Instant ebooks textbook A comparison process for mouse pairs john r. steel download all chapters

Page 1


A comparison process for mouse pairs John R. Steel

Visit to download the full and correct content document: https://ebookmass.com/product/a-comparison-process-for-mouse-pairs-john-r-steel/

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

Town Mouse and Country Mouse (Penguin Young Readers, Level 1) 1st Edition Arlene Wong

https://ebookmass.com/product/town-mouse-and-country-mousepenguin-young-readers-level-1-1st-edition-arlene-wong/

Random Process Analysis With R Marco Bittelli

https://ebookmass.com/product/random-process-analysis-with-rmarco-bittelli/

Extraction Techniques for Environmental Analysis John R. Dean

https://ebookmass.com/product/extraction-techniques-forenvironmental-analysis-john-r-dean/

Guidelines for Revalidating a Process Hazard Analysis

2nd Edition Ccps (Center For Chemical Process Safety)

https://ebookmass.com/product/guidelines-for-revalidating-aprocess-hazard-analysis-2nd-edition-ccps-center-for-chemicalprocess-safety/

Without a Trace Danielle Steel

https://ebookmass.com/product/without-a-trace-danielle-steel/

Corrosion of Steel in Concrete 3rd Edition John P. Broomfield

https://ebookmass.com/product/corrosion-of-steel-in-concrete-3rdedition-john-p-broomfield/

Pulse Crop Effects on Gut Microbial Populations, Intestinal Function, and Adiposity in a Mouse Model of Diet-Induced Obesity John N. Mcginley

https://ebookmass.com/product/pulse-crop-effects-on-gutmicrobial-populations-intestinal-function-and-adiposity-in-amouse-model-of-diet-induced-obesity-john-n-mcginley/

Handbook for Process Safety in Laboratories and Pilot Plants: A Risk■based Approach Ccps (Center

For Chemical Process Safety)

https://ebookmass.com/product/handbook-for-process-safety-inlaboratories-and-pilot-plants-a-risk%e2%80%90based-approach-ccpscenter-for-chemical-process-safety/

3D Printing: A Revolutionary Process for Industry Applications

Richard Sheng

https://ebookmass.com/product/3d-printing-a-revolutionaryprocess-for-industry-applications-richard-sheng/

CHAPTER 1.INTRODUCTION .........................................1

1.1.Largecardinalsandtheconsistencystrengthhierarchy.........1

1.2.Innermodeltheory.........................................3

1.3.Miceanditerationstrategies.................................5

1.4.HODinmodelsofdeterminacy..............................6

1.5.Leastbranchhodpairs......................................8

1.6.Comparisonandthemousepairorder........................9

1.7.Hodpaircapturing.........................................12

1.8.Constructingmousepairs...................................13

1.9.Thecomparisonargument...................................15

1.10.Planofthebook............................................17

CHAPTER 2.PRELIMINARIES ........................................19

2.1.Extendersandultrapowers..................................19

2.2.Pureextenderpremice......................................21

2.3.Projectaandcores..........................................24

2.4.Elementarityofmaps.......................................33

2.5. rΣk cofinalityandnearelementarity..........................40

2.6.Iterationtreesonpremice...................................49

2.7.Iterationstrategies..........................................55

2.8.Comparisonandgenericityiterations.........................62

2.9.Coarsestructure............................................65

CHAPTER 3.BACKGROUND-INDUCEDITERATIONSTRATEGIES ........71

3.1.Fullbackgroundextenderconstructions......................72

3.2.Resurrectionmaps.........................................78

3.3.AShiftLemmaforconversionstages.........................81

3.4.Conversionsystems........................................85

3.5.Inducediterationstrategies..................................92

3.6.Internalconsistencyforiterationstrategies....................93

3.7.Measurableprojecta........................................95

3.8.Projectawithmeasurablecofinality..........................99

CHAPTER 4.MOREMICEANDITERATIONTREES .....................103

4.1.Micewithprojectum-freespaces.............................104

4.2.Othersoundnesspatterns....................................112

4.3.Elementarityforpremouseembeddings.......................115

4.4.Plustrees..................................................129

4.5.Copymaps,liftedtrees,andlevelsofelementarity.............136

4.6.Iterationstrategiesandcomparison...........................149

4.7.PFSconstructionsandtheirresurrectionmaps.................154

4.8.Conversionsystemsandinducedstrategies....................158

4.9.Backgroundsforplusextenders..............................170

4.10.SolidityinPFSconstructions................................178

4.11.TheBicephalusLemma.....................................199

CHAPTER 5.SOMEPROPERTIESOFINDUCEDSTRATEGIES .............203

5.1.Copyingcommuteswithconversion..........................203

5.2.Positionalityandstrategycoherence..........................208

5.3.Pullbackconsistency........................................211

5.4.Internalliftconsistency.....................................215

5.5.Areductionto λ -separatedtrees.............................219

CHAPTER 6.NORMALIZINGSTACKSOFITERATIONTREES ............221

6.1.Normalizingtreesoflength2................................222

6.2.Normalizing T F ........................................231

6.3.Theextendertree V ext ......................................245

6.4.Treeembeddings...........................................246

6.5.Normalizing T U ..........................................251

6.6.Thebranchesof W (T , U ) ...................................266

6.7.Quasi-normalizingstacksofplustrees........................278

6.8.Copyingcommuteswithnormalization.......................284

6.9.Normalizinglongerstacks..................................290

CHAPTER 7.STRATEGIESTHATCONDENSEANDNORMALIZEWELL ...293

7.1.Thedefinitions.............................................294

7.2.Coarse Γ-Woodinsand Γ-universality........................302

7.3.Stronguniqueiterabilityfrom UBH ..........................307

7.4.Finestrategiesthatnormalizewell...........................317

7.5.Finestrategiesthatcondensewell............................329

7.6.Pureextenderpairs.........................................336

CHAPTER 8.COMPARINGITERATIONSTRATEGIES ....................345

8.1.Iteratingintoabackgroundedpremouse......................346

8.2.Extendingtreeembeddings..................................351

8.3.Resurrectionembeddingsasbranchembeddings...............357

8.4.Iteratingintoabackgroundedstrategy........................360

CHAPTER 9.FINESTRUCTUREFORTHELEASTBRANCHHIERARCHY ..391

9.1.Leastbranchpremice.......................................392

9.2.Leastbranchhodpairs......................................398

9.3.MousepairsandtheDodd-JensenLemma....................400

9.4.Backgroundconstructions...................................405

9.5.Comparisonandthehodpairorder...........................413

9.6.Theexistenceofcores......................................418

CHAPTER 10.PHALANXITERATIONINTOACONSTRUCTION ..........443

10.1.TheBicephalusLemma.....................................443

10.2.ThePseudo-premouseLemma...............................451

10.3.ProofofLemma9.6.5......................................464

10.4.Somesuccessfulbackgroundconstructions...................483

10.5. UBH holdsinhodmice.....................................486

CHAPTER 11.HOD INTHEDERIVEDMODELOFAHODMOUSE .......499

11.1.Genericinterpretability.....................................499 11.2.Mouselimits...............................................501

11.3.HODasamouselimit......................................504

11.4.HODmicesatisfy V = K ...................................513

11.5.Furtherresults.............................................518

PREFACE

Thisbookbeganlifeasalongresearcharticletitled Normalizingiterationtrees andcomparingiterationstrategies.Ifoundthemainideasbehindthecomparison processthatmotivatesitinSpring2015,andcirculatedahandwrittenmanuscript shortlyafterward.IcirculatedapreliminaryformofthepresentbookinApril2016, andhaverevisedandexpandeditmanytimessincethen,asvarioussignificant gapsanderrorsshowedup.Thelastmajorrevisionstookplacein2020-2021.1

Beyondmakingthebookcorrect,oneofmygoalshasbeentomakeitaccessible. Iwasencouragedherebythefactthatthenewdefinitionsandresultsareactually quiteelementary.Theyrestonthetheoryof Finestructureanditerationtrees (FSIT),andcanbeseenascompletingthattheoryinacertainway.Thecomparison theoremforpureextendermicethatisattheheartofFSITisdeficient,inthathow twomicecomparedependsonwhichiterationstrategiesarechosentocompare them.Hereweremedythatdefect,bydevelopingamethodforcomparingthe strategies.Theresultisacomparisontheoremfor mousepairs paralleltotheFSIT comparisontheoremforpureextendermice.Wethenusethecomparisonprocess underlyingthattheoremtodevelopafinestructuretheoryfor strategymice parallel tothefinestructuretheoryforpureextendermiceofFSIT.

Therearepointsatwhichdescriptivesettheoryunderdeterminacyhypotheses becomesrelevant.Atthesepoints,itwouldhelptohavereadthelatersectionsof [65].However,IhaveincludedenoughmaterialthatthereaderfamiliarwithFSIT butshakyondeterminacyshouldbeabletofollowtheexposition.Ourworkhere ismotivatedbytheproblemofanalyzingordinaldefinabilityinmodelsofAxiom ofDeterminacy,buttheprerequisiteforfollowingmostofitisjustinnermodel theoryatthelevelofFSIT.

1Thisisapre-publicationcopyonly.Thefinal,publishedversionofthebookcanbepurchased throughCambridgeUniversityPressandotherstandarddistributionchannels.Thisprepublicationcopy ismadeavailableforpersonaluseonlyandmustnotbesoldorre-distributed.

INTRODUCTION

Inthisbookweshalldevelopageneralcomparisonprocessforiterationstrategies, andshowhowtheprocesscanbeusedtoanalyzeordinaldefinabilityinmodels oftheAxiomofDeterminacy.Inthisintroduction,welookatthecontextand motivationforthetechnicalresultstocome.

Webeginwithabroadoverviewof innermodeltheory,thesubjecttowhichthis bookbelongs.Eventuallywereachanoutlineoftheideasandresultsthatarenew here.Thejourneyisorganizedsothatthetechnicalbackgroundneededtofollow alongincreasesasweproceed.

1.1.Largecardinalsandtheconsistencystrengthhierarchy

Strongaxiomsofinfinity,orastheyaremoreoftencalled,largecardinalhypotheses,playacentralroleinsettheory.Thereareatleasttworeasons.

First,largecardinalhypothesescanbeusedtodecideinanaturalwaymany questionswhichcannotbedecidedonthebasisof ZFC (thecommonlyaccepted systemofaxiomsforsettheory,andhenceallofmathematics).Manysuch questionscomefrom descriptivesettheory,thetheoryofsimplydefinablesetsof realnumbers.Forexample,thehypothesisthatthereareinfinitelymanyWoodin cardinalsyieldsasystematicanddetailedtheoryofthe projective setsofreals,those thataredefinableinthelanguageofsecondorderarithmeticfromrealparameters. ZFC byitselfyieldssuchatheoryatonlythesimplestlevelsofsecondorder definability.

Second,largecardinalhypothesesprovideawayoforganizingandsurveying allpossiblenaturalextensionsof ZFC.Thisisduetothefollowingremarkable phenomenon:foranynaturalextension T of ZFC whichsettheoristshavestudied, thereseemstobeanextension S of ZFC axiomatizedbylargecardinalhypotheses suchthattheconsistencyof T isprovably(in ZFC)equivalenttothatof S.The consistencystrengthsofthelargecardinalhypothesesarelinearlyordered,and usuallyeasytocompare.Thusallnaturalextensionsof ZFC seemtofallinto

ahierarchylinearlyorderedbyconsistencystrength,andcalibratedbythelarge cardinalhypotheses.2

Thesetwoaspectsoflargecardinalhypothesesareconnected,inthattheconsistencystrengthorderonnaturaltheoriescorrespondstotheinclusionorderon thesetoftheir“sufficientlyabsolute”consequences.Forexample,if S and T arenaturaltheoriesextending ZFC,and S hasconsistencystrengthlessthanor equaltothatof T ,thenthearithmeticconsequencesof S areincludedinthose of T .Ifinaddition, S and T haveconsistencystrengthatleastthatof“thereare infinitelymanyWoodincardinals”,thentheconsequencesof S inthelanguageof secondorderarithmeticareincludedinthoseof T .Thispatternpersistsatstill higherconsistencystrengths,withstillmorelogicallycomplicatedconsequences aboutrealsandsetsofrealsbeingbroughtintoauniformorder.Thisbeautifuland suggestivephenomenonhasapracticaldimensionaswell:onewaytodevelopthe absoluteconsequencesofastrongtheory T istocomputeaconsistencystrength lowerbound S for T intermsoflargecardinalhypotheses,andthenworkinthe theory S.Foroneofmanyexamples,theProperForcingAxiom(PFA)yieldsa canonicalinnermodelwithinfinitelymanyWoodincardinalsthatiscorrectfor statementsinthelanguageofsecondorderarithmetic,andtherefore PFA implies allconsequencesoftheexistenceofinfinitelymanyWoodincardinalsthatcanbe statedinthelanguageofsecondorderarithmetic.

Onecanthinkoftheconsistencystrengthofatheoryasthedegreetowhich itiscommittedtotheexistenceofthehigherinfinite.Largecardinalhypotheses maketheircommitmentsexplicitly:theysimplysayoutrightthattheinfinitiesin questionexist.Itisthereforeusuallyeasytocomparetheirconsistencystrengths. Othernaturaltheoriesoftenhavetheircommitmentstotheexistenceoftheinfinite wellhidden.Nevertheless,settheoristshavedevelopedmethodswherebythese commitmentscanbebroughttothesurface,andcompared.Thesemethodshave revealedtheremarkablephenomenondescribedinthelastparagraph,thatnatural theoriesappeartobewellorderedbythedegreestowhichtheyarecommittedto theinfinite,andthatthisdegreeofcommitmentcorrespondsexactlytothepower ofthetheorytodecidequestionsaboutconcreteobjects,likenaturalnumbers,real numbers,orsetsofrealnumbers.

Weshouldemphasizethattheparagraphsabovedescribeageneralpatternof existingtheorems.Therearemanyexamplesofnaturaltheorieswhoseconsistency strengthshavenotyetbeencomputed,andperhapsthey,orsomenaturaltheory yettobefound,willprovidecounterexamplestothepatterndescribedabove. Thepervasivenessofthepatternwhereweknowhowtocompareconsistency strengthsisevidencethatthiswillnothappen.3 Thetwomethodswherebyset

2Let con(T ) besomenaturalformalizationoftheassertionthat T isconsistent.Theconsistency strengthorderisgivenby: S ≤con T iff ZFC provescon(T ) → con(S)

3Thepatternextendstoweaksubtheoriesof ZFC aswell.Thisbookisconcernedonlywith theorieshavingverystrongcommitmentstoinfinity,andsoweshallignoresubtheoriesof ZFC,butthe linearityoftheconsistencystrengthsbelowthatof ZFC isevidenceoflinearityhigherup.

theoristscompareconsistencystrengths,forcingandinnermodeltheory,seemto leadinevitablytothepattern.Inparticular,thewellorderofnaturalconsistency strengthsseemstocorrespondtotheinclusionorderoncanonicalminimalinner modelsforlargecardinalhypotheses.Forcingandinnermodeltheoryseem sufficientlygeneraltocompareallnaturalconsistencystrengths,butatthemoment, thisisjustinformedspeculation.Soonereasonableapproachtounderstanding thegeneralpatternofconsistencystrengthsistodevelopourcomparisonmethods further.Inparticular,innermodeltheoryisingreatneedoffurtherdevelopment, astherearequiteimportantconsistencystrengthsthatitdoesnotyetreach.

1.2.Innermodeltheory

Theinnermodelprogramattemptstoassociatetoeachlargecardinalhypothesis H acanonicalminimaluniverseofsets MH (an innermodel)inwhich H istrue. Thestronger H is,thelarger MH willbe;thatis, G ≤con H ifandonlyif MG ⊆ MH Someofourdeepestunderstandingoflargecardinalhypothesescomesfromthe innermodelprogram.

Theinnermodelswehavesofarconstructedhaveaninternalstructurewhich admitsasystematic,detailedanalysis,a finestructuretheory ofthesortpioneered byRonaldJensenaround1970([16]).Thusbeingabletoconstruct MH givesus averygoodideaastowhatauniversesatisfying H mightlooklike.Innermodel theorytherebyprovidesevidenceoftheconsistencyofthelargecardinalhypotheses towhichitapplies.(Theauthorbelievesthatthiswillsomedayincludeallthe largecardinalhypothesescurrentlystudied.)Sinceforcingseemstoreduceany consistencyquestiontotheconsistencyquestionforsomelargecardinalhypothesis, itisimportanttohaveevidencethatthelargecardinalhypothesesthemselvesare consistent!Noevidenceismoreconvincingthananinnermodeltheoryforthe hypothesisinquestion.

Thesmallestofthecanonicalinnermodelsistheuniverse L ofconstructiblesets, isolatedbyKurtG¨odel([14])inhis1937proofthat CH isconsistentwith ZFC.It wasnotuntilthemid1960’sthatJ.SilverandK.Kunen([57],[23])developedthe theoryofacanonicalinnermodelgoingproperlybeyond L,byconstructing MH for H = “thereisameasurablecardinal”.4 Sincethen,progressivelylarger MH forprogressivelystronger H havebeenconstructedandstudiedindetail.(Seefor example[7],[27],and[28].)Atthemoment,wehaveagoodtheoryofcanonical innermodelssatisfying“thereisaWoodincardinal”,andevenslightlystronger hypotheses.(See[26],[30],and[61],forexample.)Oneofthemostimportantopen problemsinsettheoryistoextendthistheorysignificantlyfurther,withperhaps

4ZFC isofcoursetooweak,consistency-wise,toprovethatthereissuchamodel.Silverand Kunenworkedinthetheory ZFC+ “thereisameasurablecardinal”.Inthemid1970s,Doddand Jensendevelopedgeneralmethodsforconstructingthecanonicalinnermodelwithameasurableunder awideassortmentofhypotheses.See[7].

themostwell-knowntargetbeingmodelssatisfying“thereisasupercompact cardinal”.

Innermodeltheoryisacrucialtoolincalibratingconsistencystrengths:inorder toprovethat H ≤con T ,where H isalargecardinalhypothesis,onegenerally constructsacanonicalinnermodelof H insideanarbitrarymodelof T .Because wedonothaveafullinnermodeltheoryveryfarpastWoodincardinals,welack themeanstoprovemanywell-knownconjecturesoftheform H ≤con T ,where H issignificantlystrongerthan“thereisaWoodincardinal”.Broadlyspeaking,there aregreatdefectsinourunderstandingoftheconsistencystrengthhierarchybeyond Woodincardinals.

Innermodeltheoryisalsoacrucialtoolindevelopingtheconsequencesforreal numbersoflargecardinalhypotheses.Indeed,thebasicsofinnermodeltheory forWoodincardinalswerediscoveredin1985-86byD.A.Martinandtheauthor, atroughlythesametimetheydiscoveredtheirproofofProjectiveDeterminacy, or PD.(Martin,Moschovakis,andothershadshowninthe1960’sand70’sthat PD decidesinanaturalwayalltheclassicalquestionsaboutprojectivesetsleft undecidedby ZFC alone.)Thissimultaneousdiscoverywasnotanaccident,as thefundamentalnewtoolinbothcontextswasthesame: iterationtrees,andthe iterationstrategies whichproducethem.Sincethen,progressininnermodeltheory hasgivenusadeeperunderstandingofpuredescriptivesettheory,andthemeans tosolvesomeoldproblemsinthatfield.

Thefundamentalopenproblemofinnermodeltheoryistoextendthetheoryto modelssatisfyingstrongerlargecardinalhypotheses.“Thereisasupercompact cardinal”isanoldandstillquitechallengingtarget.Oneverywellknowntest questionhereiswhether (ZFC+“thereisasupercompactcardinal”) ≤con ZFC + PFA.Theanswerisalmostcertainlyyes,andtheproofalmostcertainlyinvolves aninnermodeltheorythatisfiringonallcylinders.5 Thatkindofinnermodel theorywehavenowonlyatthelevelofmanyWoodincardinals,butsignificant partsofthetheorydoexistalreadyatmuchhigherlevels.6

5Aparallel,andstillolder,questioniswhether(ZFC +“thereisasupercompactcardinal”) ≤con ZFC+ “thereisastronglycompactcardinal”.

6J.Baumgartnershowedintheearly1980sthat ZFC + PFA ≤con ZFC+“thereisasupercompact cardinal”.SupercompactsarefarbeyondWoodincardinals,inthesensethattherearemanyinteresting consistencystrengthsstrictlybetweenthetwo,andinthesensethatconstructingcanonicalinner modelsforsupercompactspresentssignificantnewdifficulties.Manysettheoreticprincipleshave beenshownconsistentrelativetotheexistenceof(sometimesmany)supercompactcardinals,so inner-model-theoreticevidenceoftheirconsistencywouldbevaluable.

1.3.Miceanditerationstrategies

Thecanonicalinnermodelsweseekareoftencalled mice.Therearetwo principalvarieties,thepureextendermiceandthestrategymice.7

Apureextenderpremouseisamodeloftheform Lα [E] where E isacoherent sequenceofextenders.Hereanextenderisasystemofultrafilterscodingan elementaryembedding,andcoherencemeansroughlythattheextendersappearin orderofstrength,withoutleavinggaps.ThesenotionswereintroducedbyMitchell inthe1970s8,andtheyhavebeenafoundationforworkininnermodeltheory sincethen.

Inthisbook,weshallassumethatourpremicehavenolongextendersontheir coherentsequences.9 Suchpremicecanmodelsuperstrong,andevensubcompact, cardinals.Theycannotmodel κ +-supercompactness.Longextendersleadtoan additionalsetofdifficulties.

An iterationstrategy isawinningstrategyforplayer II intheiterationgame.For anypremouse M,theiterationgameon M isatwoplayergameoflength ω1 + 1 10 Inthisgame,theplayersconstructatreeofmodelssuchthateachsuccessivenode onthetreeisobtainedbyanultrapowerofamodelthatalreadyexistsinthetree. I istheplayerthatdescribeshowtoconstructthisultrapower.Hechoosesan extender E fromthesequenceofthelastmodel N constructedsofar,thenchooses anothermodel P inthetreeandtakestheultrapowerof P by E.Iftheultrapoweris ill-foundedthenplayer I wins;otherwisetheresultingultrapoweristhenextnode onthetree.Player II movesatlimitstages λ bychoosingabranchofthetreethat hasbeenvisitedcofinallyoftenbelow λ ,andissuchthatthedirectlimitofthe embeddingsalongthebranchiswell-founded.Ifhefailstodoso,heloses.IfII managestostayinthecategoryofwellfoundedmodelsthroughall ω1 + 1 moves, thenhewins.Awinningstrategyfor II inthisgameiscalledan iterationstrategy for M,and M issaidtobe iterable justincasethereisaniterationstrategyforit. Iterablepureextenderpremicearecalled pureextendermice

Pureextendermicearecanonicalobjects;forexample,anyrealnumberbelongingtosuchamouseisordinaldefinable.Letussaythatapremouse M is pointwise definable ifeveryelementof M isdefinableover M.Foranyaxiomatizabletheory T ,theminimalmousesatisfying T ispointwisedefinable.Thecanonicityofpure extendermiceisduetotheiriterability,which,viathefundamental Comparison Lemma,impliesthatthepointwisedefinablepureextendermicearewellordered byinclusion.Thisisthe mouseorder onpointwisedefinablepureextendermice.

7Strategymicearesometimescalled hodmice,becauseoftheirroleinanalyzingthehereditarily ordinaldefinablesetsinmodelsoftheAxiomofDeterminacy.

8See[27]and[28].

9Anextenderisshortifallitscomponentultrafiltersconcentrateonthecriticalpoint.Otherwise,it islong.

10Iterationgamesofotherlengthsarealsoimportant,butthislengthiscrucial,soweshallfocuson it.

Theconsistencystrengthof T isdeterminedbytheminimalmouse M havinga genericextensionsatisfying T ,andthustheconsistencystrengthorderonnatural T ismirroredinthemouseorder.However,inthecaseofthemouseorder,we have proved thatwehaveawellorder;whatwecannotyetdoistienatural T at highconsistencystrengthstoit.Asweclimbthemouseorder,themicebecome correct(reflectwhatistrueinthefulluniverseofsets)athigherandhigherlevels oflogicalcomplexity.

Iterationstrategiesforpointwisedefinablepureextendermicearealsocanonicalobjects;forexample,apointwisedefinablemousehasexactlyoneiteration strategy.11 Theexistenceofiterationstrategiesisattheheartofthefundamental problemofinnermodeltheory,andforapointwisedefinable M,toprovethe existenceofaniterationstrategyistodefineit.Inpractice,itseemsnecessary togiveadefinitioninthesimplestpossiblelogicalform.Aswegohigherinthe mouseorder,thelogicalcomplexityofiterationstrategiesmustincrease,inaway thatkeepspacewiththecorrectnessofthemicetheyidentify.

Ourmostpowerful,all-purposemethodforconstructingiterationstrategiesis the coremodelinductionmethod.Becauseiterationstrategiesmustactontrees oflength ω1,theyarenotcodedbysetsofreals.Nevertheless,thefragmentof theiterationstrategyforacountablemousethatactsoncountableiterationtrees is codedbyasetofreals.Ifthissethappenstobeabsolutelydefinable(thatis, UniversallyBaire)thenthestrategycanbeextendedtoactonuncountableiteration treesinauniqueway.Thereisnootherwayknowntoconstructiterationstrategies actingonuncountabletrees.Thus,havinganabsolutelydefinableiterationstrategy forcountabletreesistantamounttohavingafulliterationstrategy.Thekey ideainthecoremodelinductionistousetheconceptsofdescriptivesettheory, underdeterminacyhypotheses,toidentifyanextrelevantlevelofcorrectnessand definabilityforsetsofreals,atargetlevelatwhichthenextiterationstrategyshould bedefinable.

Absolutedefinabilityleadstodeterminacy.Thusatreasonablyclosedlimitsteps inacoremodelinduction,onehasamodel M of AD + V = L(P(R)) thatcontains therestrictionstocountabletreesoftheiterationstrategiesalreadyconstructed. UnderstandingthestructureofHODM isimportantforgoingfurther.

1.4. HOD inmodelsofdeterminacy

HOD istheclassofallhereditarilyordinaldefinablesets.Itisamodelof ZFC12 , butbeyondthat, ZFC doesnotdecideitsbasictheory,andthesameistrueof ZFC augmentedbyanyoftheknownlargecardinalhypotheses.Theproblemisthat thedefinitionsonehasallowedarenotsufficientlyabsolute.Incontrast,thetheory

11ThisfollowsfromTheorem4.11of[65],andthefactthatanyiterationstrategyforapointwise definable M hastheWeakDodd-Jensenpropertywithrespecttoallenumerationsof M

12See[31].

of HOD indeterminacymodelsiswell-determined,notsubjecttothevagariesof forcing.13

Thestudyof HOD inmodelsof AD hasalonghistory.Thereadershould see[67]forasurveyofthishistory. HOD wasstudiedbypurelydescriptiveset theoreticmethodsinthelate70sand80s,andpartialresultsonbasicquestions suchaswhether HOD |= GCH wereobtainedthen.Itwasknownthenthatinner modeltheory,ifonlyonecoulddevelopitinsufficientgenerality,wouldberelevant tocharacterizingtherealsin HOD.Itwasknownthat HODM iscloseto M in variousways;forexample,if M |= AD+ + V = L(P(R))14,then M canberealized asasymmetricforcingextensionof HODM ,sothatthefirstordertheoryof M is partofthefirstordertheoryofitsHOD. 15

Justhowrelevantinnermodeltheoryistothestudyof HOD inmodelsof AD becameclearin1994,whentheauthorshowedthatifthereare ω Woodincardinals withameasurableabovethemall,then HODL(R) upto θ L(R) isapureextender mouse.16(See[60].)Shortlyafterward,thisresultwasimprovedbyHughWoodin, whoreduceditshypothesisto ADL(R),andidentifiedthefull HODL(R) asamodel oftheform L[M, Σ],where M isapureextenderpremouse,and Σ isapartial iterationstrategyfor M. HODL(R) isthusanewtypeofmouse,sometimescalled a strategymouse,sometimescalleda hodmouse.See[77]foranaccountofthis work.

Sincethemid-1990s,therehasbeenagreatdealofworkdevotedtoextending theseresultstomodelsofdeterminacybeyond L(R).Woodinanalyzed HOD in modelsof AD+ belowtheminimalmodelof ADR finestructurally,andSargsyan extendedtheanalysisfurther,firsttodeterminacymodelsbelow ADR + “θ is regular”(see[37]and[38]),andmorerecently,tomodelsofstillstrongerformsof determinacy. 17 Partofthemotivationforthisworkisthatitseemstobeessential inthecoremodelinduction:ingeneral,thenextiterationstrategyseemstobea strategyforahodmouse,notforapureextendermouse.Thisideacomesfrom workofWoodinandKetchersidaround2000.(See[21]and[47].)

13Wemeanheredeterminacymodelsoftheform M = L(Γ, R),where Γ isaproperinitialsegment oftheuniversallyBairesets.IftherearearbitrarilylargeWoodincardinals,thenforanysentence ϕ , whether ϕ istrueinallsuch HODM isabsoluteundersetforcing.ThisfollowseasilyfromWoodin’s theoremonthegenericabsolutenessof (Σ2 1)uB statements.See[64,Theorem5.1].

14AD+ isatechnicalstrengtheningof AD.Itisnotknownwhether AD ⇒ AD+ ,butinevery modelof AD constructedsofar, AD+ alsoholds.Inparticular,themodelsof AD thatarerelevantin thecoremodelinductionsatisfy AD+

15ThisisatheoremofWoodinfromtheearly1980s.Cf.[67].

16Inadeterminacycontext, θ denotestheleastordinalthatisnotthesurjectiveimageofthereals.

17See[39].Partofthisworkwasdoneincollaborationwiththeauthor;see[68],[74],and[69].The determinacyprinciplesdealtwithhereareallweakerthanaWoodinlimitofWoodincardinals.

1.5.Leastbranchhodpairs

Thestrategymiceusedintheworkjustdescribedhavetheform M = L[E, Σ], where E isacoherentsequenceofextenders,and Σ isaniterationstrategyfor M

Thestrategyinformationisfedintothemodel M slowly,inawaythatisdictated inpartbythedeterminacymodelwhose HOD isbeinganalyzed.Onesaysthatthe hierarchyof M is rigidlylayered,or extenderbiased.Theobject (M, Σ) iscalleda rigidlylayered(extenderbiased) hodpair.

Perhapsthemainmotivationfortheextenderbiasedhierarchyisthatitmakesit possibletoproveacomparisontheorem.Thereisnoinnermodeltheorywithout suchatheorem.Comparingstrategymicenecessarilyinvolvescomparingiteration strategies,andcomparingiterationstrategiesissignificantlymoredifficultthan comparingextendersequences.Rigidlayeringletsoneavoidthedifficulties inherentinthegeneralstrategycomparisonproblem,whileprovingcomparisonfor aclassofstrategymiceadequatetoanalyze HOD intheminimalmodelof ADR + “θ isregular”,andsomewhatbeyond.Thekeyisthatinthisregion, HOD doesnot havecardinalsthatarestrongpastaWoodincardinal.

Unfortunately,rigidlayeringdoesnotseemtohelpincomparingstrategymice thathavecardinalsthatarestrongpastaWoodin.Moreover,ithasseriouscosts. Thedefinitionof“hodpremouse”becomesverycomplicated,andindeeditisnot clearhowtoextendthedefinitionofrigidlylayeredhodpairsmuchpastthatgiven in[39].Thedefinitionof“rigidlylayeredhodpremouse”isnotuniform,inthatthe extentofextenderbiasdependsonthedeterminacymodelwhose HOD isbeing analyzed.Finestructure,andinparticularcondensation,becomemoreawkward. Forexample,itisnottrueingeneralthatthepointwisedefinablehullofalevelof M isalevelof M.(Theproblemisthatthehullwillnotgenerallybesufficiently extenderbiased.)

Themorenaivenotionofhodpremousewouldabandonextenderbias,and simplyaddtheleastmissingpieceofstrategyinformationatessentiallyevery stage.ThiswasoriginallysuggestedbyWoodin.18 Thefocusofthisbookisa generalcomparisontheoremforiterationstrategiesthatmakesitpossibletouse thisapproach,atleastintherealmofshortextenders.Theresultingpremiceare called leastbranchpremice (lpm’s),andthepairs (M, Σ) arecalled leastbranch hodpairs (lbrhodpairs).Combiningresultsofthisbookand[73],onehas THEOREM 1.5.1([73]). Assume AD++ “thereisan (ω1, ω1) iterationstrategy forapureextenderpremousewithalongextenderonitssequence”;then (1) forany Γ ⊆ P(R) suchthat L(Γ, R) |= ADR+ “thereisno (ω1, ω1) iteration strategyforapureextenderpremousewithalongextenderonitssequence”, HODL(Γ,R) isaleastbranchpremouse,and

18Therearesomefine-structuralproblemswiththeprecisemethodforinsertingstrategyinformation originallysuggestedbyWoodin.Themethodforstrategyinsertionthatiscorrectindetailisdueto SchlutzenbergandTrang.Cf.[56].

(2) thereisa Γ ⊆ P(R) suchthat L(Γ, R) |= ADR+ “thereisno (ω1, ω1) iterationstrategyforapureextenderpremousewithalongextenderonits sequence”,and HODL(Γ,R) |= “thereisasubcompactcardinal”.

Ofcourse,onewouldliketoremovethemouseexistencehypothesisof1.5.1, andproveitsconclusionunder AD+ alone.Findingawaytodothisisone manifestationofthelongstandingiterabilityproblemwehavediscussedabove. Althoughwedonotyetknowhowtodothis,thetheoremdoesmakeithighlylikely thatinmodelsof ADR thathavenotreachedaniterationstrategyforapureextender premousewithalongextender, HOD isaleastbranchpremouse.Italsomakesit verylikelythattherearesuch HOD’swithsubcompactcardinals.Subcompactness isoneofthestrongestlargecardinalpropertiesthatcanberepresentedwithshort extenders.19

AlthoughweshallnotproveTheorem1.5.1here,weshallproveanapproximationtoitthatmakesthesamepoints.ThatapproximationisTheorem11.3.13 below.

Leastbranchpremicehaveafinestructuremuchclosertothatofpureextender modelsthanthatofrigidlylayeredhodpremice.Inthisbookwedevelopthe basics,includingthesolidityanduniversalityofstandardparameters,andaformof condensation.In[76],theauthorandN.Tranghaveprovedasharpercondensation theorem,whosepureextenderversionwasusedheavilyintheSchimmerlingZemanwork([44])on inpureextendermice.Itseemslikelythattherestofthe Schimmerling-Zemanworkextendsaswell.

Thusleastbranchhodpairsgiveusagoodtheoryof HOD intheshortextender realm,providedthereareenoughsuchpairs.20 Below,weformulateaconjecture thatwecall HodPairCapturing,or HPC,thatmakesprecisethestatementthat thereareenoughleastbranchhodpairs. HPC isthemainopenprobleminthe theorytowhichthisbookcontributes.

1.6.Comparisonandthemousepairorder

Letusfirstsaymoreaboutthenatureofleastbranchhodpairs (M, Σ).There arefourrequirementson Σ inthedefinition:stronghullcondensation,quasinormalizingwell,internalliftconsistency,andpushforwardconsistency.Weshall describetheserequirementsinformally,omittingsomeofthefinepoints,andgive thefulldefinitionslater.

Recallthataniterationtreeonapremouse M is normal ifftheextenders EW α usedin W havelengthsincreasingwith α ,andeach EW α isappliedtothelongest

19Untilnow,therewasnoverystrongevidencethatthe HOD ofadeterminacymodelcouldsatisfy thattherearecardinalsthatarestrongpastaWoodincardinal.

20Atleastinthecasethatthebackgrounddeterminacymodelsatisfies ADR + V = L(P(R)).Some formofextenderbiasmaybeappropriateinothercases.

possibleinitialsegmentoftheearliestpossiblemodelin W.Fortechnicalreasons weneedtoconsideraslightweakeningofthelength-increasingrequirement;we calltheresultingtrees quasi-normal.Ouriterationstrategieswillactonfinite stacks ofquasi-normaltrees,thatis,sequences s = T0,..., Tn suchthatforall k ≤ n 1, Tk+1 isaquasi-normaltreeonsomeinitialsegmentofthelastmodelin Tk .Wewrite M∞(s) forthelastmodelof Tn,ifthereisone.

DEFINITION 1.6.1. Let Σ beaniterationstrategyforapremouse P. (1) (Tailstrategy)If s isastackby Σ and Q ✂ M∞(s),then Σs,Q isthestrategyfor Q givenby: Σs,Q(t)= Σ(s Q, t ). 21 (2) (Pullbackstrategy)If π : N → P iselementary,then Σπ isthestrategyfor N givenby: Σπ (s)= Σ(π s),where π s istheliftof s by π toastackon P

In(2),elementaritymustbeunderstoodfinestructurally;ourconventionisthat everypremouse P hasadegreeofsoundnessattachedtoit,andelementaritymeans elementarityatthatquantifierlevel.

Perhapsthemostimportantregularitypropertyofiterationstrategiesis strong hullcondensation.Todefineitweneedthenotionofa treeembedding Φ : T→U , where T and U arenormaltreesonthesame M.Theideaofcourseisthat Φ should preserveacertainamountoftheiterationtreestructure,butsomecareisneededin spellingoutexactlyhowmuch. Φ isdeterminedbyamap u :lh(T ) → lh(U ) and maps πα : MT α →MU u(α ) havingvariousproperties.See §6.4.

DEFINITION 1.6.2. Let Σ beaniterationstrategyforapremouse M;then Σ has stronghullcondensation iffwhenever s isastackofnormaltreesby Σ and N ✂ M∞(s),and U isanormaltreeon N by Σs,N ,and Φ : T→U isatreeembedding, withassociatedmaps πα : MT α →MU u(α ),then

(a) T isby Σs,N ,and (b) forall α < lh(T ), Σs N,T α +1 =(Σs N,U u(α )+1 )πα

Stronghullcondensationisastrongerversionofthehullcondensationproperty isolatedbySargsyanin[37].

Thesecondimportantpropertyisquasi-normalizingwell.Givenan M-stack T , U withlastmodel N suchthat T and U arenormal,shufflingtheextenders of U into T inaminimalwayproducesanormaltree W = W (T , U ).If U has alastmodel R,wegetnearlyelementarymap π : N → R.Wecall W (T , U ) the embeddingnormalization of T , U .Theideaissimple,buttherearemany technicaldetails.22 Itprovesusefultoconsideraslightlylessminimalshuffling V (T , U ) thatwecallthe quasi-normalization of T , U .Evenif T and U are normal, V (T , U ) maynotbelength-increasing,butitisnearlyso.Thereader shouldseeChapter6forfulldefinitions.

21Forpremice Q and R, Q ✂ R iffthehierarchyof Q isaninitialsegmentofthatof R

22MuchofthegeneraltheoryofnormalizationwasdevelopedindependentlybySchlutzenberg.See [54].Seealso[19]and[58].

DEFINITION 1.6.3. Let Σ beaniterationstrategyforapremouse M.Wesay that Σ quasi-normalizeswell iffwhenever s isan M-stackby Σ,and T , U isa 2-stackby Σs suchthat T and U arenormaltreeshavinglastmodels,then

(a) V (T , U ) isby Σs,and (b) letting V = V (T , U ) and π : MU ∞ →MV ∞ bethemapgeneratedbyquasinormalization,wehavethat Σs T ,U =(Σs V )π .

Thefinalbasicregularitypropertyofiterationstrategiesforpureextender premiceis internalliftconsistency.Supposethat s isastackby Σ and P ✂ Q ✂ M∞(s).Stacks t on P canbeliftedtostacks t + on Q inanaturalway.Wesaythat Σ isinternallyliftconsistentiffforallsuch s, P, and Q, Σs,P(t)= Σs,Q(t +).See §5.4.

Forpairs (M, Σ) suchthat M isastrategymouse,werequirealsothatthe internalstrategypredicateof M beconsistentwith Σ.Moregenerally,letting ˙ Σ bethepredicatesymbolusedtorecordstrategyinformation,wesaythat (M, Σ) is pushforwardconsistent iffwhenever s isastackby Σ and N ✂ M∞(s),then ΣN ⊆ Σs,N If M isapureextenderpremouse,and Σ isastrategyfor M thathasstronghull condensation,quasi-normalizeswell,andisinternallyliftconsistent,thenwecall (M, Σ) a pureextenderpair.If M isaleastbranchpremouse,and Σ isastrategy for M thathasstronghullcondensation,quasi-normalizeswell,isinternallylift consistentandpushforwardconsistent,thenwecall (M, Σ) a leastbranchhodpair. Apairofoneofthetwotypesisa mousepair.

If (M, Σ) isamousepair,and s isastackby Σ withlastmodel N,thenwecall (N, Σs) an iterateof (M, Σ).Ifthebranch M-to-N of s doesnotdrop,wecallita non-droppingiterate.Inthatcase,wehaveaniterationmap is : M → N.Letus write

(M, Σ) ✂ (R, Λ) iff M ✂ R and Σ = ΛM .

Wehavenohopeofshowinganythingaboutmousepairs (M, Σ) unlesswe assumeabsolutedefinabiltyfortheiterationstrategy.Hereweassume Σ hasscope HC,i.e.that M iscountableand Σ isdefinedonfinitestacksofcountabletrees, andweassumethatweareinamodelof AD+ . 23 Thefollowingisthemainnew resultofthebook.

THEOREM 1.6.4. (ComparisonLemma)Assume AD+,andlet (P, Σ) and (Q, Ψ) bestronglystable24 mousepairswithscope HC ofthesamekind;thenthereare iterates (R, Λ) of (P, Σ) and (S, Ω) of (Q, Ψ),obtainedbynormaltrees T and U , suchthateither (1) (R, Λ) ✂ (S, Ω) andP-to-Rdoesnotdrop,or

23Onecouldrequirethattheybedefinedoncountablestacks.

24Strongstabilityisamildfinestructuralrequirement.Onecanavoiditbyslightlycomplicating thenotionof iterate.See4.4.5and4.6.12.

(2) (S, Ω) ✂ (R, Λ) andQ-to-Sdoesnotdrop.

Evenforpureextenderpairs,thistheoremisnew,becauseoftheagreement betweentailstrategiesitrequires.Infact,itisnoeasiertoprovethetheoremfor pureextenderpairsthanitistoproveitforleastbranchhodpairs.Theproofin bothcasesisthesame,anditmakesuseofthepropertiesoftheiterationstrategies wehaveisolatedinthedefinitionofmousepair.

WorkinginthecategoryofmousepairsenablesustostateageneralDodd-Jensen lemma.Letussay π : (P, Σ) → (Q, Ψ) iselementaryiff π iselementaryfrom P to Q,and Σ = Ψπ .Weshallshowthatanelementarysubmodelofamousepairisa mousepair,andthattheiterationmapsassociatedtonon-droppingiterationsofa mousepairareelementary.25

THEOREM 1.6.5(Dodd-Jensenlemma). Let (P, Σ) beamousepair,and (Q, Ψ) beaniterateof (P, Σ) viathestack s.Suppose π : (P, Σ) → (Q, Ψ) iselementary; thensdoesnotdrop,andforallordinals η ∈ P,is(η ) ≤ π (η ).

TheproofisjusttheusualDodd-Jensenproof;thepointisjustthatthelanguage ofmousepairsenablesustoformulatethetheoreminitspropergenerality.There isnoneedtorestricttomicewithuniqueiterationstrategies,asisusuallydone. Similarly,wecandefinethemouseorderinitspropergenerality,withoutrestrictingtomicewithuniqueiterationstrategies.If (P, Σ) and (Q, Ψ) arepairsof thesametype,then (P, Σ) ≤∗ (Q, Ψ) iff (P, Σ) canbeelementarilyembeddedinto aniterateof (Q, Ψ).TheComparisonandDodd-Jensentheoremsimplythat ≤∗ is aprewellorderoneachtype.

1.7.Hodpaircapturing

Leastbranchhodpairscanbeusedtoanalyze HOD inmodelsof AD+,provided thatthereareenoughsuchpairs.

DEFINITION 1.7.1(AD+). (a) HodPairCapturing (HPC) istheassertion: foreverySuslin-co-Suslinset A,thereisaleastbranchhodpair (P, Σ) such that A isdefinablefromparametersover (HC, ∈, Σ). (b) L[E] capturing (LEC) istheassertion:foreverySuslin-co-Suslinset A,there isapureextenderpair (P, Σ) suchthat A isdefinablefromparametersover (HC, ∈, Σ)

Anequivalent(under AD+)formulationwouldbethatthesetsofrealscoding strategiesofthetypeinquestion,undersomenaturalmapoftherealsonto HC, areWadgecofinalintheSuslin-co-Suslinsetsofreals.TherestrictiontoSuslinco-Suslinsets A isnecessary,for AD+ impliesthatif (P, Σ) isapairofoneofthe

25Neitherisobvious.Thatiterationmapsareelementaryisapropertyoftheiterationstrategy knownaspullbackconsistency.Itfollowsfromstronghullcondensation.

twotypes,thenthecodesetof Σ isSuslinandco-Suslin.Thisisthemainresultof [73],whereitisalsoshownthattheSuslinrepresentationconstructedisofoptimal logicalcomplexity.

Remark 1.7.2. HPC isacousinofSargsyan’s GenerationofFullPointclasses. See[37]and[38], §6.1.

Assuming AD+ , LEC isequivalenttothewellknownMouseCapturing:for reals x and y, x isordinaldefinablefrom y iff x isinapureextendermouseover y.Thisequivalenceisshownin[63].(SeeespeciallyTheorem16.6.)Weshow inTheorem10.4.3belowthatunder AD+ , LEC implies HPC.Wedonotknow whether HPC implies LEC.

Granted ADR and HPC,wehaveenoughhodpairstoanalyzeHOD.

THEOREM 1.7.3([73]). Assume ADR and HPC;then Vθ ∩ HOD istheuniverse ofaleastbranchpremouse.

Sometechniquesdevelopedin[59]and[73]areneededtoprovethetheorem,so weshallnotproveithere.

Thenaturalconjectureisthat LEC and HPC holdinallmodelsof AD+ thathave notreachedaniterationstrategyforapremousewithalongextender.Becauseour capturingmicehaveonlyshortextendersontheirsequences, LEC and HPC cannot holdinlargermodelsof AD+

DEFINITION 1.7.4. NLE (“Nolongextenders”)istheassertion:thereisno countable, ω1 + 1-iterablepureextenderpremouse M suchthatthereisalong extenderonthe M-sequence.

CONJECTURE 1.7.5. Assume AD+ and NLE;then LEC

CONJECTURE 1.7.6. Assume AD+ and NLE;then HPC.

Asweremarkedabove,1.7.5implies1.7.6.Conjecture1.7.5isequivalentto aslightstrengtheningoftheusualMouseSetConjecture MSC.(Thehypothesis of MSC isthatthereisnoiterationstrategyforapureextenderpremousewitha superstrong,whichisslightlystrongerthan NLE.) MSC hasbeenacentraltarget forinnermodeltheoristsforalongtime.

1.8.Constructingmousepairs

Thebasicsourceformousepairsisabackgroundconstruction.Inthesimplest case,suchaconstruction C buildspairs (Mν ,k , Ων ,k ) inductively,puttingextenders onthe Mν ,k -sequencethatarerestrictionsofniceextendersin V .Theiteration strategy Ων ,k isinducedbyaniterationstrategyfor V ,andifweareconstructing strategypremice,therelevantinformationabout Ων ,k isinsertedinto Mν ,k atthe

appropriatepoints. Mν ,k+1 isthecoreof Mν ,k .Theconstructionbreaksdownif thestandardparameterof Mν ,k behavespoorly,sothatthereisnocore.

Thereisofcoursemoretosayhere,andweshalldosolaterinthebook.For now,letusnotethatthebackgrounduniverseforsuchaconstructionshouldbea modelof ZFC thathaslotsofextenders,andyetknowshowtoiterateitself.Inthe AD+ context,thefollowingtheoremofWoodinapplies.26

THEOREM 1.8.1(Woodin). Assume AD+,andlet Γ beagoodpointclasssuch thatallsetsin Γ areSuslinandco-Suslin;thenforanyreal x thereisacoarse Γ-Woodinpair (N, Σ) suchthatx ∈ N.

Here,roughlyspeaking, N isacountabletransitivemodelof ZFC withaWoodin cardinalandatermforauniversal Γ set,and Σ isaniterationstrategyfor N that movesthistermcorrectly,andissuchthat Σ ∩ N isdefinableover N.SeeDefinition 7.2.3.

ThefollowingisessentiallyTheorem10.4.1tofollow.Ittooisoneofthemain newresultsofthebook.

THEOREM 1.8.2. Assume AD+,andlet (N, Σ) beacoarse Γ-Woodinpair.Let C bealeastbranchconstructionin N;then C doesnotbreakdown.Moreover,each ofitslevels (MC ν ,k , ΩC ν ,k ) isaleastbranchhodpairin N,andextendscanonically toaleastbranchhodpairinV.

Backgroundconstructionsofthesortdescribedinthistheoremhaveanimportant roletoplayinourcomparisonprocess.Assume AD+,andlet (M, Ω) and (N, Σ) bemousepairsofthesametype.Wecompare (M, Ω) with (N, Σ) byputting M and N intoacommon Γ-Woodinuniverse N∗,where Σ and Ω arein Γ ∩ ˇ Γ.We theniterate (M, Σ) and (N, Ω) intolevelsofafullbackgroundconstruction(ofthe appropriatetype)of N∗.Herearesomedefinitionsencapsulatingthemethod.

DEFINITION 1.8.3. Let (M, Σ) and (N, Ω) bemousepairsofthesametype; then

(a) (M, Σ) iteratespast (N, Ω) iffthereisa λ -separatediterationtree T by Σ on M whoselastpairis (N, Ω)

(b) (M, Σ) iteratesto (N, Ω) iffthereisa λ -separated T asin(a)suchthatthe branch M-to-N of T doesnotdrop.

(c) (M, Σ) iteratesstrictlypast (N, Ω) iffititeratespast (N, Ω),butnotto (N, Ω). λ -separationisasmallstrengtheningofnormalitythatisdefinedin4.4.8.One reasonthatitisimportantisthatif T is λ -separatedand U isanormaltreeon MT ∞ ,then W (T , U )= V (T , U ).Thatis,embeddingnormalizationcoincideswith quasi-normalizationinthiscase.

DEFINITION 1.8.4(AD+). Let (P, Σ) beamousepair;then (*)(P, Σ) isthe followingassertion:Let (N, Ψ) beanycoarse Γ-Woodinpairsuchthat P ∈ HCN∗ ,

26See[22],and[66,Lemma3.13].

and Σ ∈ Γ ∩ ˇ Γ.Let C beabackgroundconstructiondonein N∗ oftheappropriate type,andlet (R, Φ) bealevelof C.Supposethat (P, Σ) iteratesstrictlypastall levelsof C thatarestrictlyearlierthan (R, Φ);then (P, Σ) iteratespast (R, Φ).

If (M, Ω) isamousepair,and N isaninitialsegmentof M,thenwewrite ΩN fortheiterationstrategyfortreeson N thatisinducedby Ω.Wecanunpackthe conclusionof1.8.4asfollows:supposethecomparisonof P with R hasproduced anormaltree T on P withlastmodel Q,with T by Σ,and S isaninitialsegment ofboth Q and R;then ΣT ,S = ΦS.Thustheleastdisagreementbetween Q and R isanextenderdisagreement.Moreover,if E on Q and F on R aretheextenders involvedinit,then F = ∅.

Weshallshow(cf.Theorems8.4.3and9.5.6below)

THEOREM 1.8.5. Assume AD+;then (*)(P, Σ) holds,forallstronglystable mousepairs (P, Σ).

Thistheoremletsuscomparetwo(ormore)mousepairsofthesametype indirectly,bycomparingthemtothelevelsofanappropriateconstruction,done ina Γ-Woodinmodel,wherebothstrategiesarein Γ ∩ ˇ Γ.Onecanshowusingthe Woodinnessthat C reachesnon-droppingiteratesofbothpairs27.Thisgivesusa stage (M, Ω) of C suchthatoneofthepairsiteratestoit,whiletheotheriterates pastit.

1.9.Thecomparisonargument

Inwhatfollows,weshallgivefairlycompleteproofsofthetheoremsabove. Thebookislong,partlybecausewewantedtomakeitaccessible,andpartly becauseweshallbeforcedtorevisethebasicdefinitionsof[30]and[81]invarious ways,sothereisalimittowhatwecansimplyquote.Inaddition,theneed tocomparestrategiesaddsalayerofcomplexitytotheproofsofthemainfine structuraltheoremsaboutstrategymice.Nevertheless,themainnewideasbehind thestrategy-comparisonprocessitselfarereasonablysimple.Wedescribethem now.

Thefirststepistofocusonproving (*)(P, Σ).Thatis,ratherthandirectly comparingtwostrategies,weiteratethembothintoacommonbackgroundconstructionanditsstrategy.Inthecomparison-of-micecontext,thismethodgoes backtoKunen([23]),andwasfurtherdevelopedbyMitchell,Baldwin([5]),and theauthor.28 WoodinandSargsyanhadusedthemethodforstrategycomparison inthehodmousecontext.Allthesecomparisonscouldbereplacedbydirect comparisonsofthetwomiceorstrategiesinvolved,butinthegeneralcaseof comparisonofstrategies,thereareseriousadvantagestotheindirectapproach.

27See8.1.4.

28Inunpublished1985notestitled“Largecardinalsand ∆1 3 wellorders”.

Thereisnoneedtodecidewhattodoifoneencountersastrategydisagreement, becauseoneisprovingthatthatneverhappens.Thecomparisonprocessisjustthe usualoneofcomparingleastextenderdisagreements.Insteadofthedualproblems ofdesigningaprocessandprovingitterminates,onehasagivenprocess,and knowswhyitshouldterminate:nostrategydisagreementsshowup.Theproblem isjusttoshowthis.Theseadvantagesledtheauthortofocus,since2009,ontrying toprove (*)(P, Σ).

ThemainnewideathatmakesthispossibleismotivatedbySargsyan’sproofin [37]thatif Σ hasbranchcondensation,then (*)(P, Σ) holds.29Branchcondensation istoostrongtoholdonce P hasextendersoverlappingWoodincardinals;we cannotconcludethat Σ(T )= b fromhavingmerelyrealized MT b intoa Σ-iterate of P.Weneedsomekindofrealizationoftheentirephalanx Φ(T b) inorderto concludethat Σ(T )= b.Thisleadstoaweakeningofbranchcondensationthat onemightcall“phalanxcondensation”,inwhichoneasksforafamilyofbranchcondensation-likerealizationshavingsomenaturalagreementwithoneanother. Phalanxcondensationisstillstrongenoughtoimply (*)(P, Σ),andmightwell betrueingeneralforbackground-inducedstrategies.Unfortunately,Sargsyan’s constructionofstrategieswithbranchcondensationdoesnotseemtoyieldphalanx condensationinthemoregeneralcase.Foronething,itinvolvescomparison arguments,andinthegeneralcase,thislookslikeaviciouscircle.Itwasduring oneoftheauthor’smanyattemptstobreakintothiscirclethatherealizedthat certainpropertiesrelatedtophalanxcondensation,namelynormalizingwelland stronghullcondensation,couldbeobtaineddirectlyforbackground-induced strategies,andthatthesepropertiessufficefor (*)(P, Σ)

Letusexplainthislastpartbriefly.Supposethatweareinthecontextof Theorem1.8.5.Wehaveapremouse P withiterationstrategy Σ thatnormalizes wellandhasstronghullcondensation.Wehave N apremouseoccuringinthe fullybackgroundedconstructionof N∗,where P ∈ HCN∗ and N∗ captures Σ.We compare P with N byiteratingawaytheleastextenderdisagreement.Ithas beenknownsince1985thatonly P willmove.Wemustprovethatnostrategy disagreementshowsup.

Supposewehaveproduceda λ -separatediterationtree T on P withlastmodel Q,that Q|α = N|α ,andthat U isanormaltreeon R = Q|α = N|α oflimitlength playedbyboth ΣT ,R (thetailof Σ)and ΩR,where Ω isthe N∗-inducedstrategyfor N.Wewishtoshowthat ΣT ,R(U )= ΩR(U ).Because Σ isinternallyliftconsistent, wecanreducetothecasethat Q = R.

Let b = ΩR(U ).Wemustsee b = ΣT ,R(U ),thatis,that b = Σ( T , U ).Since T is λ -separated,embeddingnormalizationcoincideswithquasi-normalization.

29Roughly,aniterationstrategy Σ for M has branchcondensation iffwhenever T isaniterationtree oflimitlengthby Σ, b isacofinalbranchof T withassociatediterationmap ib : M → MT b , π : M → N isaniterationmapby Σ,andthereisa k : MT b → N suchthat π = k ◦ ib ,then Σ(T )= b.See[37]for moredetail.

Letusconsider

Wc = W (T , U c)= V (T , U c)

forarbitrarycofinalbranches c of U .Weshallsee:

(1) ΣT ,R(U )= c iff Wc isby Σ.The ⇒ directionfollowsatoncefromthefact that Σ quasi-normalizeswell,andthe ⇐ directionisprovedin §6.6.

(2) Letting i∗ b : N∗ → N∗ b comefromlifting iU b to N∗ viatheiteration-strategy constructionof[30],thereisatreeembeddingof Wb into i∗ b(T ).Thisisthe keystepintheproof.ItiscarriedoutinChapter8.

(3) i∗ b(Σ) ⊆ Σ because Σ wascapturedby N∗,so i∗ b(T ) isby Σ.

(4) Thus Wb isby Σ,because Σ hasstronghullcondensation.

(5) Soby(1), ΣT ,R(U )= b.

Hereisadiagramofthesituation:

IGURE 1.9.1.Proofof (*)(P, Σ). Wb isapsuedo-hullof i∗ b(T ).

1.10.Planofthebook

Chapters2and3collectandorganizesomestandarddefinitionsandresultsfrom innermodeltheory.Thebookisaimedatpeoplewhohavealreadyencountered

thismaterial,via[30],[65],or[81]forexample,butthesechapterswillserveasa bridgetotherestofthebook.

In §3.6weexplainwhythisstandardtheoryisnotcompletelyadequatetothe problemofcomparingiterationstrategies.Roughlyspeaking,theproblemisthat theinducediterationstrategiesforthelevelsofabackgroundconstructionarenot connectedsufficientlywelltotheiterationstrategyforthebackgrounduniverse. §3.7and §3.8analyzeoneofthetwosourcesofthisshortfall,andChapter4 removesbothofthem.Thisinvolvesrevisingthenotionsof premouse and iteration tree slightly,andre-provingthestandardfinestructuralresultsinthenewsetting.

Chapter5showsthatthenewdefinitionsleadtobackground-inducediteration strategiesthatarebetterbehavedinseveralways.Chapters6and7pushfurtherin thisdirection,leadingultimatelytoTheorem7.6.2,whichsaysthatpureextender backgroundconstructions,doneinanappropriatelyiterablebackgrounduniverse, producepureextenderpairs.

InChapter8weprovethemaincomparisontheoremforpureextenderpairs, Theorem8.4.3.Weshalladapttheproofof8.4.3toleastbranchhodpairsand tophalanxcomparisonsinChapters9and10,butthemainstepsallshowupin thissimplersituation,sowehavebegunwithit.Whenweusetheproofagainin Chapters9and10,weshallcondenselongstretchesbypointingtotheproofof 8.4.3.

Chapters9and10usethestrategy-comparisonprocesstodevelopthetheory ofleastbranchhodpairs.Chapter11usesthistheorytoanalyze HOD incertain modelsof ADR,andconcludeswithadiscussionoffurtherresultsthathavebeen provedbythemethodswedevelophere.

PRELIMINARIES

Innermodeltheorydealswithcanonicalobjects,butinnermodeltheoristshave presentedtheminvariousways.Theconventionsweusehereare,forthemost part,fairlycommon.Forbasicfinestructuralnotionssuchasprojecta,cores, standardparameters,fineultrapowers,anddegreesofelementarity,weshallstay closetoMitchell-Steel[30]andthepaper[49]bySchindlerandZeman.We shalluseJensenindexingforthesequencesofextendersfromwhichpremiceare constructed;seeforexampleZeman’sbook[81].InChapter4weshallmodifythe notionofpremouseslightly,byenlargingthestandardparametersandassociated cores.Untilwegettothatpoint,ournotionofpremouseisjustthestandardone determinedbytheconventionsof[30],[49],and[81].30

Mostofourterminologytodowithiterationtreesanditerationstrategiestraces backtoMartin-Steel[26]andMitchell-Steel[30],andisbynowprettystandard. Wedoneedtoconsidercarefullyiterationstrategiesdefinedonawiderclassof iterationtreesthaniscommon,andsothereissomelessfamiliarterminology definedinsections2.6and2.7.

2.1.Extendersandultrapowers

Ournotationforextendersisstandard.

DEFINITION 2.1.1. Let M betransitiveandrudimentarilyclosed;then E = Ea | a ∈ [θ ]<ω isa (κ , θ )-extenderover M withspaces µa | a ∈ [θ ]<ω ifand onlyif

(1) Each Ea isan (M, κ )-completeultrafilterover P([µa]|a|) ∩ M,with µa being theleast µ suchthat [µ ]|a| ∈ Ea.

(2) (Compatibility)For a ⊆ b and X ∈ M, X ∈ Ea ⇐⇒ X ab ∈ Eb

(3) (Uniformity) µ{κ } = κ

(4) (Normality)If f ∈ M and f (u) < max(u) for Ea a.e. u,thenthereisa β < max(a) suchthatfor Ea∪{β } a.e. u, f a,a∪{β }(u)= u{β },a∪{β } .

30Thenotionofpremousein[81],anditsrelatedfinestructure,originateinJensen’smanuscripts [17]and[18].

Theunexplainednotationherecanbefoundin[49, §8].Weshalloftenidentify E withthebinaryrelation (a, X ) ∈ E iff X ∈ Ea.Onecanalsoidentifyitwith theothersection-functionofthisbinaryrelation,whichisessentiallythefunction X → iM E (X ) ∩ θ .Wecall θ the length of E,andwrite θ = lh(E).The spaceof E is

sp(E)= sup{µa | a ∈ [lh(E)]<ω }.

The domainof E isthefamilyofsetsitmeasures,thatis, dom(E)= {Y |∃(a, X ) ∈

E(Y = X ∨ Y =[µa]|a| X )}.If M isapremouseofsomekind,wealsowrite M|η = dom(E),where η isleastsuchthat ∀(a, X ) ∈ E(X ∈ M|η ).Byacceptability, η = sup({µ +,M a | a ∈ [θ ]<ω }).Weshallfurtherabusenotationbywriting η = dom(E) when M isdeterminedbycontext.

Thecriticalpointofa (κ , θ )-extenderis κ ,andweuseeither crit(E) or κE to denoteit.Givenanextender E over M,weformthe Σ0 ultrapower

Ult0(M, E)= {[a, f ]M E | a ∈ [lh(E)]<ω and f ∈ M}, asin[49,8.4].Our M willalwaysberudimentarilyclosedandsatisfytheAxiom ofChoice,sowehaveLos’theoremfor Σ0 formulae,andthecanonicalembedding

iM E : M → Ult0(M, E) iscofinaland Σ0 elementary,andhence Σ1 elementary.By(1)and(3), κE = crit(iM E ).Bynormality, a =[a, id]M E ,so lh(E) isincludedinthe(alwaystransitivized)wellfoundedpartofUlt0(M, E).Moregenerally,

[a, f ]M E = iM E ( f )(a)

If X ⊆ lh(E),then E X = {(a,Y ) ∈ E | a ⊆ X } E X hasthepropertiesofan extender,exceptpossiblynormality,sowecanform Ult0(M, E X ),andthereisa naturalfactorembedding τ :Ult0(M, E X ) → Ult0(M, E) givenby τ ([a, f ]M E X )=[a, f ]M E

Inthecasethat X = ν > κE isanordinal, E ν isanextender,and τ ν isthe identity.Wesay ν isa generatorof E iff ν isthecriticalpointof τ ,thatis, ν =[a, f ]M E whenever f ∈ M and a ⊆ ν .Let

ν (E)= sup({ν + 1 | ν isageneratorof E }).

So ν (E) ≤ lh(E),and E isequivalentto E ν (E),inthatthetwoproducethesame ultrapower.

Wewrite

λ (E)= λE = iM E (κE ).

Notethatalthough E maybeanextenderovermorethanone M, sp(E), κE , lh(E), dom(E), ν (E),and λ (E) dependonlyon E itself.If N isanothertransitive, rudimentarilyclosedset,and P(µa) ∩ N = P(µa) ∩ M forall a ∈ [lh(E)]<ω ,then E isalsoanextenderover N;moreover iM E agreeswith iN E on dom(E).However, iM E and iN E maydisagreebeyondthat.Wesay E is short iff ν (E) ≤ λ (E).Itiseasyto

seethat E isshortif lh(E) ≤ sup(iM E “((κ + E )M )).If E isshort,thenallitsinteresting measuresconcentrateonthecriticalpoint.When E isshort, iM E iscontinuousat κ +,M ,andif M isapremouse,then dom(E)= M|κ +,M E .Inthisbook,weshalldeal almostexclusivelywithshortextenders.

Ifwestartwith j : M → N withcriticalpoint κ ,andanordinal ν suchthat κ < ν ≤ o(N),thenfor a ∈ [ν ]<ω welet µa betheleast µ suchthat a ⊆ j(µ ),and for X ⊆ [µa]|a| in M,weput

(a, X ) ∈ E j ⇐⇒ a ∈ j(X ).

E j isanextenderover M,calledthe (κ , ν )-extenderderivedfrom j.Wehavethe diagram

i where i = iM E j ,and k(i( f )(a))= j( f )(a)

Ult0(M, E j )

k ν istheidentity.If E isanextenderover M,then E isderivedfrom iM E

The Jensencompletion ofashortextender E oversome M isthe (κE , iM E (κ +,M E )) extenderderivedfrom iM E . E anditsJensencompletion E∗ areequivalent,inthat

ν (E)= ν (E∗),and E = E∗ lh(E).

2.2.Pureextenderpremice

Ourmainresultsapplytopremiceofvariouskinds,bothstrategypremice andpureextenderpremice,with λ -indexingorms-indexingfortheirextender sequences.31,32 Thecomparisontheoremforiterationstrategiesthatisourfirst maingoalholdsinallthesecontexts.Buttheproofofthistheoremrequiresa detailedfinestructuralanalysis,andtheparticularsofthefinestructurebecome importantatcertainpoints.Weshallfirstprovethecomparisontheoreminthecase ofiterationstrategiesforwhatweshallcall pfs premice.Theseareavarianton pureextenderpremicewith λ -indexing,thedifferencebeingthatthesoundness

31Why“mouse”?Like“quark”,itisshortandeasilyremembered.Ithasafinetradition,goingback toadiscovereroftheconcept.Thelonger,colorless“extendermodel”doeshaveitsplace,but“mouse” ismoreflexibleanddistinctive.

32Peoplesometimesspeakof“strategic”miceorextendermodels,butthisseemswrongtous.A strategicX(bomber,position,move,etc.)isanXthatisincorporatedinsomestrategy,notanXthat hasastrategyincorporatedinit.Amousethatincorporatesastrategyisastrategymouse,justasapie thatincorporatesapplesisanapplepie.

requirementhasbeenrelaxed.TheyareformallydefinedinChapter4.Untilweget tothatchapter,weshalldealprimarilywiththestandard λ -indexedpureextender premice,asdefinedin[81].33

Thereadershouldsee[4,Def.2.4]forfurtherdetailsonthefollowingdefinition. A potentialJensenpremouse isanacceptableJ-structure

M = JE α , ∈, E, γ , F

withvariousproperties. o(M)= OR ∩ M = ωα .Thelanguage L0 of M has ∈, predicatesymbols ˙ E and ˙ F ,andaconstantsymbol γ .Wecall L0 the languageof (pureextender)premice. If M isapotentialJensenpremouse,then E M isasequenceofextenders,and either ˙ F M isempty(i.e. M is passive),or ˙ F M codesanewextenderbeingaddedto ourmodelby M.Themainrequirementsare

(1) (λ -indexing)If F = ˙ F M isnonempty(i.e., M is active),then M |= crit(F )+ exists,andfor µ = crit(F )+M , o(M)= iM F (µ )= lh(F ). ˙ F M isjustthegraph of iM F (M|µ )

(2) (Coherence) iM F ( ˙ E M ) o(M)+ 1 = ˙ E M /0

(3) (Initialsegmentcondition, J-ISC)If G isawholeproperinitialsegmentof F ,thentheJensencompletionof G mustappearin ˙ E M .Ifthereisalargest wholeproperinitialsegment,then ˙ γ M istheindexofitsJensencompletion in ˙ E M .Otherwise, ˙ γ M = 0.

(4) If N isaproperinitialsegmentof M,then N isapotentialJensenpremouse. Hereaninitialsegment G = F η of F is whole iff η = λG

SincepotentialJensenpremiceareacceptable J-structures,thebasicfinestructuralnotionsapplytothem.Werecallsomeoftheminthenextsection.Wethen definea Jensenpremouse asapotentialJensenpremouseallofwhoseproperinitial segmentsaresound.

Figure2.2.1illustratesacommonsituation,onethatoccursatsuccessorstepsin aniterationtree,forexample.

ThereisasignificantstrengtheningoftheJenseninitialsegmentcondition(3) above.If M isanactivepremouse,thenweset

ν (M)= max(ν (F M ), crit(F M )+,M ).

˙ F M ν (M) isequivalentto ˙ F M ,andsoitisnotin M.But

DEFINITION 2.2.1. Let M beanactivepremousewithlastextender F ;then M satisfiesthe ms-ISC (oris ms-solid)iffforany η < ν (M), F η ∈ M.

Clearlythe ms-ISC impliestheweakeningof J-ISC inwhichweonlydemand thatthewholeproperinitialsegmentsof ˙ F M belongto M.Butforiterable M,this thenimpliesthefull J-ISC.(See[48].)

33Theessentialequivalenceof λ -indexingwithms-indexinghasbeencarefullydemonstratedby Fuchsin[11]and[10].

(N, E)

FIGURE 2.2.1. E isonthecoherentsequenceof M, κ = crit(E),and λ = λ (E). P(κ )M = P(κ )N = dom(E),so

Ult0(M, E) and Ult0(N, E) makesense.Theultrapowersagree with M belowlh(E),andwitheachotherbelowlh(E)+ 1.

THEOREM 2.2.2(ms-ISC). Let M beanactivepremousewithlastextender F , andsupposeMis1-soundand (1, ω , ω1 + 1)-iterable;thenMisms-solid.

Thisisessentiallytheinitialsegmentconditionof[30],butstatedforJensen premice.[30]goesontosaythatthetrivialcompletionof F η iseitheronthe M-sequence,oranultrapoweraway.Thisiscorrectunless F η istypeZ.If F η istypeZ,thenitistheextenderof F ξ -then-U ,where ξ isitslargestgenerator, and U isanultrafilteron ξ ,andwestillget F η ∈ M.(See[48].Theorem2.7of [48]isessentially2.2.2above.)

If M isactive,weletits initialsegmentordinal be ι (M)= sup({η + 1 | F M η ∈ M}).

So M isms-solidiff ι (M)= ν (M).Theorem2.2.2becomesfalsewhenitssoundnesshypothesisisremoved,sinceif N = Ult0(M, E) where ν (M) ≤ crit(E) < λF , then ι (N)= ι (M)= ν (M),butcrit(E) isageneratorof iM E (F ).

TheproofofTheorem2.2.2requiresacomparisonargumentbasedoniterability, andsointhecontextofthisbook,itisclosertotheendofthedevelopmentthanto thebeginning.Inthetheoryof[30],thestrongformofms-solidityisanaxiomon premicefromthebeginning,butcomparisonargumentsareneededtoshowthat thepremiceoneconstructssatisfyit.34 Onemightsimilarlymakems-solidityan axiomonJensenpremicefromthebeginning,andsohaveitavailableearlierinthe

34Thesearetheresultsof[30, §9, §10]concerningbicephaliandpsuedo-premice.

game,sotospeak.Thiswouldsimplifyafewthings,butitisnotstandard,andwe shallnotdoithere.35

Weshallnotusems-premice,sohenceforthweshallrefertopotentialJensen premiceaspotentialpremice,orlater,whenweneedtodistinguishthemfrom strategypremice,as potentialpureextenderpremice.

2.3.Projectaandcores

Finestructuretheoryreliesonacarefulanalysisofthecondensationproperties ofmice;thatis,oftheextenttowhichSkolemhullsofamouse M collapseto initialsegmentsof M.Jensen’stheoryofprojecta,standardparameters,andcores isthefoundationforthisanalysis.

Soundpremiceandtheirreducts

Let M beanacceptable J-structure.36 Wedefinetheprojecta ρi(M),standard parameters pi(M),andreducts(“Σi mastercodes”) Mi = Mi,pi (M) byinduction.At thesametimewedefine k-solidityand k-soundnessfor M.Westartwith A0 = /0 , and M0 =(M, A0), ρ0(M)= o(M), p0(M)= /0 .

M isautomatically0-soundand0-solid.Thesuccessorstepis

ρi+1(M)= least α s.t. ∃A ⊆ α (A isboldface ΣMi 1 and A / ∈ Mi), = ρ1(Mi) and pi+1(M)= pi(M) ∪ ri+1, where ri+1 isthelexicographicallyleastdescendingsequenceofordinalsfrom whichanewsubsetof ρ1(Mi) canbe Σ1 definedover Mi.(ri+1 = /0 ispossible.)

Wethenset Mi+1 =(M||ρi+1(M), Ai+1), where Ai+1 = ThMi 1 (Mi||ρi+1 ∪ ri+1) = { ϕ , x | ϕ (u, v) is Σ1 ∧ Mi |= ϕ [x, ri+1]}.

Wesaythat M is i + 1-solid iff ri+1 issolidanduniversal,andsoisitsimagein

35Ourdefinitionof pfs premiceinChapter4doesincludeasmallfragmentofms-solidity.See4.1.1. Doddsolidity isausefulandstillstrongerformoftheinitialsegmentconditionthatiterablepremice satisfy.Seeforexample[48].

36See[49].

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.