Autor(en): Fichtenberger, Hendrik
Titel: Property testing of graphs and the role of neighborhood distributions
Sprache (ISO): en
Zusammenfassung: Property testing considers decision problems in the regime of sublinear complexity. Most classical decision problems require at least linear time complexity in order to read the whole input. Hence, decision problems are relaxed by introducing a gap between “yes” and “no” instances: A property tester for a property Π (e. g., planarity) is a randomized algorithm with constant error probability that accepts objects that have Π (planar graphs) and that rejects objects that have linear edit distance to any object from Π (graphs with a linear number of crossing edges in every planar embedding). For property testers, locality is a natural and crucial concept because they cannot obtain a global view of their input. In this thesis, we investigate property testing in graphs and how testers leverage the information contained in the neighborhoods of randomly sampled vertices: We provide some structural insights regarding properties with constant testing complexity in graphs with bounded (maximum vertex) degree and a connection between testers with constant complexity for general graphs and testers with logarithmic space complexity for random-order streams. We also present testers for some minor-freeness properties and a tester for conductance in the distributed CONGEST model.
Schlagwörter: Property testing
Sublinear algorithms
Graph algorithms
Schlagwörter (RSWK): Graphentheorie
URI: http://hdl.handle.net/2003/39198
http://dx.doi.org/10.17877/DE290R-21116
Erscheinungsdatum: 2020
Enthalten in den Sammlungen:LS 02 Komplexitätstheorie und Effiziente Algorithmen

Dateien zu dieser Ressource:
Datei Beschreibung GrößeFormat 
fichtenberger-phd_thesis.pdfDNB1.23 MBAdobe PDFÖffnen/Anzeigen


Diese Ressource ist urheberrechtlich geschützt.



Diese Ressource ist urheberrechtlich geschützt. rightsstatements.org