Parameterized Resiliency Problems. / Crampton, Jason; Gutin, Gregory; Koutecky, Martin; Watrigant, Remi.

In: Theoretical Computer Science, 07.08.2019, p. 1-14.

Research output: Contribution to journalArticle

E-pub ahead of print

Documents

  • Accepted Manuscript

    Accepted author manuscript, 569 KB, PDF document

    Embargo ends: 7/08/20

    Licence: CC BY-NC-ND Show licence

Abstract

We introduce an extension of decision problems called \textit{resiliency problems}. In a resiliency problem, the goal is to decide whether an instance remains positive after any (appropriately defined) perturbation has been applied to it. To tackle these kinds of problems, some of which might be of practical interest, we introduce a notion of resiliency for Integer Linear Programs (ILP) and show how to use a result of Eisenbrand and Shmonin (Math. Oper. Res., 2008) on Parametric Linear Programming to prove that \textsc{ILP Resiliency} is fixed-parameter tractable (FPT) under a certain parameterization.

To demonstrate the utility of our result, we consider natural resiliency variants of several concrete problems, and prove that they are FPT under natural parameterizations. Our first results concern a four-variate problem which generalizes the {\sc Disjoint Set Cover} problem and which is of interest in access control. We obtain a complete parameterized complexity classification for every possible combination of the parameters. Then, we introduce and study a resiliency variant of the {\sc Closest String} problem, for which we extend an FPT result of Gramm et al. (Algorithmica, 2003). We also consider problems in the fields of scheduling and social choice. We believe that many other problems can be tackled by our framework.
Original languageEnglish
Pages (from-to)1-14
Number of pages14
JournalTheoretical Computer Science
Early online date7 Aug 2019
DOIs
Publication statusE-pub ahead of print - 7 Aug 2019
This open access research output is licenced under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License.

ID: 34353085