CMPUT 675 : Topics on Approximation Algorithms and Approximability Fall 2007 Lecture 8 : Oct 2

M. Salavatipour,Zhuang Guo
Abstract:The above LP-formulation has an exponential number of constraints. However, we can still solve this LP using the Ellipsoid method. Given a (possible) solution vector ~x (assignments to xe, e ∈ E), we can check if it is feasible in polynomial time (this implies that the separation oracle for this LP is in P). To do so we interpret variable xe as a distance label on edge e, for each e ∈ E; then we compute the lengths of shortest paths between each source-sink pair (si, ti) w.r.t the current distance labels. If all the lengths are ≥ 1, then all the paths between each pair (si, ti) must have lengths ≥ 1, and therefore, we can conclude that all constraints are satisfied and the solution is feasible. If the shortest path is < 1 then we obtain a violated constraint.
What problem does this paper attempt to address?