Eva Tardos
Associate Professor
eva@cs.cornell.edu
http://www.cs.cornell.edu/home/eva/eva.html
Ph.D. Eotvos University, Hungary, 1984


My current research interest focuses on combinatorial optimization problems relevant for the design, maintenance, and management of high-speed communication networks. These networks, the most important communication platform of the future, present new technological challenges. Effective development of future large-scale networks will be impossible without the development of new theoretically sound algorithmic approaches. My research is concerned with developing fast approximation algorithms that can serve as tools for successfully answering these challenges. This year, I worked on two basic problems: finding disjoint paths in graphs (virtual circuit routing) and packet switching.

In a sequence of two papers joint with Jon Kleinberg, we give an O(1) approximation for the maximum disjoint paths problem in densely embedded, nearly-Eulerian graphs, a class of graphs that includes the two-dimensional mesh and many other planar and locally planar interconnection networks. For networks that are not explicitly required to be "high capacity", this is the first constant-factor approximation for the problem in any class of graphs other than trees. We also consider the problem in the on-line setting, relevant to applications in which connection requests arrive over time and must be processed immediately. Here we obtain an asymptotically optimal O(log n)-competitive on-line algorithm for the same class of graphs.

In a paper joint with Yuval Rabani (visiting scientist at Cornell), we developed a new distributed algorithm for the packet switching problem. The packet routing problem is simply that of moving packets in a communication network from their sources to their desired destinations as quickly, as reliably, using as few resources (such as queues) as possible. A solution to this problem consists of two distinct (though by no means independent) parts: a selection of paths for the packets and a schedule of the motion of packets along their selected paths. Earlier, Leighton, Maggs, and Rao considered the second part, the packet scheduling problem when a single packet has to traverse each path. Our results improve the previous results for problems with relatively high dilation (when packets have to travel along long paths in the network). We also extend the techniques to handle the case of infinite streams of regularly scheduled packets along every path.

I also continued my work on algorithms for fundamental problems of combinatorial optimization. The traveling salesman problem was one of the very first problems where the cutting-plano algorithms and the branch-and-cut algorithms were developed and tested. Much of the research has focused on finding new classes of facets for the TSP polytope, and much less attention has been paid to the algorithm to separate among these classes of facets. In joint work with Lisa Fleischer, we consider the problem of finding violated comb inequalities for the traveling salesman problem.


Awards

University Activities

Professional Activities

Lectures

Publications


Return to:

1995-1996 Annual Report Home Page
Departmental Home Page

If you have questions or comments please contact: www@cs.cornell.edu.


Last modified: 2 November 1996 by Denise Moore (denise@cs.cornell.edu).