A Strongly Polynomial Algorithm for Line Search in Submodular Polyhedra.pdfVIP

A Strongly Polynomial Algorithm for Line Search in Submodular Polyhedra Kiyohito NAGANO Department of Mathematical Informatics, Graduate School of Information Science and Technology, University of Tokyo, Bunkyo-ku, Tokyo 113-8656, Japan. kiyohito nagano@mist.i.u-tokyo.ac.jp Revised May 2, 2005 Abstract A submodular polyhedron is a polyhedron associated with a submodular function. This paper presents a strongly polynomial time algorithm for line search in submodular polyhedra with the aid of a fully combinatorial algorithm for submodular function minimization. The algorithm is based on the parametric search method proposed by Megiddo. 1 Introduction Let U be a finite nonempty set. A function ρ defined on 2U is submodular if ρ(X ) + ρ(Y ) ≥ ρ(X ∪ Y ) + ρ(X ∩ Y ), ∀X, Y ⊆ U. (1) Iwata, Fleischer and Fujishige [13] and Schrijver [21] independently presented combinatorial, strongly polynomial time algorithms for submodular function minimization. Iwata [11] pre- sented a fully combinatorial strongly polynomial time algorithm, which uses only additions, subtractions, comparisons, and the oracle calls for function values. For a vector x ∈ RU and u ∈ U , we denote by x(u) the component of x on u. For a submodular function ρ : 2U → R with ρ(∅) = 0, the submodular polyhedron P(ρ) and the base polyhedron B(ρ) are defined by P(ρ) = { x ∈ RU |x(X ) ≤ ρ(X ) (∀X ⊆ U ) } , (2) B(ρ) = { x ∈ RU |x ∈ P(ρ), x(U ) = ρ(U )} , (3) where x(X ) = u∈X x