Algorithm Engineering: Selected Results and Surveys

428 Pages·2016·10.734 MB·English

by  Lasse Kliemann, Peter Sanders (eds.)

t Lasse Kliemann
r
A
- Peter Sanders (Eds.)
e
h
t
-
f
o
y
-e
e
v
t
ar
u
t
SS
0
2
2
9

S
C Algorithm Engineering
N
L
Selected Results and Surveys
123
Lecture Notes in Computer Science 9220
Commenced Publication in 1973
Founding and Former Series Editors:
Gerhard Goos, Juris Hartmanis, and Jan van Leeuwen
Editorial Board
David Hutchison
Lancaster University, Lancaster, UK
Takeo Kanade
Carnegie Mellon University, Pittsburgh, PA, USA
Josef Kittler
University of Surrey, Guildford, UK
Jon M. Kleinberg
Cornell University, Ithaca, NY, USA
Friedemann Mattern
ETH Zurich, Zurich, Switzerland
John C. Mitchell
Stanford University, Stanford, CA, USA
Moni Naor
Weizmann Institute of Science, Rehovot, Israel
C. Pandu Rangan
Indian Institute of Technology, Madras, India
Bernhard Steffen
TU Dortmund University, Dortmund, Germany
Demetri Terzopoulos
University of California, Los Angeles, CA, USA
Doug Tygar
University of California, Berkeley, CA, USA
Gerhard Weikum
Max Planck Institute for Informatics, Saarbrücken, Germany
More information about this series at http://www.springer.com/series/7407
Lasse Kliemann Peter Sanders (Eds.)
(cid:129)
Algorithm Engineering
Selected Results and Surveys
123
Editors
LasseKliemann PeterSanders
Kiel University Karlsruhe Institute of Technology
Kiel Karlsruhe
Germany Germany
ISSN 0302-9743 ISSN 1611-3349 (electronic)
Lecture Notesin Computer Science
ISBN 978-3-319-49486-9 ISBN978-3-319-49487-6 (eBook)
DOI 10.1007/978-3-319-49487-6
LibraryofCongressControlNumber:2016957380
LNCSSublibrary:SL1–TheoreticalComputerScienceandGeneralIssues
©SpringerInternationalPublishingAG2016
Thisworkissubjecttocopyright.AllrightsarereservedbythePublisher,whetherthewholeorpartofthe
material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information
storageandretrieval,electronicadaptation,computersoftware,orbysimilarordissimilarmethodologynow
knownorhereafterdeveloped.
Theuseofgeneraldescriptivenames,registerednames,trademarks,servicemarks,etc.inthispublication
doesnotimply,evenintheabsenceofaspecificstatement,thatsuchnamesareexemptfromtherelevant
protectivelawsandregulationsandthereforefreeforgeneraluse.
Thepublisher,theauthorsandtheeditorsaresafetoassumethattheadviceandinformationinthisbookare
believedtobetrueandaccurateatthedateofpublication.Neitherthepublishernortheauthorsortheeditors
give a warranty, express or implied, with respect to the material contained herein or for any errors or
omissionsthatmayhavebeenmade.
Printedonacid-freepaper
ThisSpringerimprintispublishedbySpringerNature
TheregisteredcompanyisSpringerInternationalPublishingAG
Theregisteredcompanyaddressis:Gewerbestrasse11,6330Cham,Switzerland
Preface
AlgorithmEngineeringisamethodologyforalgorithmicresearchthatcombinestheory
withimplementationandexperimentationinordertoobtainbetteralgorithmswithhigh
practicalimpact.Traditionally,thestudyofalgorithmswasdominatedbymathematical
(worst-case) analysis. In Algorithm Engineering, algorithms are also implemented and
experimentsconductedinasystematicway,sometimesresemblingtheexperimentation
processes known from fields such as biology, chemistry, or physics. This helps in
counteracting an otherwise growing gap between theory and practice. The possible
benefits are manifold:
– We can identify practically good algorithms, even where theoretical analysis is
lackingordoesnotmakeanaccurateenoughstatement.Thiscanleadtosuccessful
algorithms being recognized and put into practice even when they look mathe-
matically inaccessible. It can also lead to theoretically appealing algorithms being
recognized as less useful in practice. For example, suppose that we have one
algorithmwithaquadraticworst-caserunningtimeandanotherwithacubicbound,
both proven by theory. Which of the two is faster in practice? This cannot be
answered by merely looking at the degrees of the polynomials.
– Experimentation can be supportive of design and analysis. It can help to identify
bottlenecks that could be eliminated by a re-design. It can help to build and to test
falsifiable hypotheses, e.g., whether a particular algorithm after a modification that
makes it faster will still deliver the same solution quality. This creates a feedback
loop where each particular sub-methodology benefits from the others.
– Observations of good performance in practice of an algorithm for which we only
know unsatisfactory theoretical guarantees have sparked new types of theoretical
algorithm analyses. In particular, smoothed analysis instead of the traditional
worst-case analysis has been shown to be successful. In smoothed analysis, the
input is first chosen as in worst-case analysis (i.e., we consider all possible inputs)
but then small random perturbations are performed. We then conduct an
average-case analysis using the resulting distribution on the set of instances.
– Therequirementtoimplementandtoexperimentallyevaluatealgorithmsmotivates
algorithm designers to think about practicability and to consider more realistic
computational models. New possibilities and their restrictions given by modern
hardware are moved into focus, such as memory hierarchies and parallelism.
– Testing algorithms on real-world input stimulates the exchange and cooperation
between thealgorithmicscommunityandother fieldswherealgorithmsareusedas
tools.
This volume reviews important algorithmic developments and results that were
made possible or were supported by Algorithm Engineering. Chapters are ordered
alphabetically by first author. This work emerged from the Priority Programme
“Algorithm Engineering” (1307) funded by the German Research Foundation (DFG),
VI Preface
which started in 2007 and lasted six years. In total, 28 projects received funding
through this program. In addition there were six associated projects. We gratefully
acknowledge this support.
Eachsubmissionforthisvolumewaspeer-reviewed.Wesincerelythanktheauthors
and the reviewers for their work, diligence, and cooperation.
In total, we have 12 chapters, including extensive surveys and case studies:
Chapter 1 A simple but powerful stochastic local search algorithm for the SAT
problem is presented and analyzed. Experiments are used for tuning and for
comparisonwithotheralgorithms.Itisconcludedthatwhenflippingavariable,
itismoreimportanttopayattentiontothenumberofnewlyunsatisfiedclauses
than to the number of newly satisfied ones.
Chapter2isasurveyonpracticalalgorithmsforroutingintransportationnetworks,
including road networks, schedule-based public transport networks, as well as
multimodal scenarios. Experiments show that it is possible to find good jour-
neys within milliseconds in large-scale networks. Several of the described
approaches have been included in mainstream production systems.
Chapter 3 surveys different ways to theoretically analyze the k-means clustering
algorithm. Several of the theoretical activities, e.g., smoothed analysis, were
motivated by observations in experiments.
Chapter 4 surveys practical algorithms for balanced graph partitioning. A large
variety of different approaches are presented, and implementation aspects and
benchmarking are discussed.
Chapter 5 In randomized and derandomized rounding, for many applications it is
required that the solution satisfies certain constraints with probability one. In
order to obtain such solutions, there exist two very different algorithmic
approaches, which, however, have very similar theoretical properties. This
chapter surveys theoretical foundations, experimental studies, and applications
for those two original approaches and new ones derived from them.
Chapter 6 is a review of external-memory search for state space problems, giving
detailed descriptions of algorithms and data structures, complemented by
concrete examples. Implementation on a GPU is discussed and speedups are
substantiated by experiment.
Chapter 7 presents a framework for the development and evaluation of real-time
rendering algorithms. A central concept is a meta rendering algorithm that
automatically selects an algorithm for the visualization of highly complex
scenes.
Chapter 8 applies the Algorithm Engineering cycle of design, analysis, imple-
mentation, and experimentation to robust optimization. In such problems, the
exact data is not known but bounded by a set of possible realizations. The
importance of considering real-world applications is demonstrated.
Chapter 9 gives a survey on concepts and algorithms for finding clusters in net-
works that change over time. Data sets for experimentation, comprised of
real-world and synthetic data, are thoroughly discussed.
Preface VII
Chapter 10 Many industrial production planning problems have both sequencing
and allocation aspects. This chapter describes and experimentally evaluates a
framework based on genetic algorithms that can smoothly integrate both
aspects.
Chapter 11 A streaming algorithm for the bipartite matching problem is gradually
improvedusingexperimentation.Thisfinallyresultsinaversionthatinheritsall
the good theoretical properties of the original version while being much faster
in practice.
Chapter 12 is a survey and a unified, extensive experimental comparison of algo-
rithms for the art gallery problem – a classic and important problem from
computational geometry. Moreover, a new and superior implementation is
presented, which combines the best techniques identified in previous
approaches.
February 2016 Lasse Kliemann
Peter Sanders
Contents
Engineering a Lightweight and Efficient Local Search SAT Solver . . . . . . . . 1
Adrian Balint and Uwe Schöning
Route Planning in Transportation Networks . . . . . . . . . . . . . . . . . . . . . . . . 19
Hannah Bast, Daniel Delling, Andrew Goldberg,
Matthias Müller-Hannemann, Thomas Pajor, Peter Sanders,
Dorothea Wagner, and Renato F. Werneck
Theoretical Analysis of the k-Means Algorithm – A Survey. . . . . . . . . . . . . 81
Johannes Blömer, Christiane Lammersen, Melanie Schmidt,
and Christian Sohler
Recent Advances in Graph Partitioning . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
Aydın Buluç, Henning Meyerhenke, Ilya Safro, Peter Sanders,
and Christian Schulz
How to Generate Randomized Roundings with Dependencies
and How to Derandomize Them. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
Benjamin Doerr and Magnus Wahlström
External-Memory State Space Search. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
Stefan Edelkamp
Algorithm Engineering Aspects of Real-Time Rendering Algorithms. . . . . . . 226
Matthias Fischer, Claudius Jähn, Friedhelm Meyer auf der Heide,
and Ralf Petring
Algorithm Engineering in Robust Optimization. . . . . . . . . . . . . . . . . . . . . . 245
Marc Goerigk and Anita Schöbel
Clustering Evolving Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 280
Tanja Hartmann, Andrea Kappes, and Dorothea Wagner
Integrating Sequencing and Scheduling: A Generic Approach
with Two Exemplary Industrial Applications . . . . . . . . . . . . . . . . . . . . . . . 330
Wiebke Höhn and Rolf H. Möhring
Engineering aBipartite Matching Algorithm in theSemi-Streaming Model. . . 352
Lasse Kliemann
X Contents
Engineering Art Galleries. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 379
Pedro J. de Rezende, Cid C. de Souza, Stephan Friedrichs,
Michael Hemmer, Alexander Kröller, and Davi C. Tozoni
Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 419