but one of these edges, with the remaining edge reserved for connecting to the existing variable gadget. For each expansion module added to a variable gadget, \((\lambda -1)^2-1\) interface edges are added (one interface edge is consumed when connecting to the variable gadget, hence the \(-1\)). See Fig. 9. Just as before, we must adjust the weights of a small subset of edges so that True and False assignments incur the same cost. To do so, arbitrarily select a fundamental block that is not the hub (i.e., not the fundamental block which connects to the existing variable gadget) and set the outgoing edges from its central 1-vertex to a weight of \(\frac{2\lambda -3}{\lambda -1}\).
Recall that the variable gadget described in Sect. 5.1 creates \(c(v_i)\lambda \) interface edges. In truth, the fundamental block creates \(\lambda \) interface edges while each of the \(c(v_i)-1\) expansion modules add \((\lambda -1)^2-1\) interface edges. When \(\lambda = 3\), these formulae are equivalent, each variable has \(c(v_i)\lambda \) interface edges, and each of the \(c(v_i)\) clause gadgets will connect to \(\lambda \) interface edges.
When \(\lambda > 3\), the variable gadgets will have \(\lambda + \left[ c(v_i)-1\right] \left[ (\lambda -1)^2-1\right] \) interface edges. As this is not evenly divisible by \(c(v_i)\), connecting to the clause gadgets becomes problematic. To correct for this, a modified expansion module, called a seed module, is used to start each variable gadget rather than a fundamental block. Rather than connecting to an existing variable gadget, the edge that was previously reserved for this connection is instead attached to one of the existing interface edges (recall that in actuality a \((\lambda -1)\)-vertex is being shared, not that two edges are being connected directly). By doing so, we now have a seed module that has \((\lambda -1)^2-1\) interface edges (Fig. 10) and, when used in conjunction with unaltered expansion modules, creates a variable gadget with \(c(v_i)\left[ (\lambda -1)^2-1\right] \) interface edges.
Clause gadgets are constructed as described in Sect. 5.2, with the caveat that they require \((\lambda -1)^2-1\) vertices rather than \(\lambda \). Each vertex will still only have three incoming edges, as each clause will only ever have three literals, and a demand of 1.
Despite a small increase in the size of each gadget, the reduction can still be done in polynomial time.
Finally, the cost threshold must be generalized in Lemma 5.4. In the general form, if F is satisfiable there will be a satisfying flow with
Rights and permissions
About this article
Cite this article
Dasler, P., Mount, D.M. Modular Circulation and Applications to Traffic Management. Algorithmica 81, 4098–4117 (2019). https://doi.org/10.1007/s00453-018-0491-9
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00453-018-0491-9