Instant download Combinatorial algorithms 25th international workshop iwoca 2014 duluth mn usa octob

Page 1


International Workshop IWOCA 2014

Duluth MN USA October 15 17 2014

Revised Selected Papers 1st Edition Kratochvíl Jan

Visit to download the full and correct content document: https://textbookfull.com/product/combinatorial-algorithms-25th-international-workshop -iwoca-2014-duluth-mn-usa-october-15-17-2014-revised-selected-papers-1st-editionkratochvil-jan/

More products digital (pdf, epub, mobi) instant download maybe you interests ...

Algorithms and Computation 25th International Symposium ISAAC 2014 Jeonju Korea December 15 17 2014 Proceedings 1st Edition Hee-Kap Ahn

https://textbookfull.com/product/algorithms-and-computation-25thinternational-symposium-isaac-2014-jeonju-koreadecember-15-17-2014-proceedings-1st-edition-hee-kap-ahn/

Medical Computer Vision Algorithms for Big Data

International Workshop MCV 2014 Held in Conjunction with MICCAI 2014 Cambridge MA USA September 18 2014

Revised Selected Papers 1st Edition Bjoern Menze

https://textbookfull.com/product/medical-computer-visionalgorithms-for-big-data-international-workshop-mcv-2014-held-inconjunction-with-miccai-2014-cambridge-ma-usaseptember-18-2014-revised-selected-papers-1st-edition-bjoernmenze/

Mathematical and Engineering Methods in Computer Science 9th International Doctoral Workshop MEMICS 2014 Tel■ Czech Republic October 17 19 2014 Revised Selected Papers 1st Edition Petr Hlin■ný

https://textbookfull.com/product/mathematical-and-engineeringmethods-in-computer-science-9th-international-doctoral-workshopmemics-2014-telc-czech-republic-october-17-19-2014-revisedselected-papers-1st-edition-petr-hlineny/

Digital Forensics and Watermarking 13th International Workshop IWDW 2014 Taipei Taiwan October 1 4 2014

Revised Selected Papers 1st Edition Yun-Qing Shi

https://textbookfull.com/product/digital-forensics-andwatermarking-13th-international-workshop-iwdw-2014-taipei-taiwanoctober-1-4-2014-revised-selected-papers-1st-edition-yun-qingshi/

Combinatorial Algorithms 27th International Workshop

IWOCA 2016 Helsinki Finland August 17 19 2016

Proceedings 1st Edition Veli Mäkinen

https://textbookfull.com/product/combinatorial-algorithms-27thinternational-workshop-iwoca-2016-helsinki-finlandaugust-17-19-2016-proceedings-1st-edition-veli-makinen/

Modelling and Simulation for Autonomous Systems First International Workshop MESAS 2014 Rome Italy May 5 6 2014 Revised Selected Papers 1st Edition Jan Hodicky (Eds.)

https://textbookfull.com/product/modelling-and-simulation-forautonomous-systems-first-international-workshop-mesas-2014-romeitaly-may-5-6-2014-revised-selected-papers-1st-edition-janhodicky-eds/

Information

Search Integration and Personalization 9th

International Workshop ISIP 2014 Kuala Lumpur Malaysia

October 9 10 2014 Revised Selected Papers 1st Edition

Dimitrios Kotzinos

https://textbookfull.com/product/information-search-integrationand-personalization-9th-international-workshop-isip-2014-kualalumpur-malaysia-october-9-10-2014-revised-selected-papers-1stedition-dimitrios-kotzinos/

Constructive Side Channel Analysis and Secure Design

5th International Workshop COSADE 2014 Paris France April 13 15 2014 Revised Selected Papers 1st Edition

Emmanuel Prouff (Eds.)

https://textbookfull.com/product/constructive-side-channelanalysis-and-secure-design-5th-international-workshopcosade-2014-paris-france-april-13-15-2014-revised-selectedpapers-1st-edition-emmanuel-prouff-eds/

Bayesian and grAphical Models for Biomedical Imaging

First International Workshop BAMBI 2014 Cambridge MA USA September 18 2014 Revised Selected Papers 1st

Edition M. Jorge Cardoso

https://textbookfull.com/product/bayesian-and-graphical-modelsfor-biomedical-imaging-first-international-workshopbambi-2014-cambridge-ma-usa-september-18-2014-revised-selectedpapers-1st-edition-m-jorge-cardoso/

Jan Kratochvíl · Mirka Miller

Dalibor

Froncek (Eds.)

Combinatorial Algorithms

25th International Workshop, IWOCA 2014 Duluth, MN, USA, October 15–17, 2014

Revised Selected Papers

LectureNotesinComputerScience8986

CommencedPublicationin1973

FoundingandFormerSeriesEditors: GerhardGoos,JurisHartmanis,andJanvanLeeuwen

EditorialBoard

DavidHutchison

LancasterUniversity,Lancaster,UK

TakeoKanade

CarnegieMellonUniversity,Pittsburgh,PA,USA

JosefKittler UniversityofSurrey,Guildford,UK

JonM.Kleinberg CornellUniversity,Ithaca,NY,USA

FriedemannMattern

ETHZurich,Zürich,Switzerland

JohnC.Mitchell StanfordUniversity,Stanford,CA,USA

MoniNaor

WeizmannInstituteofScience,Rehovot,Israel

C.PanduRangan

IndianInstituteofTechnology,Madras,India

BernhardSteffen

TUDortmundUniversity,Dortmund,Germany

DemetriTerzopoulos UniversityofCalifornia,LosAngeles,CA,USA

DougTygar UniversityofCalifornia,Berkeley,CA,USA

GerhardWeikum

MaxPlanckInstituteforInformatics,Saarbrücken,Germany

Moreinformationaboutthisseriesathttp://www.springer.com/series/7407

JanKratochvíl • MirkaMiller

DaliborFroncek(Eds.)

Combinatorial Algorithms

25thInternationalWorkshop,IWOCA2014 Duluth,MN,USA,October15–17,2014

RevisedSelectedPapers

Editors

CharlesUniversity

Praha

CzechRepublic

MirkaMiller UniversityofNewcastle

Newcastle,NSW Australia

UniversityofMinnesota,Duluth

Duluth,MN USA

ISSN0302-9743

LectureNotesinComputerScience

ISSN1611-3349(electronic)

ISBN978-3-319-19314-4ISBN978-3-319-19315-1(eBook) DOI10.1007/978-3-319-19315-1

LibraryofCongressControlNumber:2015940420

SpringerChamHeidelbergNewYorkDordrechtLondon © SpringerInternationalPublishingSwitzerland2015

Thisworkissubjecttocopyright.AllrightsarereservedbythePublisher,whetherthewholeorpartofthe materialisconcerned,specificallytherightsoftranslation,reprinting,reuseofillustrations,recitation, broadcasting,reproductiononmicrofilmsorinanyotherphysicalway,andtransmissionorinformation storageandretrieval,electronicadaptation,computersoftware,orbysimilarordissimilarmethodologynow knownorhereafterdeveloped.

Theuseofgeneraldescriptivenames,registerednames,trademarks,servicemarks,etc.inthispublication doesnotimply,evenintheabsenceofaspecificstatement,thatsuchnamesareexemptfromtherelevant protectivelawsandregulationsandthereforefreeforgeneraluse.

Thepublisher,theauthorsandtheeditorsaresafetoassumethattheadviceandinformationinthisbookare believedtobetrueandaccurateatthedateofpublication.Neitherthepublishernortheauthorsortheeditors giveawarranty,expressorimplied,withrespecttothematerialcontainedhereinorforanyerrorsor omissionsthatmayhavebeenmade.

Printedonacid-freepaper

SpringerInternationalPublishingAGSwitzerlandispartofSpringerScience+BusinessMedia (www.springer.com)

Preface

The25thInternationalWorkshoponCombinatorialAlgorithms(IWOCA)washeld duringOctober15–17,2014,inthepicturesqueharbortownDuluth,locatedinthe south-westcornerofLakeSuperiorinMinnesota,USA.Autumnisafavoritetime oftheyearforvisitingDuluth,owingtotheamazingrangeofcolorsoftreeandshrub folliageondisplayatthistimeoftheyear.TheIWOCA2014OrganizingCommittee timedtheeventperfectly!

IWOCA – theworkshopthatoriginated25yearsagoasthe(Australasian)AWOCA – hasovertheyearsestablisheditselfasatrulyinternationalconference.Thename change(toIWOCA)refl ectedtheexpanseoftheconferencebeyondlocalboundaries, motivatedbythegrowingglobalinterestintheconference.The fi rstIWOCAevents werestillheldinAustraliain2007,andthesubsequentyearsbroughtittoJapan (2008),theCzechRepublic(2009),theUK(2010),Canada(2011),India(2012), France(2013),andtotheUSAthisyear.Duringthelastsixyearstheproceedingshave beenpublishedbySpringerintheLNCSseries.

IWOCA2014received68submissions,mostofthemofveryhighquality.The ProgramCommitteewasfacedwithhardworkandsometimesdifficultdecisionsand weregrettedthatsomegoodpapershadtoberejectedbecauseofthelimitedcapacity oftheconferenceschedule.Intheend,32contributedtalkswerepresentedduringthe conference.

Wewouldliketothankallwhohavesenttheirsubmissionsandtocongratulateall theauthorsoftheacceptedpapers.Weextendspecialthankstothedistinguished invitedspeakersJosepDomingo-Ferrer,PinarHeggernes,SakethSaurab,andXuding Zhu.Wealsothankalltheauthorswhosubmittedpostersforthepostersession(which are,however,notincludedintheseproceedings).

Finally,wethankallthemembersoftheProgramCommittee,allexternalreviewers, andallthemembersoftheOrganizingCommitteeforallthehardworktheyhavedone. Whileallcommitteemembersworkedwellasateam,somenamesmustbesingledout: SpecialthanksgotoSergeiBezrukovfortirelesslyupdatingthewebsiteandrunning thetechnologysupportduringtheworkshop,andtoXiaofengGuforhandlingtechnicalissuesofpapersincludedinboththepre-workshopproceedingsandthisvolume.

March2015

DaliborFroncek JanKratochvíl MirkaMiller

Organization

ProgramCommittee

JemalAbawajy DeakinUniversity,Australia

HideoBannai KyushuUniversity,Japan

LjiljanaBrankovic UniversityofNewcastle,Australia

StoltingBrodal AarhusUniversity,Denmark

PinoCaballero-GilUniversityofLaLaguna,Spain

CharlieColbourn ArizonaStateUniversity,USA

MaximeCrochemoreKing’sCollegeLondon,UK

PinarDundar EgeUniversity,Turkey

JiriFiala CharlesUniversityinPrague,CzechRepublic

DaliborFroncek UniversityofMinnesotaDuluth,USA

RobertoGrossi Università diPisa,Italy

JanHolub CzechTechnicalUniversityinPrague,CzechRepublic CostasIliopoulos King’sCollegeLondon,UK

RalfKlasing LaBRIandCNRS,France

ChristianKomusiewiczTUBerlin,Germany

JanKratochvíl CharlesUniversityinPrague,CzechRepublic

DieterKratsch UniversitedeLorraine-Metz,France

GregoryKucherov UniversityParis-EstMarne-la-ValleeandCNRS, France

ThierryLecroq Université deRouen,France

ZsuzsannaLipták Università diVerona,Italy

PaulManuel KuwaitUniversity,Kuwait

MirkaMiller UniversityofNewcastle,Australia

IanMunro UniversityofWaterloo,Canada

KunsooPark SeoulNationalUniversity,SouthKorea

SolonPissis King’sCollegeLondon,UK

HebertPérez-Ros ésUniversityofLleida,Spain

SohelRahman BangladeshUniversityofEngineeringandTechnology, Bangladesh

VojtaRodl EmoryUniversity,USA

FrankRuskey UniversityofVictoria,Canada

BillSmyth McMasterUniversity,Canada

LynetteVanZijl StellenboschUniversity,SouthAfrica

ProgramCommitteeCo-chairs

JanKratochvíl CharlesUniversityinPrague,CzechRepublic MirkaMiller UniversityofNewcastle,Australia

ProblemSessionCo-chairs

UweLeck UniversityofWisconsin-Superior,USA ZsuzsannaLipták Università diVerona,Italy

ProceedingsTechnicalEditor

XiaofengGu UniversityofWisconsin-Superior,USA

SteeringCommittee

CostasIliopoulos King’sCollegeLondon,UK MirkaMiller UniversityofNewcastle,Australia BillSmyth McMasterUniversity,Canada

OrganizingCommittee

SergeiBezrukov UniversityofWisconsin-Superior,USA DaliborFroncek(Chair)UniversityofMinnesotaDuluth,USA XiaofengGu UniversityofWisconsin-Superior,USA StevenRosenberg UniversityofWisconsin-Superior,USA UweLeck UniversityofWisconsin-Superior,USA

AdditionalReviewers

Avrachenkov,Konstantin Baca,Martin Baisya,Dipankar Bari,Md.Faizul Bevern,René Van Boeckenhauer,Hans-Joachim Burcsi,Péter Caballero-Gil,Candido CaceresCruz,Jose Cechlarova,Katarina Chen,Jiehua Cheng,Eddie Cicalese,Ferdinando D’Arco,Paolo Dev,Himel

Dogan,Derya Escoffer,Bruno Feria-Puron,Ramiro Fernau,Henning Fertin,Guillaume Fici,Gabriele Firoz,Jesun Foucaud,Florent Gerbner,Daniel Grigorious,Cyriac Gu,Xiaofeng Harju,Tero Hartung,Sepp Hocquard,Hervé Hoksza,David

Hossain,Md.Iqbal Hsieh,Sun-Yuan Hüffner,Falk I.,Tomohiro Inenaga,Shunsuke Irvine,Veronika Islam,A.S.M.Sohidull Keil,Mark Lefebvre,Arnaud Li,Rao Liu,Daphne Lukovszki,Tamas Martin-Fernandez,Francisco Mary,Arnaud Medvedev,Paul Miltzow,Tillmann Molina-Gil,Jezabel Mondal,Debajyoti Nisse,Nicolas Oner,Tahsin Ordin,Burak Peterlongo,Pierre Phanalasy,Oudone Pineda-Villavicencio, Guillermo Prieur-Gaston,Elise

Rajan,Bharati Rios-Solis,Yasmin Rivals,Eric Rosenberg,Steve Ryan,Joe Ryjacek,Zdenek Saba,Sahand Salikhov,Kamil Scholtzova,Jirina Sebe,Francesc Sgall,Jiri Sgall,Jiří Shaw,DipanLal Sorge,Manuel Stephen,Sudeep Suchy,Ondrej Tantau,Till Tiwary,HansRaj Valla,Tomáš Valtr,Pavel Vialette,Stéphane Walen,Tomasz Weimann,Oren Xu,Min Zhu,Xuding Zohora,FatemaTuz

Contents

OntheComplexityofVariousParameterizationsofCommonInduced SubgraphIsomorphism......................................1 FaisalN.Abu-Khzam, ÉdouardBonnet,andFlorianSikora

ApproximationandHardnessResultsfortheMaximumEdges inTransitiveClosureProblem..................................13 AnnaAdamaszek,GuillaumeBlin,andAlexandruPopa

QuantifyingPrivacy:ANovelEntropy-BasedMeasureofDisclosureRisk...24 MousaAlfalaylehandLjiljanaBrankovic

OntheGaloisLatticeofBipartiteDistanceHereditaryGraphs...........37 NicolaApollonio,MassimilianoCaramia,andPaoloGiulioFranciosa

FastandSimpleComputationsUsingPrefixTablesUnderHamming andEditDistance..........................................49 CarlBarton,CostasS.Iliopoulos,SolonP.Pissis,andWilliamF.Smyth BorderCorrelations,Lattices,andtheSubgraphComponentPolynomial....62 FrancineBlanchet-Sadri,MichelleCordier,andRachelKirsch

ComputingMinimumLengthRepresentationsofSetsofWords ofUniformLength.........................................74 FrancineBlanchet-SadriandAndrewLohr

ComputingPrimitively-RootedSquaresandRunsinPartialWords........86 FrancineBlanchet-Sadri,JordanNikkel,J.D.Quigley,andXufanZhang

3-ColoringTriangle-FreePlanarGraphswithaPrecolored9-Cycle........98 IlkyooChoi,JanEkstein,PřemyslHolub,andBernardLidický

ComputingHeatKernelPagerankandaLocalClusteringAlgorithm.......110 FanChungandOliviaSimpson

A C-magicRectangleSetandGroupDistanceMagicLabeling...........122 SylwiaCichacz

SolvingMatchingProblemsEfficientlyinBipartiteGraphs.............128 SelmaDjelloul

A3-ApproximationAlgorithmforGuardingOrthogonalArtGalleries withSlidingCameras.......................................140 StephaneDurocherandSaeedMehrabi

OnDecomposingtheCompleteGraphintotheUnion ofTwoDisjointCycles......................................153

SaadI.El-Zanati,UthoompornJongthawonwuth,HeatherJordon, andCharlesVandenEynden

ReconfigurationofVertexCoversinaGraph.......................164 TakehiroIto,HiroyukiNooka,andXiaoZhou

SpaceEfficientDataStructuresforNearestLargerNeighbor............176 VarunkumarJayapaul,SeungbumJo,VenkateshRaman, andSrinivasaRaoSatti

PlayingSeveralVariantsofMastermindwithConstant-SizeMemory isnotHarderthanwithUnboundedMemory.......................188 GeroldJägerandMarcinPeczarski

OnMaximumCommonSubgraphProblemsinSeries-ParallelGraphs......200 NilsKriege,FlorianKurpicz,andPetraMutzel

Profile-BasedOptimalMatchingsintheStudent/ProjectAllocationProblem....213 AugustineKwanashie,RobertW.Irving,DavidF.Manlove, andColinT.S.Sng

TheMin-maxEdge q-ColoringProblem...........................226 TommiLarjomaaandAlexandruPopa

SpeedingupGraphAlgorithmsviaSwitchingClasses.................238

NathanLindzey

Studyof jðDÞ for D ¼f2; 3; x; yg ...............................250 DanielCollisterandDaphneDer-FenLiu

SomeHamiltonianPropertiesofOne-Con flictGraphs.................262 ChristianLaforestandBenjaminMomège

SequenceCoveringArraysandLinearExtensions....................274 PatrickC.MurrayandCharlesJ.Colbourn

Minimum r -StarCoverofClass-3OrthogonalPolygons...............286 LeonidasPaliosandPetrosTzimas

EmbeddingCirculantNetworksintoButterflyandBenesNetworks........298 R.SundaraRajan,IndraRajasingh,PaulManuel,T.M.Rajalaxmi, andN.Parthiban

KineticReverse k -NearestNeighborProblem.......................307 ZahedRahmati,ValerieKing,andSueWhitesides

EfficientlyListingBoundedLength st -Paths........................318

RomeoRizzi,GustavoSacomoto,andMarie-FranceSagot

MetricDimensionforAmalgamationsofGraphs.....................330

RinoviaSimanjuntak,SaladinUttunggadewa,andSuhadiWidoSaputro

ASuffixTreeOrNotaSuffixTree?.............................338

TatianaStarikovskayaandHjalteWedelVildhøj

DeterministicAlgorithmsfortheIndependentFeedbackVertex SetProblem..............................................351

YumaTamura,TakehiroIto,andXiaoZhou

LosslessSeedsforSearchingShortPatternswithHighErrorRates........364

ChristopheVroland,MikaëlSalson,andHélèneTouzet

AuthorIndex

OntheComplexityofVariousParameterizations ofCommonInducedSubgraphIsomorphism

FaisalN.Abu-Khzam1 , ´ EdouardBonnet2 ,andFlorianSikora2(B)

1 LebaneseAmericanUniversity,Beirut,Lebanon faisal.abukhzam@lau.edu.lb

2 PSL,Universit´eParis-Dauphine,LAMSADE,UMRCNRS7243,Paris,France {florian.sikora,edouard.bonnet}@dauphine.fr

Abstract. MaximumCommonInducedSubgraph (henceforth MCIS)isamongthemoststudiedclassical NP-hardproblems.MCIS remains NP-hardonmanygraphclassesincludingbipartitegraphs,planargraphsand k -trees.Littleisknown,however,abouttheparameterizedcomplexityoftheproblem.Whenparameterizedbythevertex covernumberoftheinputgraphs,theproblemwasrecentlyshowntobe fixed-parametertractable.Capitalizingonthisresult,weshowthatthe problemdoesnothaveapolynomialkernelwhenparameterizedbyvertex coverunless NP ⊆ coNP/poly .Wealsoshowthat MaximumCommon ConnectedInducedSubgraph (MCCIS),whichisavariantwhere thesolutionmustbeconnected,isalsofixed-parametertractablewhen parameterizedbythevertexcovernumberofinputgraphs.Bothproblemsareshowntobe W [1]-completeonbipartitegraphsandgraphsof girthfiveand,unless P = NP,theydonotbelongtotheclass XP when parameterizedbyaboundonthesizeoftheminimumfeedbackvertex setsoftheinputgraphs,thatissolvingtheminpolynomialtimeisvery unlikelywhenthisparameterisaconstant.

1Introduction

Acommoninducedsubgraphoftwographs G1 and G2 isagraphthatisisomorphictoinducedsubgraphsofeach.Theproblemoffindingacommoninduced subgraphofmaximumnumberofvertices(oredges)hasmanyapplicationsin anumberofdomainsincludingbioinformaticsandchemistry[11–13, 16, 17].In thedecisionversionoftheproblem,wearegivenaninteger k andthequestion istodecideifthereisasolutionwithatleast k vertices.Wesaythat k isthe naturalparameteroftheproblem,thatisthesolutionsize.

Concerningitsclassicalcomplexity, MaximumCommonInducedSubgraph is NP-complete,andremainssoonbipartitegraphsandgraphswith boundedtreewidth.However,theproblemisin P fortrees[10]andgraphsof (both)boundedtreewidthandboundeddegree[3].

Adecisionsubproblemof MaximumCommonInducedSubgraph isthe wellknown InducedSubgraphIsomorphism (ISI)problem,whichconsistsof decidingwhether G1 isisomorphictoaninducedsubgraphof G2 .Inotherwords, c SpringerInternationalPublishingSwitzerland2015 J.Kratochv´ıletal.(Eds.):IWOCA2014,LNCS8986,pp.1–12,2015. DOI:10.1007/978-3-319-19315-1 1

2F.N.Abu-Khzametal.

itisequivalentto MaximumCommonInducedSubgraph where k = |G1 |. Inthiscase G1 iscalledthepatterngraphwhile G2 isthehostgraph.ISIis W [1]-hardingeneral,byastraightforwardreductionfrom k -Clique.Therefore MCISisalso W [1]-hard.Ontheotherhand,ifISIisin FPT onacertaingraph class,thensoisMCIS.Toseethis,notethatanarbitraryinstance(G1 ,G2 ,k ) ofMCIScanbereducedinfpt-timetotwoinstancesofISIbyenumeratingall possiblegraphson k verticesandcheckingwhethereachisaninducedsubgraph ofeachofthetwoinputgraphs.ThisimpliesthatISIandMCIShavethesame parameterizedcomplexitywhenparameterizedbythesolutionsize,whichwe refertoasthenaturalparameterinthispaper.Ofcourse,thelatterreduction takestime O (2k 2 )(multipliedbythetimeneededtosolveISIonthegiven graphclass),whichmakesitprohibitivelyimpractical.Weshallprovideasimpler reductionthattakes O (ck )-timeonaclassofgraphsthatincludes H -minorfree graphsandgraphsofboundeddegree.

Anotherwaytodealwiththehardnessofaproblemistostudyitscomplexity withrespecttoauxiliary(orstructural)parameters,tobetterunderstandthe behavioroftheproblem(seeforexample[8]).MCISisalreadyhardongraphs withboundedtreewidth,being NP-hardonforests,asweshallobservebased onaclassicalresultfromGareyandJohnson[10].Accordingly,theproblemis W [1]-hardwhenparameterizedbythetreewidthoftheinputgraphs.Therefore weneedtolookforbiggerparameters.Weshallstudytheproblemwithrespect tothesizeofa(minimum)feedbackvertexsetthatofa(minimum)vertexcover ofinputgraphs.WeobservethatMCISisnotin XP whenparameterizedby thefeedbackvertexsetnumberoftheinputgraphs.Thisalsoimpliesthatthe problemisnotin XP whenparameterizedbytreewidth.

WeobservethatISIremains W [1]-hardongraphswhere G1 hasa k -vertex coverbyareductionfromthe W [1]-hard InducedBipartiteMatching problem[14]:ifthepatternconsistsof k disjointedges,itsvertexcoveris k .Therefore, MCISis W [1]-hardwhentheparameteristhevertexcoverofoneoftheinput graphs,eveniftheothergraphisbipartite.However,iftheparameter k isthe combinationofthevertexcoverofbothinputgraphs,thentheproblemisin FPT,witharunningtimeof O ((24k )k )[1].WeshallproveinSect. 3 thatMCIS doesnothaveapolynomial-sizekernelinthiscaseunless NP ⊆ coNP/poly

Wealsoconsiderthe MaximumCommonConnectedInducedSubgraph problem.Weobservethattheproblemisin FPT ongraphsofboundeddegree andshowittobe W [1]-completeontheclassofbipartitegraphs,eveniftheinput graphis C4 -free.Consequently,MCCISis W [1]-completeongraphsofgirthfive. Finally,weshowthatMCCISisfixed-parametertractablewhenparameterized byaboundontheminimumvertexcoversoftheinputgraphs.

2Preliminaries

Twofinitegraphs G1 =(V1 ,E1 )and G2 =(V2 ,E2 )are isomorphic ifthereis abijection π : V1 → V2 suchthat ∀u,v ∈ V1 : uv ∈ E1 ⇔ π (u)π (v ) ∈ E2 . Givenagraph G =(V,E ),agraph G =(V ,E )isan inducedsubgraph of G

if V ⊆ V and E = E (V )= {uv ∈ E | u,v ∈ V },i.e. E istheedgesetwith bothextremitiesin V .Wealsosaythat G isthesubgraphof G inducedby V .

The girth ofagraph G isthelengthoftheshortestcyclecontainedin G. Contractinganedge uv consistsofdeleting uv andreplacingthevertices u and v byasinglevertex w intheincidencerelation(edgesincidenton u or v become incidenton w ).Agraph H isa topologicalminor ofgraph G if H isobtained fromasubgraphof G byapplyingzeroormoreedgecontractions.Givenafixed graph H ,afamily F ofgraphsissaidtobe H -minorfree if H isnotaminorof anyelementof F

The MaximumCommonInducedSubgraph problemisdefinedformally asfollows.

MaximumCommonInducedSubgraph(MCIS):

• Input:Twographs G1 =(V1 ,E1 )and G2 =(V2 ,E2 ).

• Output:Aninducedsubgraph G1 of G1 isomorphictoaninduced subgraph G2 of G2 withamaximumnumberofvertices.

MaximumCommonConnectedInducedSubgraph (MCCIS)isdefined asMCISwiththeadditionalrestrictionthatthesolutionmustbeconnected. InducedSubgraphIsomorphism isdefinedsimilarly:

InducedSubgraphIsomorphism(ISI):

• Input:Twographs G1 =(V1 ,E1 )and G2 =(V2 ,E2 ).

• Output:Aninducedsubgraph G1 of G1 isomorphicto G2

Parameterizedcomplexity. Aparameterizedproblem(I,k )issaid fixed-parameter tractable (orintheclass FPT)w.r.t.(withrespectto)parameter k ifitcanbe solvedin f (k ) ·|I |c time(i.e.infpt-time),where f isanycomputablefunctionand c isaconstant(see[7, 15]formoredetailsaboutfixed-parameter tractability).Theparameterizedcomplexityhierarchyiscomposedoftheclasses FPT ⊆ W [1] ⊆ W [2] ⊆···⊆ XP.Theclass XP containsproblemssolvablein time f (k ) ·|I |g (k ) ,where f and g areunrestrictedfunctions.A W [1]-hardproblemisnotfixed-parametertractable(unless FPT = W [1])andonecanprove W [1]-hardnessbymeansofa parameterizedreduction froma W [1]-hardproblem.Thisisamappingofaninstance(I,k )ofaproblem A1 in g (k ) ·|I |O (1) time(foranycomputablefunction g )intoaninstance(I ,k )for A2 suchthat (I,k ) ∈ A1 ⇔ (I ,k ) ∈ A2 and k ≤ h(k )forsomefunction h.

Apowerfultechniquetodesignparameterizedalgorithmsis kernelization.In short,kernelizationisapolynomial-timeself-reductionalgorithmthattakesan instance(I,k )ofaparameterizedproblem P asinputandcomputesanequivalentinstance(I ,k )of P suchthat |I | h(k )forsomecomputablefunction h and k k .Theinstance(I ,k )iscalleda kernel inthiscase.Ifthefunction h ispolynomial,wesaythat(I ,k )isapolynomialkernel.Itiswellknown thataproblemisin FPT iffithasakernel,butthisequivalenceyieldssuperpolynomialkernels(ingeneral).Todesignefficientparameterizedalgorithms,a

4F.N.Abu-Khzametal.

kernelofpolynomial(orevenlinear)sizein k isimportant.However,somelower boundsonthesizeofthekernelcanbeshownunlesssomepolynomialhierarchycollapses.Toshowthisresult,wewillusethecrosscompositiontechnique developedbyBodlaenderetal.[4].

Definition1(PolynomialEquivalenceRelation[4]). Anequivalencerelation R on Σ∗ issaidtobe polynomial ifthefollowingtwoconditionshold:(i) Thereisanalgorithmthatgiventwostrings x,y ∈ Σ∗ decideswhether x and y belongtothesameequivalenceclassintime (|x| + |y |)O (1) .(ii)Foranyfinite set S ⊆ Σ∗ theequivalencerelation R partitionstheelementsof S intoatmost (maxx∈S |x|)O (1) classes.

Definition2(OR-Cross-Composition[4]). Let L ⊆ Σ∗ beasetandlet Q ⊆ Σ∗ × N beaparameterizedproblem.Wesaythat L cross-composes into Q ifthereisapolynomialequivalencerelation R andanalgorithmwhich,given t strings x1 ,x2 ,...,xt belongingtothesameequivalenceclassof R,computes aninstance (x∗ ,k ∗ ) ∈ Σ∗ × N intimepolynomialin t i=1 |xi | suchthat:(i) (x∗ ,k ∗ ) ∈ Q ⇔ xi ∈ L forsome 1 i t.(ii) k ∗ isboundedbyapolynomialin maxt i=1 |xi | +log t.

Proposition1([4]). Let L ⊆ Σ∗ beasetwhichis NP-hardunderKarpreductions.If L cross-composesintotheparameterizedproblem Q,then Q hasno polynomialkernelunless NP ⊆ coNP/poly

Aparameterizedproblemissaidtobe fixed-parameterenumerable ifallfeasible solutionscanbeenumeratedin O (f (k )|I |c )where f isacomputablefunctionof theparameter k only,and c isaconstant.

3StructuralParameterizationofMaximumCommon InducedSubgraph

Letusfirstrecallthat tw (G) fvs(G) vc(G),where tw (G)(resp. fvs(G), vc(G))representsthetreewidth(resp.thefeedbackvertexsetnumber,thevertexcovernumber)of G [8].Asnotedbefore,iftheparameteristhecombination of tw (G1 )and tw (G2 )thenMCISisknowntobe W [1]-hard.Evenmore,ifthe parameteristhecombinationof fvs(G1 )and fvs(G2 )(whichisbiggerthanthe combinationofthetreewidth),thentheproblemisnotevenin XP since MaximumCommonInducedSubgraph and InducedSubgraphIsomorphism are NP-hardonforests,acasewheretheparameterisequalto0.Indeed,onecan modifythereductionfrom 3-partition donebyGareyandJohnsonin[10]for SubforestIsomorphism toourproblem,bybuildingchainsof B +3vertices insteadof B +1in G2 suchthateachchainof G1 isseparatedbyavertex.The followingtheoremfollows.

Theorem1. Unless P = NP, MaximumCommonInducedSubgraph isnot in XP whenparameterizedbyaboundontheminimumfeedbackvertexsetsof thepairofinputgraphs.

ThehardnessofMCISonforestalsoimpliesthefollowing.

Corollary1. Unless P = NP, MaximumCommonInducedSubgraph is notin XP whenparameterizedbythetreewidthoftheinputgraphs.

Itwasshownin[1]thatMCISisin FPT iftheparameteristhecombinationof vc(G1 )and vc(G2 ).Accordingly,theproblemhasakernel,butnopolynomial boundisknownonitssize.Weshowthat,inthiscase,thekernelcannotbe polynomialunless NP ⊆ coNP/poly

Theorem2. Unless NP ⊆ coNP/poly , MaximumCommonInducedSubgraph hasnopolynomialkernelwhenparameterizedbythesumofthesizesof vertexcoversinthetwoinputgraphs.

Proof. WewilldefineanOR-cross-compositionfromthe NP-complete Clique, problem,wherethegiveninstanceisatuple(Gc ,l )andthequestioniswhether thegraph Gc containsacliqueon l vertices.

Given t instances,(Gc 1 ,l1 ), (Gc 2 ,l2 ),..., (Gc t ,lt ),of Clique,where Gc i isa graphand li ∈ N, ∀1 i t,wedefineourequivalencerelation R suchthatany stringsthatarenotencodingvalidinstancesareequivalent,and(Gc i ,li ), (Gc j ,lj ) areequivalentiff |V (Gc i )| = |V (Gc j )|,and li = lj .Hereafter,weassumethat V (Gc i )= {1,...,n} and li = l ,forany1 i t.Wewillbuildaninstanceof MaximumCommonInducedSubgraph parameterizedbythevertexcover (G1 ,G2 ,l ,Z )where G1 and G2 aretwographs, l ∈ N and Z ⊆ V (G2 )isa vertexcoverof G2 computedinfpt-time,suchthatthereisasolutionofsize l for MaximumCommonInducedSubgraph iffthereisan i, 1 i t such thatthereisasolutionofsize l in Gc i .Wewillnowdescribehowtobuild G1 and G2

Tobuild G2 (seealsoFig. 1):

V (G2 )= {p,q,r }∪{ai | 1 i t}∪{euv | 1 u<v n}∪{vi | 1 i n}, –

E (G2 )1 = {pq,pr,qr }, –

E (G2 )2 = {rai | 1 i t}, – E (G2 )3 = {ai euv | uv ∈ E (Gc i )}, –

E (G2 )4 = {euv vu ,euv vv |∀1 u<v n}, –

E (G2 )= E (G2 )1 ∪ E (G2 )2 ∪ E (G2 )3 ∪ E (G2 )4 .

Tobuild G1 (seealsoFig. 2):

– V (G1 )= {p,q,r,a}∪{ei | 1 i l 2 }∪{vi | 1 i l }, – E (G1 )1 = {pq,pr,qr,ra}, – E (G1 )2 = {aei | 1 i l 2 }, –

E (G1 )3 = {ei vu ,ei vv |∀1 i l 2 ,ei = uv }, –

E (G1 )= E (G1 )1 ∪ E (G1 )2 ∪ E (G1 )3 .

Weset l = |V (G1 )|,and Z = {p,r }∪{euv |1 u<v n}.Itiseasytoseethat Z isindeedavertexcoverfor G2 andthatitssizeisequalto n(n 1) 2 +2,which

Illustrationoftheconstructionof G2

Fig.2. Illustrationoftheconstructionof G1 .

ispolynomialin n andhenceinthesizeofthelargestinstance.Notethatthe sizeofthegraph G1 doesnotdependon t andispolynomialin n,sothesizeof itsvertexcoverisalsopolynomialin n andindependentof t.

Letusshowthat G1 isaninducedsubgraphof G2 iffatleastoneofthe Gc i ’s hasacliqueofsize l . (⇐)Supposethat Gc i hasacliqueofsize l .Wedenoteby S ⊆ V (Gc i )a cliqueofsizeexactly l in Gc i .Weshowthatthereisaninducedsubgraph S of G2 ofsize l ,isomorphicto G1 .Weset V (S )= {p,q,r }∪{ai }∪{euv | ∀uv ∈ E (S )}∪{vu |u ∈ S }.Onecaneasilycheckthatthissubgraphisisomorphic to G1 .

(⇒)Assumenowthat G1 isaninducedsubgraphof G2 .Denoteby S the subgraphof G2 isomorphicto G1 .Notethattheonlytrianglein G2 is pqr .

Fig.1.

Indeed, T (V (G2 ) \{p})isbipartite.Thetriangle pqr in G1 hasthereforeto match pqr in G2 .Moreover, r in G1 hastomatch r in G2 since p and q haveno edgesbesidestheclique pqr .Thevertex a in G1 canonlymatchavertex ai for some i ∈{1,...,t}.Then, e1 upto e( l 2) in G1 hastomatch l 2 verticesin {euv | 1 u<v n} of G2 whichcorrespondtoactualedgesin Gc i .Finally, v1 upto vl in G1 hastomatch l verticesamongthe vj ’sin G2 .Notethatthenumberof edgesin E (G1 )3 betweenthe ej ’sandthe vj ’sisexactly2 l 2 = l (l 1).More precisely,each ej touches2edgesin E (G1 )3 andeach vj touches l 1edges in E (G1 )3 .Inordertogetamatchin G2 ,oneshouldfindasetof l 2 edges inducingexactly l vertices.So,thissetof l verticesisacliquein Gc i

NotethattheparameterofMCISinthisreductionisexactlythesizeof G1 Therefore,thisnegativeresultholdsforISItoo.

DespitethefactthatISIandMCIShavethesameparameterizedcomplexity whenparameterizedbythenaturalparameter,theyexhibitdifferentcomplexitieswithrespecttostructuralparameters.Infact,thelatterisnotevenin XP whenparameterizedbythevertexcoverofonlyoneofthetwographswhileISI is FPT whenparameterizedbythevertexcoverofthesecond(host)graph.To seethis,notethatwhenthehostgraphhasa k -vertexcover,theminimumsize ofavertexcoverinthepatterngraphmustbeboundedbytheparameter k , otherwisewehaveanoinstance.Theclaimfollowsfromthefixed-parameter tractabilityofMCISinthiscase[1].

AlthoughMCISisnotin XP w.r.t.somestructuralparameterssuchas treewidthandfeedbackvertexsetnumber,itis,togetherwithMCCISandISI, in W [1]w.r.t.thenaturalparameter.

Theorem3. MCIS,MCCISandISIare W [1]-completew.r.t.thenaturalparameter.

Proof. SinceISI,MCISandMCCISare W [1]-hardbyastraightforwardreduction from k -Clique,itsufficestoshowmembershipin W [1].In[5],itisshownthatif aproblemcanbereducedinFPTtimetosimulatinganon-deterministicsingletapedTuringMachinehaltinginatmost f (k )steps,forsomefunction f ,then itisin W [1].TheTuringMachinecanhaveanalphabetandasetofstatesof sizedependingonthesizeoftheinputoftheinitialproblem.Inourcase,we candesignaTuringMachinethatguessesin2k stepsthecorrespondingright k verticesin G1 (forISIthispartisnotnecessary)andtheright k verticesin G2 (ouralphabetbeingisomorphictoanindexingof V (G1 ) ∪ V (G2 ))andthen checkintime O (k 2 )whetherthetwoinducedsubgraphsareisomorphic(and thattheyareconnectedforMCCIS).

WenowturnourattentiontothecasewheretheMCISisparameterizedby acombinationofthenaturalparameterandsomestructuralparameter.For example,considerthecasewheretheparameteristhesumofsomebound t on thefeedbackvertexsetoftheinputgraphsandthenaturalparameter k .The problemis FPT inthiscasesincegraphsof t-feedbackvertexsetare H -minor free(let H bethe“fixed”graphconsistingofadisjointunionof t +1triangles).

8F.N.Abu-Khzametal.

Moreover,weknowISIis FPT inthiscasedueto[9].However,andasstated inSect. 1,asolutiontoaninstanceofMCIS(inthiscase)isobtainedviaan exhaustiveenumerationof O (2k 2 )instancesofISI.Thiscanbeimprovedon classesofgraphsthataregivenwithsomefixedcoloring t,suchas H -minorfree graphsandgraphsofboundedmaximumdegree.Infact,iftheinputtoMCISis apairof t-coloredgraphs,thenareductionalgorithmwouldfirstcheckwhether eachofthetwographshasan(independent)colorclassofsize k .Ifso,thenboth haveanedgelesscommonsubgraphofsize k .Otherwise,theorderofatleast oneofthetwographs,say G1 ,issmallerthan tk .Insuchcase,thealgorithm proceedsbyrunninga(fixed-parameter)algorithmforISIoneachofthe O (2tk ) inducedsubgraphsof G1

In[14]itwasshownthat InducedMatching is W [1]-hardonbipartite graphs.Asmentionedearlier,thisprovesthatMCISis W [1]-hardinthiscase. WeshowthatMCISremains W [1]-hardon C4 -freebipartitegraphs,whichproves its W [1]-hardnessongraphsofgirthfive.

Theorem4. MaximumCommonInducedSubgraph is W [1]-completew.r.t. sizeofthesolution,asparameter,evenon C4 -freebipartitegraphs.

Proof. Membershipin W [1]comesfromTheorem 3.Forthehardness,consider thefollowingreductionfromthe W [1]-hardproblem Clique.Givenaninstance (G =(V,E ),k )of Clique,webuildaninstance(G1 ,G2 ,k )ofourproblemas follows.Thegraph G2 isthebipartiteincidencegraphof G (thebipartitionis betweenverticesrepresenting V andverticesrepresenting E ),thegraph G1 is thebipartiteincidencegraphof Kk ,and k = k + k 2 = |V (G1 )|.

Notethatabipartiteincidencegraphis C4 -freesince,inasimplegraph,no twoedgesareincidentonthesamepairofvertices.

Itisclearthat G1 occursasaconnectedinducedsubgraphof G2 iffthereis acliqueofsize k in G,becausew.l.o.g. k> 2andtheverticesrepresentingedges in G1 and G2 areofdegree2.

Corollary2. MaximumCommonInducedSubgraph is W [1]-complete w.r.t.sizeofthesolutionongraphsofgirthfive.

4MaximumCommonConnectedInducedSubgraph

MaximumCommonConnectedInducedSubgraph istrivially FPT whenever InducedSubgraphIsomorphism is FPT,including H -minorfreegraphs, sincetheenumerationofall O (2k 2 )possibleinducedconnectedsubgraphscanbe usedasdescribedbefore.Theconverseisalsotrue.Infact,aninstance(G1 ,G2 ,k ) ofISIcanbereducedtoanequivalentinstance(G1 ,G2 ,k +1)ofMCCISbyletting Gi bethegraphobtainedbyaddingasingle(universal)vertexto Gi that ismadeadjacenttoallotherverticesof Gi .ItfollowsthatMCISandMCCIS havethesameparameterizedcomplexitywithrespecttothenaturalparameter (i.e.,solutionsize).

NotethatMCISis NP-hardonforestswhileMCCISissolvableinpolynomialtimeinthiscase:giventwoforests G1 and G2 ,runthepolynomial-timeMCCIS algorithmof[2]oneverypairoftreesfrom G1 and G2 . InthefollowingofthissectionwestudythecomplexityofMCCISwith respecttostructuralparameters.

Lemma1. InducedconnectedSubgraphIsomorphism is NP-hardeven whenbothgraphshavefeedbackvertexsetnumberequaltoone.

Proof. Givenaninstanceof InducedSubgraphIsomorphism onforests G1 and G2 (eachwithatleast2trees),webuildaninstanceof InducedconnectedSubgraphIsomorphism byaddingauniversalvertex(connectedto everynode)in G1 andin G2 .Onecanseethatthesetwouniversalverticesmust bematchedtogethersincetheyaretheonlyoneswithsufficientlyhighdegree. Then,thereisasolutionfor InducedSubgraphIsomorphism iffthereis asolutionfor InducedconnectedSubgraphIsomorphism.Theresultof courseholdsforMCCIStoo.

Corollary3. Unless P = NP, MaximumCommonConnectedInduced Subgraph isnotin XP whenparameterizedbyaboundoftheminimumfeedback vertexsetnumberoftheinputgraphs(andhencethenwhenparameterizedbya boundonthetreewidthofeachofthetwoinputgraphs).

Giventheabovenegativeresult,thenextquestioniswhetherMCCISisin FPT w.r.t.theparametervertexcover.In[1],aparameterizedalgorithmispresented forMCISwhentheparameterisaboundontheminimumvertexcovernumber oftheinputgraphs.However,thatalgorithmcannothelpusmuchforsolving MCCISsinceitreliesontheexistenceofafeasiblesolutionofsizeatleast ≈ n k whichconsistsofmappingthetwo big independentsetsofthetwo graphsontoeachother.Ofcourse,thisisnotafeasiblesolutionforMCCIS. InthefollowingweprovethatMCCISisfixed-parametertractablew.r.t. k = max(vc(G1 ),vc(G2 )).

Theorem5. MaximumCommonConnectedInducedSubgraph parameterizedbyaboundonthevertexcoversoftheinputgraphsisfixed-parameter tractable.

Proof. Intime O ∗ (2k )(even O ∗ (1.2738k )[6]),wecanfindminimumvertexcovers C1 and C2 in G1 and G2 respectively.Let I (j ) betheindependentset V (Gj ) \ Cj for j ∈{1, 2}.Byassumption,ourparameter k ismax(C1 ,C2 ),sowecan enumeratealltripartitionsof C1 and C2 intime O ∗ (9k ).Wedenoteby C1,m , C1,u and C1,i (respectively C2,m , C2,u and C2,i )thethreesetsofatripartitionof C1 (respectively C2 ).For j ∈{1, 2}, Cj,u correspondstotheverticesof Cj that arenotmatched,sotheymaybedeleted. Cj,m comprisestheverticesmatchedto C3 j,m (thatis,tothevertexcoveroftheothergraph),and Cj,i arethevertices matchedto I (3 j ) ,theindependentsetoftheothergraph.SeeFig. 3

Weobservethatfor j ∈{1, 2}, I (j ) canbepartitionedintoatmost2k classes oftwins: I (j ) 1 ,I (j ) 2 ,...I (j ) 2k .Aclassoftwinsinthiscontextisasetofverticeswith

10F.N.Abu-Khzametal.

anidenticalneighborhoodinthevertexcoverandthereareatmost2k subsets of Cj .Potentially,someclassescanbeempty:theycorrespondtoasubsetofthe vertexcover Cj thatisnotthe(exact)neighborhoodofanyvertexin I (j ) . Atthispoint,wecanenumeratethemappingsbetween C1,m and C2,m intime O ∗ (k k )andthemappingsbetween Cj,i and I (3 j ) intime O ∗ ((2k )k )= O ∗ (2k 2 ). Indeed,tomatchavertex u withavertex v oratwinof v isequivalent.Thus, intime O ∗ ((9k )k 2k 2 )wecanenumerateallthesolutionsofMCISwhereonly verticesof I (1) couldstillbematchedtoverticesof I (2) .Theoptimalmapofthe independentsetscanbedoneinlineartimebymatchingthegreatestnumberof verticesineach equivalent twinclass(whichisthesizeofthesmallerofthetwo equivalenttwinclasses),whereatwinclass I (j ) r in I (j ) isequivalenttoatwin class I (3 j ) s in I (3 j ) iftheverticesof N (I (j ) r ) \ Cj,u and N (I (3 j ) s ) \ C3 j,u are inone-to-onecorrespondence.

Fig.3. IllustrationoftheproofofTheorem 5.Dashedboxesrepresenttheclassesinside theindependentset.Arrowsrepresentthematchingbetweensetsofvertices.

TofindasolutionforMCCIS,thealgorithmdescribedintheaboveproofenumeratesallpossiblemaximalcommoninducedsubgraphsintime O ∗ ((9k )k 2k 2 ). Assuch,itcanbeusedasanenumerationalgorithmforMCIS.

Theorem6. MaximumCommonInducedSubgraph parameterizedbyvertexcover,isfixed-parameterenumerable.

Finally,thefollowingcorollariesfolloweasilyfromtheproofsofTheorems 2 and 4 sincethegraphsusedinbothproofsareconnected.

Corollary4. MaximumCommonConnectedInducedSubgraph,parameterizedbyaboundontheminimumvertexcoversofinputgraphs,doesnot haveapolynomial-sizekernelunless NP ⊆ coNP/poly .

Corollary5. MaximumCommonConnectedInducedSubgraph is W [1]completeonbipartitegraphsandgraphsofgirthfive.

Table1. SummaryofdifferentparameterizedcomplexityresultsofISI,MCISand MCCISfordifferentstructuralparameters.

vc + vc vc + fvs fvs + fvs vc

ISI FPT;noPolyKernel (Theorem 2)

M(C)CIS FPT ([1],Theorem 5); noPolyKernel

Open / ∈ XP

FPT for vc(G2 ), / ∈ XP for vc(G1 )

Open / ∈ XP (Corollary 3) / ∈ XP

Inthefollowingtablewegiveasummaryofsomeresultsobtainedinthis paperalongwithopenquestions.NotethatforISI, vc + fvs isnotthesame parameteras fvs + vc.Inthelatter,theparameterisaboundonthevertex coverof G2 (aswellasthefeedbackvertexsetof G1 )whichmakesISIin FPT, whileitremainsopenfor vc + fvs.WealsonotethatISIisnotin XP w.r.t. vc(G1 )byasimplereductionfrom IndependentSet (let G2 beanedgeless graphon k vertices,thenitsvertexcovernumberis0).

5Conclusion

Westudiedthe MaximumCommonInducedSubgraph and MaximumCommonConnectedInducedSubgraph problemswithrespecttothesolution sizeasnaturalparameteronspecialgraphsclasses,suchasforests,bipartite graphsandgraphsofgirthfive.Thetwoproblemsarefixed-parametertractable on H -minorfreegraphs,whichincludeforests,buttheyare W [1]-completeon bipartitegraphsandgraphsofgirthfive.

Wealsoconsideredtheuseofauxiliaryparameters,suchasaboundonthe minimumvertexcoversoftheinputgraphs.AlthoughbothMCISandMCCIS arein FPT inthiscase,weprovedthatnokernelofpolynomialboundcan beobtainedunless NP ⊆ coNP/poly .WenotedthatMCISisnotevenin XP withrespecttoother(smaller)auxiliaryparameters,suchastreewidthandfeedbackvertexset(seeTable 1).Afewcorrespondingopenproblemsremaintobe addressed.Forexample,areMCIS/MCCISin FPT whenparameterizedbythe combinationofthevertexcovernumberandthefeedbackvertexsetnumber,or bythevertexcovernumberandthetreewidth?

Acknowledgements. Workpartiallysupportedbythebilateralresearchcooperation CEDREbetweenFranceandLebanon(grantnumber30885TM).

References

1.Abu-Khzam,F.N.:Maximumcommoninducedsubgraphparameterizedbyvertex cover.Inf.Process.Lett. 114(3),99–103(2014)

2.Akutsu,T.:AnRNCalgorithmforfindingalargestcommonsubtreeoftwotrees. IEICETrans.Inf.Syst. 75(1),95–101(1992)

12F.N.Abu-Khzametal.

3.Akutsu,T.:Apolynomialtimealgorithmforfindingalargestcommonsubgraph ofalmosttreesofboundeddegree.IEICETrans.Fundam.Electron.Commun. Comput.Sci. 76(9),1488–1493(1993)

4.Bodlaender,H.L.,Jansen,B.M.P.,Kratsch,S.:Kernelizationlowerboundsby cross-composition.SIAMJ.DiscreteMath. 28(1),277–305(2014)

5.Cesati,M.:Theturingwaytoparameterizedcomplexity.J.Comput.Syst.Sci. 67(4),654–685(2003)

6.Chen,J.,Kanj,I.A.,Xia,G.:Improvedupperboundsforvertexcover.Theor. Comput.Sci. 411(40–42),3736–3756(2010)

7.Downey,R.G.,Fellows,M.R.:FundamentalsofParameterizedComplexity.Texts inComputerScience.Springer,London(2013)

8.Fellows,M.R.,Jansen,B.M.P.,Rosamond,F.A.:Towardsfullymultivariatealgorithmics:parameterecologyandthedeconstructionofcomputationalcomplexity. Eur.J.Comb. 34(3),541–566(2013)

9.Flum,J.,Grohe,M.:Fixed-parametertractability,definability,andmodelchecking.SIAMJ.Comput. 31(1),113–145(2001)

10.Garey,M.R.,Johnson,D.S.:ComputersandIntractability:AGuidetotheTheory ofNP-Completeness.W.H.Freeman,SanFrancisco(1979)

11.Grindley,H.M.,Artymiuk,P.J.,Rice,D.W.,Willett,P.:Identificationoftertiary structureresemblanceinproteinsusingamaximalcommonsubgraphisomorphism algorithm.J.Mol.Biol. 229(3),707–721(1993)

12.Koch,I.,Lengauer,T.,Wanke,E.:Analgorithmforfindingmaximalcommon subtopologiesinasetofproteinstructures.J.Comput.Biol. 3(2),289–306(1996)

13.McGregor,J.,Willett,P.:Useofamaximalcommonsubgraphalgorithminthe automaticidentificationoftheostensiblebondchangesoccurringinchemicalreactions.J.Chem.Inf.Comput.Sci 21,137–140(1981)

14.Moser,H.,Sikdar,S.:Theparameterizedcomplexityoftheinducedmatchingproblem.DiscreteAppl.Math. 157(4),715–727(2009)

15.Niedermeier,R.:InvitationtoFixedParameterAlgorithms.LectureSeriesin MathematicsandItsApplications.OxfordUniversityPress,Oxford(2006)

16.Raymond,J.W.,Willett,P.:Maximumcommonsubgraphisomorphismalgorithms forthematchingofchemicalstructures.J.Comput.AidedMol.Des. 16,521–533 (2002)

17.Yamaguchi,A.,Aoki,K.F.,Mamitsuka,H.:Findingthemaximumcommonsubgraphofapartialk-treeandagraphwithapolynomiallyboundednumberof spanningtrees.Inf.Process.Lett. 92(2),57–63(2004)

ApproximationandHardnessResults fortheMaximumEdgesinTransitive ClosureProblem

1 ,GuillaumeBlin2,3 ,andAlexandruPopa4(B)

1 Max-Planck-InstitutF¨urInformatik,Saarbr¨ucken,Germany anna@mpi-inf.mpg.de

2 LaBRI,UMR5800,UniversityofBordeaux,33400Talence,France

3 CNRS,LaBRI,UMR5800,33400Talence,France

guillaume.blin@labri.fr

4 FacultyofInformatics,MasarykUniversity,Brno,CzechRepublic popa@fi.muni.cz

Abstract. Inthispaperwestudythefollowingproblem,namedMaximumEdgesinTransitiveClosure,whichhasapplicationsincomputationalbiology.Givenasimple,undirectedgraph G =(V,E )anda coloringofthevertices,removeacollectionofedgesfromthegraphsuch thateachconnectedcomponentis colorful (i.e.,itdoesnotcontaintwo identicallycoloredvertices)andthenumberofedgesinthetransitive closureofthegraphismaximized.

TheproblemisknowntobeAPX-hard,andnoapproximationalgorithmsareknownforit.Weimprovethehardnessresultbyshowingthat theproblemisNP-hardtoapproximatewithinafactorof |V |1/3 ε ,for anyconstant ε> 0.Additionally,weshowthattheproblemisAPXhardalreadyforthecasewhenthenumberofvertexcolorsequals3.We complementtheseresultsbyshowingthefirstapproximationalgorithm fortheproblem,withapproximationfactor √2 OPT.

1Introduction

TheMaximumEdgesinTransitiveClosureproblemweconsiderinthispaper belongstotheframeworkofcolorfulcomponentsproblems.

Colorfulcomponentsframework: Givenasimple,undirectedgraph G = (V,E )andacoloring σ : V → C oftheverticeswithcolorsfromagivenset C , removeacollectionofedges E ⊆ E from G suchthateachconnectedcomponent intheresultinggraph G =(V,E \E )isa colorfulcomponent (i.e.,itdoesnot containtwoidenticallycoloredvertices).Wewantthegraph G tobeoptimal accordingtosomefixed optimizationmeasure

Inourproblem,theoptimizationmeasureisthenumberofedgesinthetransitiveclosure.Foragraphconsistingof k connectedcomponents,eachcontaining respectively a1 ,a2 ,...,ak vertices,thenumberofedgesinthetransitiveclosure ofthegraphis

c SpringerInternationalPublishingSwitzerland2015 J.Kratochv´ıletal.(Eds.):IWOCA2014,LNCS8986,pp.13–23,2015. DOI:10.1007/978-3-319-19315-1 2

MaximumEdgesinTransitiveClosure(MEC): Givenasimple,undirectedgraph G =(V,E )andacoloring σ : V → C ofthevertices,remove acollectionofedges E ⊆ E from G suchthateachconnectedcomponentin theresultinggraph G =(V,E \E )iscolorful,andthenumberofedgesinthe transitiveclosureof G ismaximum.

Motivation. Thecolorfulcomponentsframeworkismotivatedbyapplicationsin comparativegenomics[8, 10],whichisafundamentalbranchofbioinformatics studyingtherelationshipofthegenomestructurebetweendifferentbiological species.Oneofthekeyproblemsinthisarea,themultiplealignmentofgene orders,canbecapturedasagraphtheoreticalproblem,usingthecolorfulcomponentsframework,wherethecolorfulgraphsrepresentsimilarityrelationships betweengenesfromdifferenthomologousgenefamilies.Apartitionintocolorful componentscorrespondsthentoapartitionofgenesintoorthologysets,where anytwogenesfromthesamegenomebelongtodifferentorthologysets.Werefer thereaderto[10]foramoredetaileddescriptionoftheconnectionbetweenthe multiplealignmentofgeneordersandthegraphtheoreticframeworkconsidered.

Theunderstandingoforthologousgenesoftwodifferentgenomesasoriginatingfromasinglegeneinthemostrecentcommonancestorofthetwospecies leadstotransitivityasapropertyoftheorthologyrelation.Thismotivatesthe studyofMEC(see[10]formoredetails,andforadiscussionwhyMECyields goodresultsinpractice).

RelatedWork. TheMaximumEdgesinTransitiveClosureproblemhasbeen introducedbyZhengetal.[10].Theypresentheuristicalgorithmsfortheproblem,withoutgivinganyworst-caseapproximationguarantee.TheyalsoconjecturetheproblemtobeNP-hard.AdamaszekandPopa[1]provethatMECis APX-hard,eveninthecaseof4vertexcolors.

ThecolorfulcomponentsframeworkappearedfirstinthepaperbyZheng etal.[10]andhasbeenformallydefinedbyAdamaszekandPopa[1],although problemswhichfitintothisframeworkhavealreadybeenstudiedearlier.We nowsummarizeknownresultsfortheseproblems.

IntheproblemnamedeitherColorfulComponents[3, 4]orMinimumOrthogonalPartition[5, 10],theobjectivefunctionistominimizethenumberofedges removedfrom G toobtainagraphinwhichallconnectedcomponentsarecolorful.Bruckneretal.[4]showthattheproblemis NP -hardforthreeormorecolors andtheystudyfixed-parameteralgorithmsfortheproblem.Their NP -hardness reductioncanbemodifiedslightlytoshowtheAPX-hardnessoftheproblem (see[1]).Zhengetal.[10]andBruckneretal.[3]studyheuristicapproaches fortheproblem,andHeetal.[5]presentanapproximationalgorithmforsome specialcaseoftheproblem.Asthegeneralproblemisaspecialcaseofthe MinimumMulti-MultiwayCutproblem,itadmitsa O (log |C |)approximation algorithm[2].

Zhengetal.[10]introducetheMinimumSingletonVerticesproblem(MSV), wherethegoalistominimizethenumberofisolatedverticesintheresulting graph.Zhengetal.[10]presentheuristicalgorithmsfortheproblem,without givinganyworst-caseapproximationguarantee.Theyalsoconjecturedthatthe problemisNP-hard.Tremblay-SavardandSwenson[9]consideraMaximum OrthogonalEdgeCoverproblem(MAX-OREC),whichisadualproblemto MSV.There,thegoalistocoveramaximumnumberofverticesofagraph usingvertex-disjoint,non-singletonconnectedcolorfulsubgraphs.In[9],a2/3approximationalgorithmforMAX-ORECispresented.AdamaszekandPopa[1] provethatMSV(andthereforealsoMAX-OREC)canbesolvedexactlyinpolynomialtime,thusdisprovingtheconjecturein[10].

AdamaszekandPopa[1]introduceanotherproblem,termedMinimumColorfulComponents,inwhichthegoalistodeleteasubsetofedgessuchthat theresultinggraphhasonlycolorfulcomponentsandthenumberofconnected componentsisminimized.Theyshowthatthisproblemcannotbeapproximated withinafactorof |V |1/14 ε unless P = NP ,andwithinafactor |V |1/2 ε unless ZPP = NP .

OurResults. InthispaperweimprovethehardnessresultsfortheMECproblem, andwepresentthefirstapproximationalgorithm.

First,weshowthatMECisAPX-hardevenforthecasewhen |C | =3.This settlesthecomplexityoftheproblemwhenthenumberofcolorsisaconstant, asfor |C | =2theMECproblemcanbesolvedexactlyinpolynomialtimeby usingamaximummatchingalgorithm.Ourproofisviaareductionfromthe MaximumBounded3-DimensionalMatchingproblem(Max3-DM-3).

Forthegeneralcase,whenthenumberofcolorsisarbitrary,weshowthat MECisNP-hardtoapproximatewithinafactorof |V |1/3 ε foranyconstant ε> 0.Thisresultholdseveniftheinputgraphisatreeandeachcolorappears atmosttwiceinthegraph.WeusethesamereductionfromtheIndependentSet asRizziandSikoraforprovinghardnessofapproximationoftheGraphMotif problem[7].

Wealsoshowthefirstpolynomial-timeapproximationalgorithmforMEC, whichhasaratioof √2 · OPT.Weusetheexactpolynomialtimealgorithmfor theMinimumSingletonVerticesproblem[1]toobtainapartitionintocolorful componentsandthenweshowthatthispartitionhasabigenoughnumberof edgesinthetransitiveclosure.

2APX-hardnessofMECfor |C | =3

Inthissection,weprovethattheMECproblemrestrictedtoinstancesusing only3colorsisAPX-hard.TheproofisviaareductionfromtheMaximum Bounded3-DimensionalMatchingproblem.Thisresultstrengthenstheonepresentedin[1],whichholdsforprobleminstancesusing4colors.

Beforewegivethereduction,wefirststatethedefinitionofMax3-DM-3and theknownhardnessresultforit.

MaximumBounded3-DimensionalMatching(Max3-DM-3): Theinput consistsofpairwisedisjointsets X , Y , Z andacollection T ⊆ X × Y × Z of

16A.Adamaszeketal.

triplessuchthateachelementfrom X , Y and Z occursinatleastoneandat mostthreetriplesin T .Theaimistofindafeasiblesubsetoftriples T ⊆ T (i.e.,notwoelementsof T agreeonanycoordinate)ofmaximumcardinality.

Theorem1(Theorem4.4in[6],Rephrased). Thereexistsaconstant ε> 0 suchthatitisNP-hardtodistinguishbetweentheinstancesofMax3-DM-3with thefollowingproperties:

1.Thereisafeasiblecollectionoftriples T ⊆ T suchthateveryelementof X , Y and Z belongstosometriplein T

2.Foreveryfeasiblecollectionoftriples T ⊆ T lessthan (1 ε) fractionof elementsfrom X ∪ Y ∪ Z belongtosometripleof T .

Withoutlossofgeneralitywecanassumethat |X | = |Y | = |Z | = n,sinceif |X |, |Y | and |Z | aredifferent,thenthecase1ofTheorem 1 cannothold.Also,define N = |T |.Itholdsthat N ≤ 3n,sinceeachelementof X ∪ Y ∪ Z appearsin atmostthreetriples.Intherestofthesection,weuseOPT3DM todenotethe sizeofanoptimalsolutionofaMax3-DM-3instance(theinstancewereferto willalwaysbeclearfromthecontext),andOPTMEC todenotethevalueofan optimalsolution(i.e.,thenumberofedgesinthetransitiveclosureofthegraph) oftheMECinstanceobtainedviathereduction.

Reduction. Givenaninstance(X,Y,Z,T )ofMax3-DM-3,wecreatean instance(G =(V,E ),σ )oftheMECprobleminthefollowingway.SeeFig. 1 for apartialillustration.Wecreatethesetofvertices V asfollows.

1.Foreachtriple tj ∈ T ,weaddsixvertices {tX j ,tY j ,tZ j ,tXY j ,tXZ j ,tYZ j }.

2.Foreachelement xi ∈ X (resp. yi ∈ Y and zi ∈ Z ),weaddacorresponding vertex xi (resp. yi and zi ).

Wehavethat |V | =6 ·|T | + |X | + |Y | + |Z | =6N +3n.Letusnowdefinethe coloring σ : V → C oftheverticesusingthesetofcolors C = {1, 2, 3}.

1.Forany1 ≤ i ≤ n and1 ≤ j ≤ N , σ (xi )= σ (tXY j )= σ (tZ j )=1.

2.Forany1 ≤ i ≤ n and1 ≤ j ≤ N , σ (yi )= σ (tYZ j )= σ (tX j )=2.

3.Forany1 ≤ i ≤ n and1 ≤ j ≤ N , σ (zi )= σ (tXZ j )= σ (tY j )=3. Finally,letusdefinethecollectionofedges E

1.Foreach1 ≤ j ≤ N ,eachof {tX j ,tXY j ,tXZ j }, {tY j ,tXY j ,tYZ j }, {tZ j ,tXZ j ,tYZ j } formsacliqueofsizethree.

2.Foreach1 ≤ i ≤ n and1 ≤ j ≤ N ,if xi (resp. yi and zi )appearsin tj , connect xi (resp. yi and zi )to tX j (resp. tY j and tZ j ).

Analysis. Informally,weshowthataninstanceofMax3-DM-3whereallthe vertices X ∪ Y ∪ Z canbecoveredbyafeasiblecollectionoftriples T correspondstoaninstanceofMECwithalargeoptimalvalue,i.e.,thegraphcan bepartitionedintocolorfulcomponentsinducingalargetransitiveclosure.On theotherhand,weshowthataninstanceofMax3-DM-3wherenomorethan (1 ε)fractionofthevertices X ∪ Y ∪ Z canbecoveredbyanyfeasiblesetof triplescorrespondstoaninstanceofMECwithamuchsmalleroptimalvalue. Wenowanalyzebothcases.

Fig.1. Asubgraphcorrespondingtoatriple tj =(xi ,yk ,zl ).Colorsoftheverticesare denotedusingthelinestyles:solid,dottedanddashedlinesrespectivelycorresponds tocolors1,2and3.

Lemma1. Let (X,Y,Z,T ) beaninstanceofMax3-DM-3whereOPT3DM = n, i.e.,wherealltheverticesof X ∪ Y ∪ Z canbecoveredbyafeasiblecollection oftriples.ThenforthecorrespondinginstanceofMEC,wehaveOPTMEC ≥ 6N +3n

Proof. ThecolorfulcomponentsoftheMECinstanceareconstructedasfollows. Foreachtriple tj ∈ T (thereare n ofthem),weaddthreecolorfulcomponents, eachcomponentconsistingofthreevertices.Givenatriple tj =(xi ,yk ,zl ),the colorfulcomponentsare {xi ,tX j ,tXZ j }, {yk ,tY j ,tXY j } and {zl ,tZ j ,tYZ j }.Foreach triple tj ∈ T \ T (thereare N n ofthem),wecreatetwocolorfulcomponents, eachconsistingofthreevertices: {tX j ,tXZ j ,tZ j } and {tXY j ,tY j ,tYZ j }.SeeFig. 2 for anillustration.

As T isafeasiblecollectionoftriples,thatisasetoftriplessuchthatnotwo elementsagreeonanycoordinate,weobtainafeasiblepartitionofthegraphinto colorfulcomponents.Clearly,thetotalnumberofedgesinthetransitiveclosure equals9n +6(N n)=6N +3n,sinceeachofthe n triplesin T inducesthree colorfulcomponentsofsizethreeandeachofthe N n othertriplesinduces twocolorfulcomponentsofsizethree.

Lemma2. Let (X,Y,Z,T ) beaninstanceofMax3-DM-3whereOPT3DM < (1 ε)n,i.e.,whereeveryfeasiblecollectionoftriplescoverslessthana (1 ε) wofvertices X ∪ Y ∪ Z .Then,forthecorrespondinginstanceofMEC,wehave OPTMEC < 6N +3n(1 ε/2).

Proof. Let(G =(V,E ),σ )betheinstanceoftheMECproblemcorresponding toaninstanceofMax3-DM-3asdefinedinthelemmastatement.Foranytriple tj =(xi ,yk ,zl ) ∈ T ,let Gtj beasubgraphof G inducedbythefollowingsetof vertices {xi ,yk ,zl ,tX j ,tY j ,tZ j ,tXY j ,tXZ j ,tYZ j },asshowninFig. 1

Letusfixanoptimalsolution S fortheMECproblemfor(G,σ ).Thissolutiondefinesapartition Γ of G intocolorfulcomponents.First,noticethateach colorfulcomponentiscontainedwithinsomesubgraph Gt .Indeed,byconstruction,theonlyverticeswhichbelongtomultiplesubgraphs Gtj arethevertices

Another random document with no related content on Scribd:

De La Pasture, E. E. M. Tension. (D ’20)

Dodd, L. W. Book of Susan. (O ’20)

Fuller, H: B. Bertram Cope’s year. (My ’20)

Gale, Z. Miss Lulu Bett. (My ’20)

Galsworthy, J: In chancery. (N ’20)

Gibbon, M. M. Jan. (Ja ’21)

Goldring, D. Margot’s progress. (Jl ’20)

Hall, H. S. Steel preferred. (N ’20)

Hawkins, A. H. Lucinda. (Ja ’21)

Hewlett, M. H: Mainwaring. (N ’20)

Holding, E. S. Invincible Minnie (My ’20)

Holdsworth, E. Taming of Nan. (Ap ’20)

Johnston, H. H. Mrs Warren’s daughter. (Jl ’20)

MacGrath, H. Man with three names. (Mr ’20)

McKenna, S. Lady Lilith. (N ’20)

Maugham, W: S. Mrs Craddock. (Ag ’20)

Peake, C. M. A. Eli of the downs. (O ’20)

Rickard, L. Cathy Rossiter. (Ap ’20)

Sadler, M. Anchor. (My ’20)

Sharp, H. M. Pawn in pawn. (Jl ’20)

Singmaster, E. Basil Everman. (Ap ’20)

Swinnerton, F. A. September. (Mr ’20)

Underhill, R. M. White moth. (Ja ’21)

Weyl, M. Happy woman. (Ap ’20)

Wyllarde, D. Temperament. (S ’20)

Cheerful stories

Abbott, J. L. Happy house. (Jl ’20)

Adams, S: H. Wanted: a husband. (Je ’20)

Champion, J. Sunshine in Underwood. (Jl ’20)

Clancy, L. B. Christine of the young heart. (D ’20)

Kelley, E. M. Outside inn. (N ’20)

Lutz, G. L. H. Cloudy Jewel. (D ’20)

Miller, A. Beauty and the bolshevist. (D ’20)

Sterrett, F. R. Nancy goes to town. (Ja ’21)

Street, J. L. Sunbeams, Inc. (D ’20)

Tompkins, J. W. Joanna builds a nest. (D ’20)

Children, Stories about

Boyer, W. S. Johnnie Kelly. (N ’20)

Butler, E. P. Swatty. (Ap ’20)

Croy, H. Turkey Bowman. (N ’20)

France, A., pseud. Little Pierre. (F ’21)

Galsworthy, J: Awakening. (Ja ’21)

Gatlin, D. Missy. (D ’20)

McKishnie, A. P. Son of courage. (D ’20)

Masters, E. L. Mitch Miller. (D ’20)

Porter, E. Mary Marie. (Je ’20)

Vorse, M. M. Growing up. (S ’20)

Christmas stories

Rinehart, M. Truce of God. (F ’21)

Court-room scenes

Hill, F: T. Tales out of court. (D ’20)

Crime and criminals

Gregory, J. Ladyfingers. (Je ’20)

Leverage, H: Where dead men walk. (Ag ’20)

Packard, F. L. From now on. (Mr ’20)

Packard, F. L. White Moll. (Jl ’20)

Roche, A. S. Uneasy street. (Mr ’20)

Rowland, H: C. Duds. (Mr ’20)

Williamson, C: N. and A. M. Second latchkey. (Je ’20)

Detective stories

See Mystery stories

Divorce

Maxwell, W: B. For better, for worse. (N ’20)

Economic problems

Cadmus and Harmonia, pseuds. Island of sheep. (Ap ’20)

Enterprise

Sullivan, A. Rapids. (S ’20)

European war

Abdullah, A. Man on horseback. (Ap ’20)

Battersby, H: F. P. Edge of doom. (Je ’20)

Bazin, R. F. N. M. Pierre and Joseph. (Je ’20)

Blasco Ibáñez, V. Enemies of women. (F ’21)

Cook, W. V: Grey fish. (S ’20)

Gibbs, G: F. Splendid outcast. (Ap ’20)

Gibbs, P. H. Wounded souls. (N ’20)

Herbert, A. P. Secret battle. (Mr ’20)

Jeffery, J. E. Side issues. (N ’20)

Kelly, T: H. What outfit, Buddy? (My ’20)

Latzko, A. Judgment of peace. (Ap ’20)

Mackay, H. G. Chill hours. (Ap ’20)

Mason, A. E: W. Summons. (D ’20)

Maurois, A. Silence of Colonel Bramble. (Jl ’20)

Mayran, C. Story of Gotton Connixloo. (Ja ’21)

Montague, M. P. England to America. (Je ’20)

Oppenheim, E: P. Devil’s paw. (D ’20)

Robinson, E. H. Maid of Mirabelle. (O ’20)

Sawyer, R. Leerie. (S ’20)

Sinclair, M. Romantic. (D ’20)

Zanella, N. By the waters of Fiume. (Jl ’20)

Belgium

Vane, G: Waters of strife. (S ’20)

Great Britain

Ayres, R. M. Richard Chatterton, V. C. (Jl ’20)

Baxter, A. B. Parts men play. (Ja ’21)

Benson, E: F: Robin Linnet. (Mr ’20)

Benson, S. Living alone. (My ’20)

Copplestone, B. Last of the Grenvilles. (Ap ’20)

Dawson, C. W: Little house. (N ’20)

Frankau, G. Peter Jameson. (My ’20)

Galsworthy, J: Tatterdemalion. (My ’20)

Hamilton, C. M. William—an Englishman. (Jl ’20)

Jenkin, A. M. N. End of a dream. (Jl ’20)

Jepson, E. Pollyooly dances. (Ag ’20)

Locke, W: J: House of Baltazar. (Mr ’20)

Maxwell, W: B. Glamour. (Ap ’20)

Onions, B. Bridge of kisses. (D ’20)

United States

Austin, M. No. 26 Jayne street. (Je ’20)

Montague, M. P. Uncle Sam of Freedom Ridge. (S ’20)

Poole, E. Blind. (D ’20)

Schem, L. C. Hyphen. (Ja ’21)

Sherwood, M. P. World to mend. (N ’20)

Fairy tales

Tarn, W: W. Treasure of the isle of mist. (Jl ’20)

High cost of living

Dodge, H: I. Skinner makes it fashionable. (Je ’20)

Historical novels

Canada

Wilson, M. Forging of the pikes. (My ’20)

Cape Colony

Juta, R. Cape Currey. (S ’20)

Discovery & exploration (17th century)

Forbes, G: Adventures in southern seas. (Ja ’21)

Egypt

Couperus, L: M. A. Tour. (Jl ’20)

Haggard, H: R. Ancient Allan. (Je ’20)

England

Hudson, W: H: Dead Man’s Plack, and An old thorn. (F ’21)

Locke, G. E. Ronald o ’ the moors. (O ’20)

McCarthy, J. H. Henry Elizabeth. (S ’20)

MacDonald, G. North door. (O ’20)

McFadden, G. V. Preventive man. (Jl ’20)

Orczy, E. His Majesty’s well-beloved. (Ap ’20)

France

Bailey, H: C. Barry Leroy. (Je ’20)

Brooks, C: S. Luca Sarto. (Ap ’20)

Iceland

Hewlett, M. H: Light heart. (Jl ’20)

Hewlett, M. H: Outlaw. (Ap ’20)

Italy

Drummond, H. Maker of saints. (S ’20)

Vorse, M. M. Ninth man. (O ’20)

Jerusalem

Mapu, A. Sorrows of Noma. (Ap ’20)

Middle ages

Ross, R. Revels of Orsera. (F ’21)

United States

Daviess, M. T. Matrix. (Ap ’20)

Fox, J:, jr. Erskine Dale, pioneer. (D ’20)

Gregg, F. M. Founding of a nation. (O ’20)

Lynn, M. Free soil. (Ja ’21)

Maule, M. K. Prairie-schooner princess. (N ’20)

Peadexter, H. Red belts. (Mr ’20)

Shafer, D. C. Barent Creighton. (S ’20)

White, S. E: Rose dawn. (D ’20)

Horse racing

Richards, G. Double life. (F ’21)

Yates, L. B. Autobiography of a race horse. (Ag ’20)

Humor and satire

Ashford, D. Daisy Ashford: her book. (S ’20)

Beerbohm, M. Seven men. (D ’20)

Cannan, G. Windmills. (Ag ’20)

Darlington, W. A. Alf’s button. (S ’20)

Dickson, H. Old Reliable in Africa. (D ’20)

Dodge, H: I. Skinner makes it fashionable. (Je ’20)

Huxley, A. L. Limbo. (Ag ’20)

Irwin, W. A. Trimmed with red. (Ag ’20)

Kelland, C. B. Efficiency Edgar. (Jl ’20)

Leacock, S. B. Winsome Winnie. (Ja ’21)

Mackall, L. Scrambled eggs. (D ’20)

Mackenzie, C. Poor relations. (Mr ’20)

Marshall, R. Enchanted golf clubs. (Mr ’20)

Putnam, N. It pays to smile. (F ’21)

Street, J. L. Sunbeams, Inc. (D ’20)

Ullman, A. E: “Line’s busy.” (N ’20)

Immigrants in America

Cournos, J: Mask. (Mr ’20)

Yezierska, A. Hungry hearts. (D ’20)

Industrial conditions

Hall, H. S. Steel preferred. (N ’20)

Jewish life

Pearl, B. Sarah and her daughter. (Jl ’20)

Journalism

Dodge, L: Whispers. (Je ’20)

George, W. L. Caliban. (O ’20)

Macaulay, R. Potterism. (D ’20)

Williams, S. C. Unconscious crusader. (Je ’20)

Williams, W. W. Goshen street. (N ’20)

Law and lawyers

Train, A. C. Tutt and Mr Tutt. (Je ’20)

Legends and folktales

Shedlock, M. L. Eastern stories and legends. (D ’20)

Flemish

Coster, C: T. H. de. Flemish legends. (N ’20)

Irish

MacManus, S. Top o ’ the mornin’. (N ’20)

Japanese

Ozaki, Y. T. Romances of old Japan. (Mr ’20)

Letters (stories in letter form)

Lucas, E: V. Verena in the midst. (N ’20)

Ridsdale, K. Gate of fulfillment. (Je ’20)

Locality, Novels of

Adirondacks

Longstreth, T: M. Mac of Placid. (O ’20)

Africa

Battersby, H: F. P. Edge of doom. (Je ’20)

Benoit, P. Atlantida. (S ’20)

Burroughs, E. R. Tarzan the untamed. (O ’20)

Dickson, H. Old Reliable in Africa. (D ’20)

Stockley, C. Pink gods and blue demons. (Ag ’20)

Alaska

Rutzebeck, H. Alaska man ’ s luck. (Ja ’21)

Arizona

Noyes, A. Beyond the desert. (D ’20)

Armenia

Inchbold, A. C. Love and the crescent. (S ’20)

Boston

Brown, A. Wind between the worlds. (S ’20)

Brazil

Elliott, L. W. Black gold. (F ’21)

Graça Aranha, J. P. da. Canaan. (Ap ’20)

California

Chase, J. S. Penance of Magdalena. (Ag ’20)

Porter, R. N. Girl from Four Corners. (My ’20)

White, S. E: Rose dawn. (D ’20)

Canada

Binns, O. Mating in the wilds. (S ’20)

Cody, H. A. Glen of the high north. (O ’20)

Cullum, R. Heart of Unaga. (N ’20)

Curwood, J. O. Valley of silent men. (O ’20)

Durkin, D. Heart of Cherry McBain. (N ’20)

Footner, H. Fur bringers. (D ’20)

Gibbon, J: M. Conquering hero. (N ’20)

Kendall, R. S. Luck of the mounted. (F ’21)

McKowan, E. Graydon of the Windermere. (F ’21)

Pinkerton, K. S. and R. E. Long traverse. (S ’20)

Pinkerton, K. S. and R. E. Penitentiary Post. (Ag ’20)

Sidgwick, C., and Garstin, C. Black knight. (O ’20)

Stringer, A. J: A. Prairie mother. (N ’20)

White, S: A. Ambush. (N ’20)

Cape Cod

Cooper, J. A. Tobias o ’ the light. (Ag ’20)

Lincoln, J. C. Portygee. (Je ’20)

Chicago

Borden, T. M. Romantic woman. (My ’20)

Webster, H: K. Mary Wollaston. (D ’20)

China

Lamb, H. Marching sands. (My ’20)

Merwin, S: Hills of Han. (D ’20)

Miln, L. J. Mr Wu. (Ap ’20)

Weale, B. L. P. Wang the ninth. (Ja ’21)

Connecticut

Minnigerode, M. Laughing house. (D ’20)

Egypt

Bradley, M. Fortieth door. (Ap ’20)

Weigall, A. E: P. B. Madeline of the desert. (D ’20)

England

Anstruther, E. H. A. Husband. (Jl ’20)

Baxter, A. B. Parts men play. (Ja ’21)

Johnston, H. H. Mrs Warren’s daughter. (Jl ’20)

O’Sullivan, Mrs D. Mr Dimock. (F ’21)

England (London)

Brunner, E. Celia and her friends. (S ’20)

Brunner, E. Celia once again. (S ’20)

Cannan, G. Time and eternity. (My ’20)

Desmond, S. Passion. (Jl ’20)

Ervine, S. G. Foolish lovers. (Jl ’20)

Pryde, A. Marqueray’s duel. (Jl ’20)

Woolf, V. Night and day. (N ’20)

England (Oxford)

Morley, C. D. Kathleen. (Ap ’20)

England (provincial and rural)

Ayscough, J:, pseud. Abbotscourt. (Je ’20)

Benson, E: F: “Queen Lucia.” (S ’20)

Blundell, M. E. Beck of Beckford. (F ’21)

Brighouse, H. Marbeck inn. (Ap ’20)

Buckrose, J. E., pseud. Young hearts. (S ’20)

Easton, D. Golden bird. (S ’20)

Galsworthy, J: In chancery. (N ’20)

Gambier, K. Girl on the hilltop. (Ag ’20)

Hicks Beach, S. E. Shuttered doors. (Je ’20)

Hocking, J. Passion for life. (Jl ’20)

Kaye-Smith, S. Tamarisk town. (Jl ’20)

Lorimer, N. O. With other eyes. (O ’20)

Marshall, A. Many Junes. (Je ’20)

Merrick, H. Mary-girl. (S ’20)

Oldmeadow, E. J. Coggin. (Mr ’20)

Pedler, M. House of dreams-come-true. (S ’20)

Phillpotts, E. Miser’s money. (Je ’20)

Rainsford, W. H. That girl March. (F ’21)

Stevenson, G: Benjy. (Je ’20)

Swinnerton, F. A. September. (Mr ’20)

Thurston, E. T. Sheepskins and grey russet. (Je ’20)

Turner, J: H. Place in the world. (My ’20)

Vachell, H. A. Whitewash. (Jl ’20)

Ward, M. A. Harvest. (My ’20)

Whitham, G. I. St John of Honeylea. (Je ’20)

Far East

Rideout, H: M. Foot-path way. (Je ’20)

Florida

Oyen, H: Plunderer. (Je ’20)

Germany

Henry, S. O. Villa Elsa. (Je ’20)

India

Comfort, W. L., and Dost, Z. K. Son of power. (D ’20)

Lowis, C. C. Four blind mice. (D ’20)

Mundy, T. Told in the East. (F ’21)

Savi, E. W. When the blood burns. (D ’20)

Tracy, L: Sirdar’s sabre. (D ’20)

Ireland

Carleton, W: Stories of Irish life. (Mr ’20)

Ervine, S. G. Foolish lovers. (Jl ’20)

Galway, C. Towards the dawn. (D ’20)

Hannay, J. O. Up, the rebels! (Mr ’20)

Hinkson, K. Love of brothers. (D ’20)

MacFarlan, A. Inscrutable lovers. (Ap ’20)

MacGill, P. Maureen. (Je ’20)

MacNamara, B. Clanking of chains. (My ’20)

O’Duffy, E. Wasted island. (N ’20)

O’Kelly, S. Golden barque; and The weaver ’ s grave. (D ’20)

O’Riordan, C. O. Adam of Dublin. (Ja ’21)

Page, G. Paddy-the-next-best-thing. (F ’21)

Poore, I. M. Rachel Fitzpatrick. (S ’20)

Somerville, E. A. O., and Martin, V. F.

Mount Music. (Je ’20)

Italy

Forster, E: M. Where angels fear to tread. (Ap ’20)

Hudson, S. Richard Kurt. (Jl ’20)

Kennard, J. S. Memmo. (Ja ’21)

Lombardi, C. Cry of youth. (Jl ’20)

Kansas

Lynn, M. Free soil. (Ja ’21)

Kentucky

Buck, C: N. Tempering. (Je ’20)

Louisiana

Perry, S. G: Palmetto. (N ’20)

Mexico

Grogan, G. William Pollok, and other tales. (Je ’20)

Minnesota

Lewis, S. Main street. (N ’20)

Mississippi river

Spears, R. S. River prophet. (Ag ’20)

Nebraska

Maule, M. K. Prairie-schooner princess. (N ’20)

New England

Bassett, S. W. Wall between. (D ’20)

Brown, A. Homespun and gold. (Ja ’21)

Gerould, G. H. Youth in Harley. (O ’20)

Howells, W: D. Vacation of the Kelwyns. (N ’20)

Spofford, H. E. Elder’s people. (My ’20)

Williams, W. W. Goshen street. (N ’20)

New York (city)

Austin, M. 26 Jayne street. (Je ’20)

Bullard, A. Stranger. (Jl ’20)

Chamberlain, G: A. Taxi. (Ag ’20)

Corbett, E. F. Puritan and pagan. (Ja ’21)

Duganne, P. Prologue. (N ’20)

Forman, H: J. Fire of youth. (Ap ’20)

Frank, W. D: Dark mother. (D ’20)

Hudson, H:, 2d, pseud. Spendthrift town. (D ’20)

Hughes, R. What’s the world coming to? (Jl ’20)

Pearl, B. Sarah and her daughter. (Jl ’20)

Raine, W: M. Big-town round-up. (F ’21)

Spadoni, A. Swing of the pendulum. (Ap ’20)

Wharton, E. N. Age of innocence. (N ’20)

New York (state)

White, G. Storm country Polly. (Jl ’20)

New Zealand

Mander, J. Story of a New Zealand river. (My ’20)

Ohio

Anderson, S. Poor white. (D ’20)

Anderton, D. Cousin Sadie. (F ’21)

Paris

Audoux, M. Marie Claire’s workshop. (D ’20)

Pennsylvania

Deland, M. W. Old Chester secret. (D ’20)

Martin, H. R. Schoolmaster of Hessville. (N ’20)

Morris, H. S. Hannah Bye. (Ja ’20)

Myers, A. B. Patchwork. (My ’20)

Philippine islands

Martin, M. W. Green god’s pavilion. (O ’20)

Rome

Richardson, N. Pagan fire. (Ja ’21)

San Francisco

Dobie, C: C. Blood red dawn. (Jl ’20)

Spadoni, A. Swing of the pendulum. (Ap ’20)

Scotland

Douglas, O. E. Penny plain. (D ’20)

Niven, F: J: Tale that is told. (D ’20)

Watson, R. Stronger than his sea. (F ’21)

South Africa

Dell, E. M. Top of the world. (N ’20)

Millin, S. G. Dark river. (D ’20)

Young, F. E. M. Almonds of life. (O ’20)

South Carolina

Oemler, M. Purple heights. (N ’20)

South seas

Barbour, R. H:, and Holt, H. P. Joan of the island. (Ag ’20)

Conrad, J. Rescue. (Je ’20)

Grimshaw, B. Terrible island. (D ’20)

Switzerland

In the mountains. (N ’20)

United States (middlewestern)

Dell, F. Moon-calf. (D ’20)

Howe, E. W. Anthology of another town. (Ja ’21)

Watts, M. S. Noonmark. (F ’21)

United States (northwestern)

Sinclair, B. W: Poor man ’ s rock. (D ’20)

United States (southern)

Griffiths, G. Lure of the manor. (Ag ’20)

Hoffman, M. E. Lindy Loyd. (O ’20)

Olmstead, F. Stafford’s Island. (Je ’20)

Ragsdale, L. Next-besters. (S ’20)

Sampson, E. S. Mammy’s white folks. (S ’20)

Turner, G: K. Hagar’s hoard. (N ’20)

United States (southwestern)

Barry, R: H. Fruit of the desert. (S ’20)

Bennet, R. A. Bloom of cactus. (Mr ’20)

Dunn, J. A. E. Dead man ’ s gold. (S ’20)

Gregory, J. Man to man. (D ’20)

Hooker, F. C. Long dim trail. (N ’20)

Shedd, G: C. Iron furrow. (Ap ’20)

White, S. E: Killer. (Jl ’20)

United States (western)

Abbott, K. Wine o ’ the winds. (Ag ’20)

Bower, B. M., pseud. Quirt. (Jl ’20)

Brand, M. Trailin’! (Jl ’20)

Burt, K. N. Hidden Creek. (O ’20)

Coolidge, D. Wunpost. (N ’20)

Croy, H. Turkey Bowman. (N ’20)

Dorrance, E. A. and J. F. Glory rides the range. (Jl ’20)

Dunn, J. A. E. Turquoise Cañon. (Ap ’20)

Grey, Z. Man of the forest. (Mr ’20)

Hendryx, J. B. Gold girl. (Ag ’20)

Lynde, F. Girl, a horse and a dog. (S ’20)

Marshall, E. Voice of the pack. (Je ’20)

Raine, W: M. Oh, you Tex! (Ag ’20)

Richards, C. E. Tenderfoot bride. (F ’21)

Ritchie, R. W. Trails to Two Moons. (D ’20)

Titus, H. Last straw. (Jl ’20)

White, W: P. Hidden trails. (O ’20)

White, W: P. Lynch lawyers. (Jl ’20)

White, W: P. Paradise Bend. (F ’21)

Virginia

Bailey, T. Trumpeter swan. (D ’20)

Johnston, M. Sweet Rocket. (D ’20)

Wales

Evans, C. My neighbors. (My ’20)

Young, F. B. and E. B. Undergrowth. (Ja ’21)

Washington (state)

Kyne, P. B. Kindred of the dust. (Ag ’20)

West Virginia

Dillon, M. C. Farmer of Roaring Run. (Mr ’20)

Marriage

Borden, T. M. Romantic woman. (My ’20)

Byrne, D. Foolish matrons. (N ’20)

Couperus, L: M. A. Inevitable. (D ’20)

Edginton, H. M. Married life. (Ag ’20)

Hamilton, C. His friend and his wife. (Je ’20)

Harris, C. M. Happily married. (Ap ’20)

Kerr, S. Painted meadows. (Je ’20)

Widdemer, M. I’ve married Marjorie. (S ’20)

Wylie, I. A. R. Children of storm. (N ’20)

Moving picture stories

Luther, M. L. Presenting Jane McRae. (S ’20)

Witwer, H. C: Kid Scanlan. (Ag ’20)

Witwer, H. C: There’s no base like home. (Ag ’20)

Musicians

Close, E. Cherry Isle. (D ’20)

Leadbitter, E. Rain before seven. (Jl ’20)

Mix, J. I. At fame’s gateway. (My ’20)

Schauffler, R. H. Fiddler’s luck. (Jl ’20)

Mystery stories

Allison, W: Secret of the sea. (Ap ’20)

Allison, W: Turnstile of night. (D ’20)

Benoit, P. Secret spring. (Je ’20)

Biss, G. Door of the unreal. (Ja ’21)

Brebner, P. J. Ivory disc. (Ag ’20)

Brown, E. A. That affair at St Peter’s. (Jl ’20)

Burt, K. Red lady. (Je ’20)

Camp, C: W. Gray mask. (Jl ’20)

Capes, B. E: J. Skeleton key. (Ag ’20)

Turn static files into dynamic content formats.

Create a flipbook
Issuu converts static files into: digital portfolios, online yearbooks, online catalogs, digital photo albums and more. Sign up and create your flipbook.