/
Relationbetweeno-equivalenceandEA-equivalenceforNihobentfunctionsD Relationbetweeno-equivalenceandEA-equivalenceforNihobentfunctionsD

Relationbetweeno-equivalenceandEA-equivalenceforNihobentfunctionsD - PDF document

alis
alis . @alis
Follow
342 views
Uploaded On 2021-01-05

Relationbetweeno-equivalenceandEA-equivalenceforNihobentfunctionsD - PPT Presentation

3SomeresultsofthispaperwerepresentedatIrsee2014conferenceBFA2018andBFA2019workshopsyDepartmentofInformaticsUniversityofBergenPB7803N5020BergenNorwayemailfDianaDavidovaLilyaBudaghyanTo ID: 826614

hkf x0000 proof 2jf x0000 hkf 2jf proof polynomial ift trm polynomials equivalence nedbyf letfbeano 1and vol nedby uptoea

Share:

Link:

Embed:

Download Presentation from below link

Download Pdf The PPT/PDF document "Relationbetweeno-equivalenceandEA-equiva..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.


Presentation Transcript

Relationbetweeno-equivalenceandEA-equiva
Relationbetweeno-equivalenceandEA-equivalenceforNihobentfunctionsDianaDavidovay,LilyaBudaghyan;yClaudeCarletz,yTorHelleseth;yFerdinandIhringerx,TimPenttila{AbstractBooleanfunctions,andbentfunctionsinparticular,areconsidereduptoso-calledEA-equivalence,whichisthemostgeneralknownequiv-alencerelationpreservingbentnessoffunctions.However,foraspecialtypeofbentfunctions,so-calledNihobentfunctionsthereisamoregen-eralequivalencerelationcalledo-equivalencewhichisinducedfromtheequivalenceofo-polynomials.Inthepresentworkwestudy,foragiveno-polynomial,ageneralconstructionwhichprovidesallpossibleo-equivalentNihobentfunctions,andweconsiderablysimplifyittoaformwhichexcludesEA-equivalentcases.Thatis,weidentifyallcaseswhichcanpotentiallyleadtopairwiseEA-inequivalentNihobentfunctionsderivedfromo-equivalenceofanygivenNihobentfunction.Furthermore,wede-termineallpairwiseEA-inequivalentNihobentfunctionsarisingfromallknowno-polynomialsviao-equivalence.Keywords:Bentfunction,Booleanfunction,EA-equivalence,maxi-mumnonlinearity,Magicaction,modi edMagicaction,Nihobentfunc-tion,o-equivalence,o-polynomials,ovals,hyperovals,Walshtransform.1IntroductionBooleanfunctionsofnvariablesarebinaryfunctionsoverthevectorspaceFn2ofallbinaryvectorsoflengthn,andcanbeviewedasfunctionsovertheGalois eldF2n,thankstothechoiceofabasisofF2noverF2.Inthispaper,weshallalwayshavethislastviewpoint.Booleanfunctionsareusedinthepseudo-randomgeneratorsofstreamciphersandplayacentralroleintheirsecurity.Bentfunctions,introducedbyRothaus[35]in1976,areBooleanfunctionshavinganevennumberofvariablesn,thataremaximallynonlinearinthesensethattheirnonlinearity,theminimumHammingdistancetoallanefunctions,SomeresultsofthispaperwerepresentedatIrsee2014conference,BFA2018andBFA2019workshops.yDepartmentofInformatics,UniversityofBergen,PB7803,N-5020Bergen,Norway,e-mail:fDiana.Davidova,Lilya.Budaghyan,Tor.Hellesethg@uib.nozDepartmentofMathematics,UniversitiesofParis8andParis13,2ruedelaliberte,93526Saint-DenisCedex,France,e-mail:claude.carlet@univ-paris8.frxDepartmentofMathematics:Analysis,LogicandDiscreteMathematics,GhentUniver-sity,Belgium,e-mail:ferdinand.ihringer@ugent.be{DepartmentofMath

ematics,ColoradoStateUniversity,FortColl
ematics,ColoradoStateUniversity,FortCollins,CO80523-1874,e-mail:penttila@math.colostate.edu1isoptimal(formoreinformationonbentfunctionssee,forinstance,[11]).ThiscorrespondstothefactthattheirWalshtransformtakesthevalues2n=2,only.Bentfunctionshaveattractedalotofresearchinterestinmathematicsbecauseoftheirrelationtodi erencesetsandtodesigns,andintheapplicationsofmathematicstocomputersciencebecauseoftheirrelationstocodingtheoryandcryptography.Despitetheirsimpleandnaturalde nition,bentfunctionsadmitaverycomplicatedstructureingeneral.Animportantfocusofresearchisto ndconstructionsofbentfunctions.Manymethodsareknownandsomeofthemallowexplicitconstructions.Wedistinguishbetweenprimaryconstruc-tionsgivingbentfunctionsfromscratchandsecondaryconstructionsbuildingnewbentfunctionsfromoneorseveralgivenbentfunctions(inthesamenumberofvariablesorindi erentones).Booleanfunctions,andbentfunctionsinparticular,areconsidereduptoso-calledEA-equivalence,whichisthemostgeneralknownequivalencerelationpreservingbentnessoffunctions[3,4].Bentfunctionsareoftenbetterviewedintheirbivariaterepresentation,intheformf(x;y),wherexandybelongtoFm2ortoF2m,wherem=n=2.ThisrepresentationhasledtothetwogeneralfamiliesofexplicitbentfunctionswhicharetheoriginalMaiorana-McFarland[29]andthePartialSpreads(PSap)classes(thislatterclassisincludedinthemoregeneralbutlessexplicitPSclass).Bentfunctionscanalsobeviewedintheirunivariateform,expressedbymeansofthetracefunctionoverF2n.Findingexplicitbentfunctionsinthistracerepresen-tationisusuallymoredicultthaninthebivariaterepresentation.Referencescontaininginformationonexplicitprimaryconstructionsofbentfunctionsintheirbivariateandunivariateformsare[9,25].ItiswellknownthatsomeoftheseexplicitconstructionsbelongtotheMaiorana-McFarlandclassandtothePSapclass.When,intheearly1970s,Dillonintroducedinhisthesis[17]thetwoabovementionedclasses,healsointroducedanotheronedenotedbyH,wherebentnesswasprovenundersomeconditionswhichwerenotobvioustoachieve.ThismadeclassHanexampleofanon-explicitconstruction:atthattime,Dillonwasabletoexhibitonlyfunctionsbelonging,uptotheaneequivalence(whichisaparticularcaseofEA-equivalence),totheMaiorana-McF

arlandclass.Itwasobservedin[10]thatthecl
arlandclass.Itwasobservedin[10]thattheclassofthe,socalled,Nihobentfunctions(introducedin[18]byDobbertinetal)is,uptoEA-equivalence,equaltotheDillon'sclassH.NotethatfunctionsinclassHarede nedintheirbivariaterepresentationandNihobentfunctionshadoriginallyaunivariateformonly.Threein nitefamiliesofNihobinomialbentfunctionswereconstructedin[18]andoneoftheseconstructionswaslatergeneralizedbyLeanderandKholosha[26]intoafunctionwith2rNihoexponents.Anotherclasswasalsoextendedin[20].In[6]itwasproventhatsomeofthesein nitefamiliesofNihobentfunc-tionsareEA-inequivalenttoanyMaiorana-McFarlandfunctionwhichimpliedthatclassesHandMaiorana-McFarlandaredi erentuptoEA-equivalence.Inthesamepaper[10],itwasshownthatNihobentfunctionsde neo-polynomialsand,conversely,everyo-polynomialde nesaNihobentfunction.Theyalsodiscoveredthatagiveno-polynomialFcanproducetwodi erent(uptoEA-equivalence)Nihobentfunctions,namely,theonesderivedfromFanditsinverseF�1.Sincetakingtheinverseofano-polynomialisaparticularcaseoftheequivalenceofo-polynomials,anaturalquestionwastoexplorethisequivalenceforconstructionoffurtherEA-inequivalentcasesofNihobentfunctions.The rstworkinthisdirectionwasdonein[7]wherethegroup2oftransformations(introducedin[14])oforder24preservingtheequivalenceofo-polynomialswasstudiedforrelationtoEA-equivalence.ItwasshownthatthesetransformationscanleadtouptofourEA-inequivalentfunctionsincludingthosederivedfromano-polynomialanditsinverse.Thatis,twonewtransformationswhichcanpotentiallyprovideEA-inequivalentfunctionsfromagiveno-polynomialwerediscovered.Hence,applicationoftheequivalenceofo-polynomialscanbeconsideredasaconstructionmethodfornew(uptoEA-equivalence)Nihobentfunctionsfromtheknownones.Notethatthegroupoftransformationsfrom[14]doesnotcoverallpossi-bletransformationswithinequivalenceofo-polynomials.Amoregeneralgroupoftransformations,so-calledtheMagicaction,waspresentedin[21],whichisanactionofagroupoftransformationsactingonprojectivelineonthesetofo-permutations.Inthispaperwestudythemodi edMagicaction,atrans-formationofo-polynomialspreservingprojectiveequivalence.Weshowthato-polynomialsareprojectivelyequivalentifandonlyiftheylie

onthesameorbitunderthemodi edMagicac
onthesameorbitunderthemodi edMagicactionandtheinversemap.Furtherweprovethat,foragiveno-polynomial,EA-inequivalentNihobentfunctionscanariseonlyfromaspeci cformulainvolvingparticularcompositionsoftransforma-tionsofthemodi edmagicactionandtheinversemap.Weshowthateacho-monomialcande neuptofourEA-inequivalentbentfunctions.Weprove,forinstance,thatthePynehyperovalcangiverisetoEA-inequivalentNihobentfunctionsde nedbyo-polynomialswhichlieon3di erentorbitsofthemodi edMagicaction.Foreveryknowno-polynomialweprovideanexplicitnumberofpairwiseEA-inequivalentNihobentfunctionswhichcanbederivedviao-equivalence.Moreover,wegiveanexplicitdescription(involvingtransfor-mationsofthemodi edmagicactionandtheinversemap)ofallo-polynomialsprovidingpairwiseEA-inequivalentNihobentfunctions.Thepaperisorganizedasfollows.InSection2werecallnecessaryback-ground,inSection3wede neNihobentfunctionsviao-polynomialsandviceversa.InSection4weprovethataneequivalenceofo-polynomialsinsomecasesyieldsEA-equivalenceofthecorrespondingNihobentfunctions.Theknownfactthateveryo-polynomialonF2mnecessarilyde nesavectorialNihobentfunctionfromF22mtoF2mcanbeseenasacorollary.InSection5themod-i edmagicactionisintroducedanditisproventhatpotentiallyEA-inequivalentNihobentfunctionscanarisefromo-polynomialswhichlieonthesameorbitunderthemodi edMagicactionandtheinversemap.ThemainresultsofthepaperarecontainedinSections6and7,whereweobtainanexactformoftheorbitonwhicho-polynomialsshouldlietoproducepotentiallyEA-inequivalentNihobentfunctions.Foreachoftheknowno-polynomialsweprovidetheex-plicitnumberandrepresentationsforallequivalento-polynomialswhichprovidepairwiseEA-inequivalentNihobentfunctions.32NotationandPreliminaries2.1TraceRepresentation,BooleanFunctionsinUnivari-ateandBivariateFormsForanypositiveintegerkandanyrdividingk,thetracefunctionTrkristhemappingfromF2ktoF2rde nedbyTrkr(x):=kr�1Xi=0x2ir=x+x2r+x22r++x2k�r:Inparticular,theabsolutetraceoverF2kisthefunctionTrk1(x)=Pk�1i=0x2i(inwhatfollows,wejustuseTrktodenotetheabsolutetrace).Recallthatthetracefunctionsatis esthetransitivitypropertyTrk=TrrTrkr.Theunivariaterep

resentationofaBooleanfunctionisde ne
resentationofaBooleanfunctionisde nedasfollows:weidentifyFn2(then-dimensionalvectorspaceoverF2)withF2nandconsidertheargumentsoffaselementsinF2n.AninnerproductinF2nisxy=Trn(xy).ThereexistsauniqueunivariatepolynomialP2n�1i=0aixioverF2nthatrepresentsf(thisistrueforanyvectorialfunctionfromF2ntoitselfandthereforeforanyBooleanfunctionsinceF2isasub eldofF2n).Thealgebraicdegreeoffisequaltothemaximum2-weightoftheexponentsofthosemonomialswithnonzerocoecientsintheunivariaterepresentation,wherethe2-weightw2(i)ofanintegeriisthenumberofonesinitsbinaryexpansion.Moreover,fbeingBoolean,itsunivariaterepresentationcanbewrittenuniquelyintheformoff(x)=Xj2�nTro(j)(ajxj)+a2n�1x2n�1;where�nisthesetofintegersobtainedbychoosingthesmallestelementineachcyclotomiccosetmodulo2n�1(withrespectto2),o(j)isthesizeofthecyclotomiccosetcontainingj,aj2F2o(j)anda2n�12F2.Thefunctionfcanalsobewritteninanon-uniquewayasTrn(P(x))whereP(x)isapolynomialoverF2n.ThebivariaterepresentationofaBooleanfunctionisde nedinthispaperasfollows:weidentifyFn2withF2mF2m(wheren=2m)andconsidertheargumentoffasanorderedpair(x;y)ofelementsinF2m.ThereexistsauniquebivariatepolynomialP0i;j2m�1ai;jxiyjoverF2mthatrepresentsf.Thealgebraicdegreeoffisequaltomax(i;j)jai;j6=0(w2(i)+w2(j)).AndfbeingBoolean,itsbivariaterepresentationcanbewrittenintheformf(x;y)=Trm(P(x;y)),whereP(x;y)issomepolynomialoftwovariablesoverF2m.Remark1.Letg(x;y)beaBooleanfunctionoverF2mF2m.Thenonecangetaunivariaterepresentationofgmakingthefollowingsubstitutions:x=t+t2mandy= t+( t)2m;where isaprimitiveelementofF22m.2.2WalshTransformandBentFunctionsLetfbeann-variableBooleanfunction.Its\sign"functionistheinteger-valuedfunctionf:=(�1)f.TheWalshtransformoffisthediscreteFourier4transformoffwhosevalueatpointw2F2nisde nedbybf(w)=Xx2F2n(�1)f(x)+Trn(wx):Forevenn,aBooleanfunctionfinnvariablesissaidtobebentifforanyw2F2nwehavebf(w)=2n2.Itiswellknown(see,forinstance,[9])thatthealgebraicdegreeofabentBooleanfunctioninn�2variablesisatmostn2.Bentnessandalgebraicdegree(whenlargerthan1)arepreservedbyextended-ane(EA-)equivalence.TwoBool

eanfunctionsfandginnvariablesarecalledEA
eanfunctionsfandginnvariablesarecalledEA-equivalentifthereexistsananepermutationAofF2nandananeBooleanfunction`suchthatf=gA+`.Ifl=0thenfandgarecalledaneequivalent.Inthecaseofvectorialfunctionsthereexistsamoregeneralnotionofequivalence,calledCCZ-equivalence,butforBooleanfunctions,itreducestoEA-equivalence,see[3](aswellasforbentvectorialfunctions[4]).TwofunctionsFandF0fromF2ntoitselfarecalledEA-equivalentifA1FA2+AforsomeanepermutationsA1andA2andforsomeanefunctionA.IfA=0thenFandF0arecalledaneequivalent.Forpositiveintegersnandt,avectorialBooleanfunctionFfromFn2toFt2iscalledbentifforanya2Fn2nf0gtheBooleanfunctionaF(x)isbent.Bentfunctionsexistifandonlyifnisevenandtn=2(see[30]).2.3Projectiveplane,Ovals,HyperovalsInthefollowingwegiveashortintroductiontotheprojectiveplane.Wereferto[16]foradetailedintroductiontoprojectivegeometry.AprojectiveplaneconsistsofasetofpointsP,asetoflinesL,andanincidencerelationIbetweenPandL.TheclassicalprojectiveplanePG(2;q)overF3qhasthe1-spacesofF3qaspointsandthe2-spacesofF3qaslines.Apointpiscontainedinaline`ifp`inF3q.Asetofpointsiscalledcollineariftheyalllieonthesameline.NotethatPG(2;q)hasq2+q+1points,q2+q+1lines,eachlinecontainsq+1points,andeachpointliesinq+1lines.ThegroupP�L(3;q)actsnaturallyonPG(2;q).Inparticular,itpreservesincidence.LetObeasetofpointsinPG(2;q)suchthatnothreepointsarecollinear.Itiswell-knownthatjOjq+1ifqisoddandjOjq+2isqiseven.Onecanseethisasfollows:ConsiderapointP2O.Eachoftheq+1linesonPcontainsatmostonemorepoints,sojOjq+2.Supposethatequalityholds.Theneachlinecontainseither0or2points.ConsiderapointR2O.ThenthereareslinesthroughRwith2pointsandq+1�slinesthroughRwith0points.Hence,q+2=2s,soqiseven.Callaline`passant,tangent,respectively,secantifj`\Oj=0,j`\Oj=1,respectively,j`\Oj=2.IfjOj=q+1,thenOiscalledanoval.FromtheargumentaboveitfollowsthatinthiscaseeachpointofOliesonexactlyonetangentandqsecants.ForqeventhesesecantsallmeetinonepointN,thenucleusofO.IfjOj=q+2,thenOiscalledahyperovalandweusuallywriteHinsteadofO.IfjOj=q+1andqeven,thenO[fNgisahyperoval.Inthefollowingwelimitourselvestoq=2meven.AframeofPG(2;q)isasetoffourpointsP=

fP1;P2;P3;P4gsuchthatany3-subsetofPspans
fP1;P2;P3;P4gsuchthatany3-subsetofPspansF3q.Thefundamentaltheoremofprojectivegeometry(for5projectiveplanes)statesthatP�L(3;q)actstransitiveonframes.AsanyfourpointsofahyperovalHareaframe,wecanassumethatanovalOcontainsh(1;0;0)i;h(0;0;1)i;h(1;1;1)i2Oandhash(0;1;0)iasitsnucleus.Inthefollowingweusuallyleaveoutthebracketshiforthesakeofreadibility.Hence,wecanwriteOasO=f(x;F(x);1):x2F2mg[f(1;0;0)g;wherethepolynomialFsatis esthefollowing:(a)FisapermutationpolynomialoverF2mofdegreeatmostq�2satisfyingF(0)=0andF(1)=1.(b)Foranys2F2mthefunctionFs(x):=(F(x+s)+F(x)xifx6=0;0otherwise.isapermutationpolynomial.HereandfurtherinthepaperwedenoteF2m=F2mnf0g.SuchapolynomialFiscalledano-polynomialand,conversely,eacho-polynomialde nesanoval.IfwedonotrequireF(1)=1,thenFiscalledano-permutation.WewriteO(F)fortheovalde nedbytheo-polynomialF,andwewriteH(F)forthehyperovalde nedbyF.NotethatthroughoutthispaperOconsistsofpointsoftheform(x;F(x);1),whileinthehyperplaneliterature,usuallytheform(1;x;f(x))isused.ForahyperovalHwehave2m+2choicesforthenucleusN2Htoob-tainanovalHnfNg.Hence,eachhyperovalHde nes2m+2o-polynomials.Twoo-polynomialsarecalled(projectively)equivalent,iftheyde neequivalenthyperovals(underthenaturalactionofP�L(3;q)).2.4NihoBentFunctionsApositiveintegerd(alwaysunderstoodmodulo2n�1withn=2m)isaNihoexponentandt!tdisaNihopowerfunctioniftherestrictionoftdtoF2mislinearor,equivalently,ifd2j(mod2m�1)forsomejn.AsweconsiderTrn(atd)witha2F2n,withoutlossofgenerality,wecanassumethatdisinthenormalizedform,i.e.,withj=0.Thenwehaveauniquerepresentationd=(2m�1)s+1with2s2m.Ifsomesiswrittenasafraction,thishastobeinterpretedmodulo2m+1(e.g.,1=2=2m�1+1).FollowingareexamplesofbentfunctionsconsistingofoneormoreNihoexponents:1.QuadraticfunctionTrm(at2m+1)witha2F2m(heres=2m�1+1).2.Binomialsoftheformf(t)=Trn( 1td1+ 2td2),where2d12m+1(mod2n�1)and 1; 22F2naresuchthat( 1+ 2m1)2= 2m+12.Equivalently,denotinga=( 1+ 2m1)2andb= 2wehavea=b2m+12F2mandf(t)=Trm(at2m+1)+Trn(btd2):6Wenotethatifb=0anda6=0thenfisabentfunctionlistedundernumber1.Thepossiblevaluesofd2

are[18,20]:d2=(2m�1)3+1;6d2=(2m�1)
are[18,20]:d2=(2m�1)3+1;6d2=(2m�1)+6(takingmeven):Thesefunctionshavealgebraicdegreemanddonotbelongtothecom-pletedMaiorana-McFarlandclass[6].3.Take1rmwithgcd(r;m)=1andde nef(t)=Trn a2t2m+1+(a+a2m)2r�1�1Xi=1tdi!;(1)where2rdi=(2m�1)i+2randa2F2nissuchthata+a2m6=0[26,27].Thisfunctionhasalgebraicdegreer+1(see[5])andbelongstothecompletedMaiorana-McFarlandclass[12].4.Bentfunctionsinabivariaterepresentationobtainedfromtheknowno-polynomials.Considerthelistedabovetwobinomialbentfunctions.Ifgcd(d2;2n�1)=dandb= dforsome 2F2nthenbcanbe\absorbed"inthepowertermtd2byalinearsubstitutionofvariablet.Inthiscase,uptoEA-equivalence,b=a=1.Inparticular,thisappliestoanybwhengcd(d2;2n�1)=1thatholdsinbothcasesexceptwhend2=(2m�1)3+1withm2(mod4)whered=5.Inthisexceptionalcase,wecangetupto5di erentclassesbuttheexactsituationhastobefurtherinvestigated.3ClassHofBentFunctionsando-polynomialsHerewerestrictourselfwith eldsF2nwithneven,n=2m.Inhisthesis[17],DillonintroducedtheclassofbentfunctionsdenotedbyH.Thefunctionsinthisclassarede nedintheirbivariateformasf(x;y)=Trm(y+xF(yx2m�2));wherex;y2F2m,andFisapermutationofF2ms.t.F(x)+xdoesn'tvanish,forany 2F2mthefunctionF(x)+ xis2-to-1.DillonwasabletoexhibitbentfunctionsinHthatalsobelongtothecompletedMaiorana-McFarlandclass.Dillon'sclassHwasmodi edin[10]intoaclassHofthefunctions:g(x;y)=8:TrmxGyx;ifx6=0Trm(y);otherwise(2)7where2F2m;G:F2m7!F2msatisfyingthefollowingconditions:F:z7!G(z)+zisapermutationoverF2m;(3)z7!F(z)+ zis2-to-1onF2mforany 2F2m:(4)Herecondition(4)impliescondition(3)anditisnecessaryandsucientforgbeingbent.FunctionsinHandtheDillonclassarethesameuptoadditionofalineartermTrm((+1)y)to(2).NihobentfunctionsarefunctionsinHintheirunivariantrepresentation.Theorem1([10]).ApolynomialFonF2msatisfyingF(0)=0andF(1)=1isano-polynomialifandonlyifz7!F(z)+ zis2-to-1onF2mforany 2F2m:(5)Hence,obviouslyeveryo-polynomialde nesaNihobentfunction.Andviceversa,everyNihobentfunctionde nesano-polynomialsinceitde nesapolynomialFsatisfyingcondition(5)ofTheorem1,andwecanderiveano-polyn

omialF0(x)=F(x)+F(0)F(1)+F(0)which
omialF0(x)=F(x)+F(0)F(1)+F(0)which xestherequirementsF0(0)=0andF0(1)=1.NotethattogetaNihobentfunctionfromapolynomialFitissucientthatFsatis esonlycondition(5)whiletheconditionsF(0)=0andF(1)=1arenotnecessary.InSection2.3wesawthateacho-polynomialcorrespondstoahyperovalandviceversa,eachhyperovalcorrespondstoano-polynomial.WesaythatNihobentfunctionsareo-equivalentiftheyde neprojectivelyequivalenthyperovals.Asshownin[7,10],o-equivalentNihobentfunctionsmaybeEA-inequivalent.Forexample,Nihobentfunctionsde nedbyo-polynomialsFandF�1areo-equivalentbuttheyare,ingeneral,EA-inequivalent.Hereisthelistofallknowno-polynomials(wealsogivenamesofthecorre-spondinghyperovals):1.F(x)=x2,regularhyperoval;2.F(x)=x2i,iandmarecoprime,i�1,irregulartranslationhyperoval;3.F(x)=x6,misodd,Segrehyperoval;4.F(x)=x32k+4,m=2k�1,GlynnI;5.F(x)=x2k+22k,m=4k�1,GlynnII;6.F(x)=x22k+1+23k+1,m=4k+1,GlynnII;7.F(x)=x2k+x2k+2+x32k+4,m=2k�1,Cherowitzohyperoval;8.F(x)=x16+x12+x56,misodd,Paynehyperoval;9.F(x)=2(x4+x)+2(1++2)(x3+x2)x4+2x2+1+x12,whereTrm(1)=1(ifm2(mod4),then=2F4),Subiacohyperoval(form=4alsoknownasLunelli-Scehyperoval);810.F(x)=1Trnm(v)Trnm(vr)(x+1)+(x+Trnm(v)x12+1)1�rTrnm(vx+v2m)r+x12,wheremiseven,r=2m�13,v2F22m;v2m+16=1;v6=1,Adelaidehyperoval.11.F(x)=x4+x16+x28+!11(x6+x10+x14+x18+x22+x26)+!20(x8+x20)+!6(x12+x24)with!5=!2+1andm=5,O'Keefe-Penttilahyperoval.Notethatano-polynomialFde nedonF2mhasthefollowingform[16]:F(x)=2m�22Xk=1b2kx2k:4VectorialNihobentfunctionsfromo-polynomialsItisknownsince2011thateveryo-polynomialde nesaBooleanNihobentfunc-tion[10].Inthissection,werevisitthefactthat,actually,everyo-polynomialonF2mde nesavectorialNihobentfunctionfromF2mF2mtoF2m.Thisconnectionhasbeenoriginallyobservedin[28].Inthepresentpaper,wederivethisresultbystudyingsomesimpletransformationsofo-polynomials.Belowweshowthatinsomecases,aneequivalenceofo-polynomialsyieldsEA-equivalenceofthecorrespondingNihobentfunctions.NotethatingeneralifafunctionF0isaneequivalenttoano-polynomialFthenF0isnotnecessarilyano-polynomial.Lemma1.LetFbeano-polynomi

alde nedonF2manda;b2F2m.ThenG(x)
alde nedonF2manda;b2F2m.ThenG(x)=aF(bx)isano-polynomialonF2mifandonlyifa=1F(b)(or,whatisthesame,b=F�1(a�1)).TheNihobentfunctionsde nedbytheo-polynomialsFandG=1F(b)F(bx)areaneequivalent.Proof.SupposeG(x)=aF(bx)isano-polynomial,thenG(0)=aF(0)=0foranya;b2F2mand1=G(1)=aF(b),henceGisano-polynomialifandonlyifa=1F(b).TheNihobentfunctioncorrespondingtotheo-polynomialFisf(x;y)=Trm(xF(yx)),andtheonecorrespondingtoGisg(x;y)=Trm(xG(yx))=Trm(xaF(byx))=Trm(xaF(abyax))=Trm(vF(uv));wherev=ax,u=aby.Hence,g=fAwithA(x;y)=(ax;aby),and,therefore,fandgareaneequivalent.Corollary1.Foreveryo-polynomialFde nedonF2mthefunctionxF(yx)fromF2mF2mtoF2misbent.Thatis,everyo-polynomialonF2mde nesavectorialNihobentfunctionxF(yx)fromF2mF2mtoF2m.Proof.FromLemma1wehavethatforagiveno-polynomialFandanya2F2mthefunctiong(x;y)=Trm(axF(byx))isNihobentwhereb=F�1(a�1).Thenthefunctiong(x;y)=Trm(axF(yx))isalsobentsincegandgareaneequiva-lent,thatis,g=gAwithA(x;y)=(x;by),andclearly,suchatransformationAkeepsgasaNihofunction.9Lemma2.LetFbeano-polynomialonF2mandA(x)=x2jbeanautomor-phismoverF2m.ThentheNihobentfunctionsde nedbyo-polynomialsFandG=AFA�1areaneequivalent.Proof.ObviouslyifFisano-polynomial,thenG(x)=(F(x2�j))2jisalsoano-polynomial.ConsidertheNihobentfunctionde nedbyG:g(x;y)=TrmxGyx=TrmxAFA�1yx=TrmxFyx2�j2j=Trmx2�jFyx2�j=TrmuFvu;whereu=x2�jandv=y2�j.Thus,fandgareaneequivalent(g=fAwithA(x;y)=(x;y)2�j).Lemma3.LetFbeano-polynomialonF2mandA1(x)=x+aandA2(x)=x+bfora;b2F2m.ThenG=A1FA2isano-polynomialonF2mifandonlyifb=F(a)andF(a+1)+F(a)=1.Furthermore,theNihobentfunctionsde nedbyo-polynomialsFandGareEA-equivalent.Proof.SupposeG(x)=A1FA2(x)=F(x+a)+bisano-polynomial.Then0=G(0)=F(a)+band,therefore,F(a)=band1=G(1)=F(1+a)+b=F(1+a)+F(a).Furtherwehaveg(x;y)=TrmxA1FA2yx=TrmxFyx+a+b=TrmxFy+axx&

#17;+Trm(bx)=TrmxFux
#17;+Trm(bx)=TrmxFux+Trm(bx),whereu=y+ax.Thus,gandfareEA-equivalent(g=fA+lwithA(x;y)=(x;y+ax)andl(x;y)=Trm(bx)).5Themodi edMagicactionLetFbethecollectionofallfunctionsF:F2m7!F2msuchthatF(0)=0.ThefollowingsetP�L(2;2m)=fx7!Ax2jjA2GL(2;F2m);0jm�1gisagroupoftransformationsactingontheprojectivelines,i.e.onthesetwiththeelementsoftheform:f(ax;ay)j(x;y)6=(0;0);x;y2F2m;a6=0g.AnactionofthegroupP�L(2;2m)onFwasintroducedanddescribedin[21].De netheimageofF2Funderthetransformation 2P�L(2;2m), :x7!Ax2j,A=abcd2GL(2;2m),0jm�1,asafunction F:F2m7!F2msuchthat F(x)=jAj�12h(bx+d)F2jax+cbx+d+bxF2jab+dF2jcdi:ThisyieldsanactionofP�L(2;2m)onF,whichiscalledthemagicaction.Themagicactiontakeso-permutationstoo-permutationsanditisasemi-linear10transformation,i.e. (F+G)= F+ G;foranyF;G2F, aF=a2j F,foranya2F2m,F2F,0jm�1.Letusrecalltwotheorems(Theorem4andTheorem6)from[21].Foragiveno-polynomialFdenoteO(F)theovalde nedbyF.Theorem2.[21]LetFbeano-permutationonF2mandlet 2P�L(2;2m)be :x7!Ax2jforA=abcd2GL(2;F2m)and0jm�1.ThenG= Fisalsoano-permutationonF2m.Infact,O(G)= (O(F)),where 2P�L(3;2m)isde nedby :x7!Ax2j,whereA=0@d0cb F(db)jAj12a F(ca)b0a1A.Notethattheformulationofthetheoremabovedi ersfromtheonein[21]becauseinthecurrentpaper(followingnotationsof[7])thepointsoftheoval(orthehyperoval)de nedbyano-polynomialFareconsideredas(x;F(x);1),meanwhilein[21]theform(1;x;F(x))isused.Theorem3.[21]LetFandGbeo-permutationsonF2m,andsupposefurtherthattheovalsde nedbyFandG,i.e.O(F)andO(G)areequivalentunderP�L(3;2m).Thenthereexists 2P�L(2;2m)suchthatG= F.ThemagicactioncanbealsodescribedbyacollectionofgeneratorsofP�L(2;2m)[21]:a:x7!a001x;aF(x)=a�12F(ax);a2F2m;c:x7!10c1x;cF(x)=F(x+c)+F(c);c2F2m;':x7!0110x;'F(x)=xF(x�1);2j:x7!x2j;2jF(x)=(F(x2�j))2j;0jm�1:(6)Weslightlymodifythemagicactiongeneratorsaandcmultiplyingthembyappropriateconstantstopreservetheimageof1at1:~aF(x)=a1

2F(a)aF(x)=1F(a)F(ax);a2F2m;
2F(a)aF(x)=1F(a)F(ax);a2F2m;~cF(x)=1F(1+c)+F(c)cF(x)=1F(1+c)+F(c)(F(x+c)+F(c));c2F2m:(7)ThenewsetofgeneratorsH=f~a;~c;';2jj0jm�1;c2F2m;a2F2mpreservesthepropertyF(1)=1ofthefunctionF.TheactionofthegroupwiththenewsetofgeneratorsHonthesetofallfunctionsFde nedonF2mwiththepropertiesF(0)=0andF(1)=1willbecalledthemodi edmagicaction.11Proposition1.Twoo-polynomialsarisefromequivalenthyperovalsifandonlyiftheylieonthesameorbitofthegroupgeneratedbyHandtheinversemap.Proof.Accordingtothe rstpartofTheorem2,themagicactiontakeso-permutationstoo-permutations.Sincethegeneratorsofthemodi edmagicactiondi erfromtheoriginalmagicactiongeneratorsonlybyconstantcoe-cient(whatallowsastopreservethepropertyofF(1)=1foranyo-polynomialF),thenthemodi edmagicactiontakeso-polynomialstoo-polynomials.AccordingtothesecondpartofTheorem2,iftwoo-permutationslieonthesameorbitunderthemagicaction,thenthecorrespondingovalsareequivalentandhave xednucleus(0;1;0).Nowsupposethattwoo-polynomialslieonthesameorbitunderthemodi edmagicactionandtheinversemap.Sinceeacho-polynomialisano-permutation,thenthecorrespondingovalsde nedbyo-polynomialsareequivalentandhavenucleus(0;1;0).Asweknow,eachovaliscontainedinauniquehyperoval,whichisobtainedbyaddingnucleustothepointsofoval.So,hyperovalsde nedbytheo-polynomialsonthesameorbitunderthemodi edmagicactionareequivalent.Alsoitiswellknownthato-polynomialsFandF�1de neequivalenthyperovals.Thus,weconcludethathyperovalsde nedbytheo-polynomialsonthesameorbitunderthemodi edmagicactionandtheinversemapareequivalent.Let'sshowtheconversestatement.SupposethathyperovalsH(F)andH(G)de nedbyo-polynomialsFandGareequivalent.ItmeansthatthereisacollineationwhichmapsH(F)toH(G).Considerthepreimageof(0;1;0)underthiscollineation,thereare3possiblecases:1.Thepreimageof(0;1;0)is(0;1;0).Itmeansthatthiscollineation xespoint(0;1;0).SodeletingthispointfromhyperovalsH(F)andH(G),wewillgetequivalentovalswith xednucleus,hencebyTheorem3,theirgeneratoro-polynomialsareonthesameorbitunderthemagicaction,henceunderthemodi edmagicaction.2.Thepreimageof(0;1;0)i

s(1;0;0).Sincehyperovalsde nedbyo-po
s(1;0;0).Sincehyperovalsde nedbyo-polynomialanditsinverseo-polynomialareequivalent,thenhyperovalH(F)isequivalenttoahyperovalH(F�1)andbythecorrespondingcollineationthepoint(1;0;0)haspreimage(0;1;0).So,attheendwehavethathyperovalsH(F�1)andH(G)areequivalentandthepreimageof(0;1;0)is(0;1;0).Hencebythepreviouscase1(andthefactthatano-polynomialanditsinversebelongtothesameorbitundermodi edactionandtheinverse)o-polynomialsFandGareonthesameorbitundermodi edmagicactionandtheinversemap.Thefollowingdiagramillustratesthepreviousdecisions.H(F�1)=H(F)=H(G)222(0;1;0)7!(1;0;0)7!(0;1;0)3.Thepreimageof(0;1;0)is(t;f(t);1).Chooseanelement'ofP�L(2;2m)taking(1;t)to(0;1)(suchauthomorphismalwaysexist,forexampleitcanbede nedbymatrixA=0010).Applying'toFwewillgetahyperovalH('F)equivalenttoH(G)wherethepreimageof(0;1;0)is(1;0;0).Becauseofthecase2,wegetthat'FandGbelongtothesameorbitunderthemodi edmagicactionandtheinversemapandsodoFandG.12Weformulatethenexttheoremwithoutproof.Firstthisresultwasan-nouncedinSeptember2014attheForthIsreeConference"FiniteGeometries"[8]bytheauthorsofthispaper,thecompleteproofcanbefoundin[1].Theorem4.TwoNihobentfunctionsareEA-equivalentifandonlyifthecorrespondingovalsareequivalent.Hence,thenumberofEA-equivalenceclassesofNihobentfunctionsarisingfromahyperovalofPG(2;2m)isthenumberoforbitsofthecollineationstabiliserofthehyperovalonthepointsofthehyperoval.6Nihobentfunctionsandthemodi edmagicactionAgroupoftransformationsoforder24with3generatorspreservingo-polynomialswasconsideredin[7].Thisgroupoftransformationsisasubgroupofthegroupwiththe(modi ed)magicactiongeneratorsandtheinversemap.Precisely,theyarethetransformationsgeneratedby',~1=1andtheinversemap.Only4ofthesetransformationscanleadtoEA-inequivalentNihobentfunctions[7].Asacontinuationoftheworkof[7],let'sconsiderthemodi edmagicactiongenerators,andtheinversemapandseewhichofthemgiverisetoEA-inequivalentNihobentfunctions.FromProposition1itisclearthato-polynomialsonthesameorbitunderthemodi edmagicactionandthein-versemapandonlytheyareprojectivelyequivalent.SinceweareinterestedinEA-inequivalentNihobentfunctionsarising

fromprojectivelyequivalento-polynomials,
fromprojectivelyequivalento-polynomials,wefocusonorbitsofthemodi edmagicactiontogetherwiththeinversemap.WeprovebelowthattogetEA-inequivalentNihobentfunctionsfromagiveno-polynomialitissucienttouseonly~and'generatorstogetherwithinversemapwhileand~donotplayanyroleinit.Moreover,weshowthatallEA-inequivalentNihobentfunctionscanbeobtainedfromaspecialformula.6.1PreliminaryresultsFollowingnotationsof[7]thegenerator'willbedenotedby0whenneeded.Let'srecallthesetofgeneratorsH=f~c;~a;0;2jjc2F2m;a2F2m;0jm�1g;where~aF(x)=1F(a)F(ax);a2F2m;~cF(x)= cFcF(x)= cF(F(x+c)+F(c));c2F2m;where cF=1cF(1);F0(x)='F(x)=xF(x�1);2jF(x)=(F(x2�j))2j;0jm�1;andproveafewstatementsaboutthegeneratorsofmagicactionandtheinversemap.13Lemma4.LetFbeano-polynomialonF2m.Thenthefollowingidentitieshold:~c~dF=~c+dF;(8)~a~bF=~abF;(9)2j2iF=2j+iF;(10)wherea;b2F2m,c;d2F2m;0i;jm�1.Proof.Toprovethe rstequalitynotethatcdF(x)=dF(x+c)+dF(c)=F(x+c+d)+F(d)+F(c+d)+F(d)=F(x+c+d)+F(c+d)=c+dF:Sincemagicactionisasemilineartransformationweget:~c~dF(x)=1F(1+d)+F(d)1~dF(1+c)+~d(c)c(d(F(x))=1F(1+d)+F(d)F(1+d)+F(d)F(1+d+c)+F(d+c)c+dF(x)=1F(1+d+c)+F(d+c)c+dF(x)=~c+dF(x):Theothertwoequalitiesarestraightforwardtoprove:~a~bF=1~bF(a)~bF(ax)=11F(b)F(ab)1F(b)F(abx)=1F(ab)F(abx)=~abF(x);2i2jF(x)=2i(F(x12j))2j=F(x12j+i)2j+i=2i+jF(x):Corollary2.LetFbeano-polynomialonF2mandkapositiveinteger.Then(~a1~a2:::~ak)F=~a1a2:::akF;(~c1~c2:::~ck)F=~c1+c2+:::ckF;(2i12i2:::2ik)F=2i1+i2+:::+ikF;wherea1;:::;ak2F2m;c1;:::;ck2F2m,0ijm�1forallj2f1;:::;kg.Proof.TheprooffollowsbyinductionusingLemma4.Lemma5.LetFbeano-polynomialonFm2.Thenthefollowingidentitieshold:(~cF)�1(x)=~F(c)F�11 cFx;(11)(~aF)�1(x)=~F(a)F�1(x);(1

2)(2jF)�1(x)=2jF�1(x);(1
2)(2jF)�1(x)=2jF�1(x);(13)wherea2F2m,c2F2mand0jm�1.14Proof.Itiseasytoseethat~F(c)F�11 cF=1,therefore(~cF)�1(x)=( cF(F(x+c)+F(c)))�1=F�11 cFx+F(c)+c=F�11 cFx+F(c)+F�1(F(c))=~F(c)F�11 cFx:Equalities(12)and(13)arestraightforwaredtoprove:(~aF)�1(x)=1F(a)F(ax)�1=1aF�1(F(a)x)=~F(a)F�1(x);(2jF)�1(x)=((F(x2�j))2j)�1=(F(x2�j)�1)2j=2jF�1(x):Lemma6.LetFbeano-polynomialonF2m.Thenthefollowingidentitieshold:~c2jF=2j~c2�jF;(14)~c~aF=~a~acF;(15)(2jF)0=2jF0(16)(~aF)0=~1aF0;(17)wherea2F2m,c2F2m;0jm�1.Proof.Toprovethe rstequality,transformitsleftandrightsides.~c2jF(x)= c2jF(2jF(x+c)+2jF(c))= c2jF((F((x+c)2�j))2j+(F(c2�j))2j)= c2jF((F(x2�j+c2�j))2j+(F(c2�j))2j)= c2jF(F(x2�j+c2�j)+F(c2�j))2jOntheotherhand,2j~c2�jF(x)=(~c2�jF(x2�j))2j=( c2�jF(F(x2�j+c2�j)+F(c2�j))2j:So,itislefttocheckthat( c2�jF)2j= c2jF.Indeed, c2jF=12jF(1+c)+2jF(c)=1(F((1+c)2�j))2j+(F(c2�j))2j=1F(1+c2�j)+F(c2�j)2j=( c2�jF)2j:Thusweprovedthat~c2jF=2j~c2�jF.Computingtheleftandtherightsidesofequality(15)weget~c~aF(x)= c~aF(~aF(x+c)+~aF(c))= c~aF(1F(a)F(a(x+c))+1F(a)F(ac));~a~acF(x)=1~acF(a) acF(F(ax+ac)+F(ac)):15Notethatthecoecients1F(a) c~aFand1~acF(a) acFareequalwhichmeansthat~c~aF=~a~acF:Indeed,1F(a) c~aF=1F(a)1~aF(1+c)+~aF(c)=1F(a)F(a)F(a(1+c))+F(ac)=1F(a+ac)+F(ac);1~acF(a) acF=F(1+ac)+F(ac)F(a+ac)+F(ac)1F(1+ac)+F(ac)=1F(a+ac)+F(ac):Theremainingtwoequalitiesareprovedsimilarly.For(16)weget2jF0(x)=(F0(x2�j))2j=(x2�jF(1x2�j))2j=x(F(1x2�j))2j=x2jF(1x)=(2jF)0(x):Transformingbothsi

desofEquality(17)weget(~aF)0(x)=x~&
desofEquality(17)weget(~aF)0(x)=x~aF1x=xF(a)Fax:~1aF0(x)=1F0(1a)F0xa=aF(a)xaFax=xF(a)Fax:6.2EA-inequivalentNihobentfunctionsandorbitsFurtherweneedthefollowingequalityfrom[7]((F0)�1)0=((F�1)0)�1(18)Let'sintroduceafewnotations.DenotebygFtheNihobentfunctionde nedbyano-polynomialF.WhenNihobentfunctionsgFandgFareEA-equivalent(EA-inequivalent),wewillwritegFEAgF(respectively,gFEAgF).Wewillusenotation"A(p)=B",whentheexpressionBisobtainedfromtheexpressionAusingequalitynumberp.Theorem5.LetFbeano-polynomial.Thenano-polynomialFobtainedfromFusingonegeneratorofthemodi edmagicactionandtheinversemapcanproduceaNihobentfunctionEA-inequivalenttothosede nedbyFandF�1onlyifF=(F0)�1.Proof.AssumeFisano-polynomialwhichisobtainedfromo-polynomialFusingonegeneratorofthemodi edmagicactionandtheinversemap,i.e.Fhasoneofthefollowingforms:hF;hF�1;(hF)�1;(hF�1)�1,whereh2H.Asweshowbelow,whenhis~a,~cor2j,Fde nesaNihobentfunctionEA-equivalenttothosede nedbyForF�1.a)Lethbe~a;a2F2m.ThenhF(x)=~aF(x)=1F(a)F(ax)andbyLemma1,thecorrespondingNihobentfunctionisEA-equivalenttothosede nedbyF.BythesamereasonhF�1=~aF�1andF�1de neEA-equivalentNihobentfunctions.Furthernotethat(hF)�1(x)=(~aF)�1(x)(12)=~F(a)F�1(x):16Hence,g(~aF)�1EAgF�1and(hF�1)�1(x)=(~aF�1)�1(x)(12)=~F�1(a)(F�1)�1(x)=~F�1(a)F(x);andthereforeg(~aF�1)�1EAgF.b)Supposehis~cwithc2F2m.ThenhF(x)=~cF(x)= cF(F(x+c)+F(c))andhF�1(x)=~cF�1de neNihobentfunctionsEA-equivalenttothosede nedbyFandF�1respectively(byLemma3).Hence,(hF)�1(x)=(~cF(x))�1(x)(11)=F(c)F�1(( cF)�1x)yieldsthatg(hF)�1EAgFandfrom(hF�1)�1(x)=(~cF�1)�1(x)(11)=F�1(c)(F�1)�11 cF�1x=F�1(c)F1 cF�1xfollowsg(hF�1)�1EAgF.c)Takenowh=2jwith0jm�1.ThenhF(x)=2jF(x)=(F(x2�i))2iandhF&#

0;1(x)=2jF�1=(F�1(x2�i))2i
0;1(x)=2jF�1=(F�1(x2�i))2i,andbyLemma2wegetthatg2jFandg2jF�1areEA-equivalenttogFandgF�1,respectively.Therefore,from(hF)�1(x)=(2jF)�1(x)(13)=2jF�1and(hF�1)�1(x)=(2jF�1)�1(13)=2jFitfollowsthatg(2jF)�1EAgF�1andg(2jF�1)�1EAgF.d)Considerh=0.TheNihobentfunctionde nedbyano-polynomialhF(x)=F0(x)=xF(x�1)isgF0(x;y)=Trm(x(F0(yx)))=Trm(xyxF((yx)�1))=Trm(yF(xy))=gF(y;x);i.e.gF0EAgF.Similarly,g(F�1)0EAgF�1.Thefunction(hF)�1(x)=(F0)�1(x)=(xF(x�1))�1cande neaNihobentfunctionEA-inequivalenttothosede nedbyFandF�1.Forexample,ano-monomialx2ide nesthreesurelyEA-inequivalentNihobentfunctionscorrespondingtoo-polynomialsF,F�1and(F0)�1[7].Usingequality(18),weimmediatelygetthataNihobentfunctionde nedbytheo-polynomial(hF�1)�1(x)=((F�1)0)�1(x)isEA-equivalenttoonede nedby(F0)�1.WerewritetheequalitiesofLemmas4,5and6inamorecompactway.Equalities(8)-(10)ashb1hb2F=hb3F;(19)wherehb1;hb2;hb3arethesamegeneratorsfromthesetHnf0gwithdi erentparametersb1;b2;b32F2m.Equalities(11)-(13)as(hb1F)�1=hb2F�1;(20)wherehb1;hb2arethesamegeneratorsfromthesetHnf0gwithdi erentpa-rametersb1;b22F2m.Notethatrightandleftpartsoftheequality(11)have17di erentarguments,butitdoesnotplayanyroleinourstudyofEA-equivalenceofresultingNihobentfunctions.Equalities(14)-(15)as~c1hbF=hb~c2F;(21)wherehb2f~a;2jg:Andequalities(16)-(17)as(hb1F)0=hb2F0;(22)wherehb1;hb2arethesamegeneratorsfromthesetf~a;2jgwithdi erentpa-rametersb1;b22F2m.Tomaketheformulationofthenexttheoremmorevisualinsteadofusingthenotation0wewillusetheinitialone,i.e.'.Wewillalsorefertotheoriginalnotation'insomepartsoftheproofwhenconvenient.Further,by"reduceo-polynomial"wemeanthattheoriginalo-polynomialandthenewone(reduced)de neEA-equivalentNihobentfunctions.Whenwearesaying"deletegenerator"wemeanthatifweskipthisgeneratorthenewo-polynomialwillde neaNihobentfunctionEA-equivalenttoonegeneratedbytheoriginalo-polynomial.Letibeapositiveintegerandki0.ByHiwedenoteacompositionoflengthkiofg

enerators'and~cfollowingeachotheras
enerators'and~cfollowingeachotherasfollows:Hi='~ci1'~ci2:::|{z}ki(23)Thatis,ifFisano-polynomialandwedenoteTj='~cij,0j(ki+1)=2thenHiF=8&#x]TJ/;ø 9;&#x.962; Tf;&#x 17.;ॗ ;� Td;&#x [00;&#x]TJ/;ø 9;&#x.962; Tf;&#x 17.;ॗ ;� Td;&#x [00;&#x]TJ/;ø 9;&#x.962; Tf;&#x 17.;ॗ ;� Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;:Fifki=0;'Fifki=1;T1:::TsiFifki=2si;T1:::Tsi'Fifki=2si+1:Inthetheorembelowweprovethatforagiveno-polynomialwecanderiveallEA-inequivalentNihobentfunctionsonlyusingtransformations',~candtheinversemapinaspecialsequence.Theorem6.LetFbeano-polynomial,gFthecorrespondingNihobentfunctionandGFtheclassofallfunctionso-equivalenttogF.Theno-polynomialsoftheform(H1(H2(H3(:::(HqF)�1:::)�1)�1)�1;(24)whereHiisde nedby(23),foralli2f1:::qg,q1,andki1fori3,ki0fori2,providerepresentativesforallEA-equivalenceclasseswithinGF.Thatis,uptoEA-equivalence,allNihobentfunctionso-equivalenttogFarisefrom(24).Proof.Note rstthatwecangetFitselfintheform(24)ifwetakeq=2,k1=k2=0.ifq=1andk1=0thenwegetF�1.Furtherwehavearestrictionki1fori3toavoidrepetitions.18AccordingtoProposition1anyfunctiono-equivalenttogFcorrespondstoano-polynomialoftheformh1h2:::hkF;(25)whereh1;h2;:::;hk(forsomek0)aregeneratorsofthemodi edmagicactionandtheinversemap.OuraimistosimplifythisexpressiontoexcludeasmanycasesleadingtoEA-equivalentfunctionsaspossible.Thatis,weexcludecertainsequencesofgeneratorswhichsurelyleadtoEA-equivalentNihobentfunctions.Byhijwedenoteageneratorofthesametypeashibutwithadi erentparameter.FromTheorem5itfollowsa)Ifh12H,thengh1h2:::hkFEAgh2:::hkFandwecanconsiderreducedo-polynomialh2:::hkF;b)Ifh1istheinversemapandh22Hnf0gthengh1h2:::hkFEAgh1h3:::hkF,sowecanconsiderthereducedo-polynomialh1h3:::hkF.Hence,ifk=1in(25)thenwecangetanEA-inequivalentcaseonlyifh1istheinversemap,anditcorrespondsto(24)wi

thq=1andk1=0.Ifk=2in(25)(anditcannotbere
thq=1andk1=0.Ifk=2in(25)(anditcannotbereducedtothecasek=1)thenwecangetEA-inequivalentcasesonlyifh1istheinversemapandh2=0,anditcorrespondsto(24)withq=1andk1=1.Ifk3wecanreduce(25)untilatsomemomentwewillgetano-polynomialhihi+1:::hkF,wherehiistheinversemapandhi+1=0,thatis,wehave((hi+2:::hkF)0)�1:(26)Notethathereandfurtherweassumethatkislargeenoughtoallowsucharedactionwhileotherwise,itiseasytoseethattheprocesswouldstopandprovideaformula(24)forsomeparameters.Ifhi+22f~a;2jgorhi+2istheinversemapthenwecandeletethegeneratorhi+2andconsiderthereducedo-polynomialhihi+1hi+3:::hkF.Indeed,supposehi+22f~a;2jgthenhihi+1hi+2:::hkF=((hi+2:::hkF)0)�1(22)=(h(i+2)1(hi+3:::hkF)0)�1(20)=h(i+2)2((hi+3:::hkF)0)�1and,accordingto(a),ghihi+1hi+2:::hkFEAghihi+1hi+3:::hkF.Inthecasewhenhi+2istheinversemap,using(18)wegetthesameresultthattheo-polynomials(((hi+3:::hkF)�1)0)�1=((hi+3:::hkF)0)�1)0and((hi+3:::hkF)0)�1=hihi+1hi+3:::hkFde neEA-equivalentNihobentfunctions.Ifhi+2is0,thenhi+1andhi+2eliminateeachother:hihi+1hi+2:::hkF=hihi+3:::hkF.Ifhi+2=~c,thenwecannoteliminateitfromtheo-polynomialhihi+1hi+2:::hkF.Furtherconsiderano-polynomialhihi+1hi+2:::hkFwherehiistheinversemap,hi+1=0,hi+2=~c,i.e.ano-polynomial((~chi+3:::hkF)0)�1:(27)Whenk=i+2thenweget((~cF)0)�1whichhastheform(24)withq=1andk1=2.Hence,in(27)wecanassumethatki+3.Furtherwecanreducehi+3from(27)unlesshi+3is0.Indeed,consider rsthi+32f~a;2jgthen19((~chi+3:::hkF)0)�1(21)=((hi+3~c1hi+4:::hkF)0)�1(22)=(h(i+3)1(~c1:::hkF)0)�1(20)=h(i+3)2((~c1hi+4:::hkF)0)�1.Thelasto-polynomialde nesaNihobentfunctionEA-equivalenttoonede nedbytheo-polynomial((~c1hi+4:::hkF)0)�1=hihi+1h(i+2)1hi+4:::hkF.Ifhi+3

=~c1,thenusing(8)weimmediatelygethi
=~c1,thenusing(8)weimmediatelygethihi+1hi+2hi+3:::hkF=hihi+1h(i+2)1hi+4:::hkF,whereh(i+2)1=~c+c1.Ifhi+3istheinversemapthenhihi+1hi+2hi+3:::hkF=((~c((hi+4:::hkF)�1))0)�1(20)=(((~c1hi+4:::hkF)�1)0)�1=(((~c1hi+4:::hkF)0)�1)0;de nesaNihobentfunctionEA-equivalenttotheonede nedby((~c1hi+4:::hkF)0)�1=hihi+1h(i+2)1hi+4:::hkF.Notethatwecouldeliminatehi+3astheinverseherebecauseitisfollowedbyhi+2=~c,hi+1=0andhiastheinversemap.Hence,if(25)producesaNihobentfunctiongEA-inequivalenttothosecorrespondingtoF,F�1,(F0)�1and((~cF)0)�1thengisEA-equivalenttothefunctioncorrespondingtoano-polynomial('~c0'hl0:::hkF)�1:(28)Nowconsiderano-polynomialoftheform:('~c1'~c2':::hl:::hkF)�1:(29)Case1.Firstwerestricttothecasehl;:::;hk2Hwhenconsidering(29).Notethatiflisanevennumberin(29),thenthegenerator'actsonhl;iflisodd,thenthegenerator~cactsonhl(forsomec2F2m).Weconsiderloddcase,i.e.l=2t+1whileforlevencasetheproofissimilarandweskipit.Ifh2t+12f~a;2jgthen('~c1'~c2':::~cth2t+1:::hkF)�1(21)=('~c1':::'h2t+1~ct1(h2t+2:::hkF))�1(22)=('~c1':::h(2t+1)1'(~ct1(h2t+2:::hkF)))�1(21)=:::(h(2t+1)t('(~c11('(:::(~ct1(h2t+2:::hkF)):::))))�1(20)=h(2t+1)t+1('(~c11('(:::(~ct1(h2t+2:::hkF)):::))))�1,hencewecanreducetheo-polynomial('~c1'~c2':::~cth2t+1:::hkF)�1;andconsider('~c11'~c21':::~ct1h2t+2:::hkF)�1.Ifh2t+1=~ct+1thenobviouslywecanconsidero-polynomial((~c1(~c2(:::(~ct+ct+1(h2t+2:::hkF))0:::)0)0)0)�1.Ifh2t+1=0thenwecannoteliminateit.Continuingthisprocesswegetforthiscasethattheo-polynomial(25)canbereduce

dto('~c1'~c2
dto('~c1'~c2':::F)�1asin(23).Thiscorrespondstothecaseq=1in(24).Case2.Nowweconsider(29)andallowhl;:::;hktobeinversestoo.Westillassumelbeoddand(aswesawearlierintheproof)w.l.o.g.hl;:::;hk2f0;~c;theinversejc2F2mg.Takehltheinverse(theotherpossibilitiesforhlwerediscussedearlierintheproof),i.e.considerthefollowingo-polynomial:('~c1'~c2':::(hl+1:::hkF)�1)�1:(30)20Ifhl+1istheinverse,thenitcancelswithhl.Ifhl+1is~ct+1,then('~c1'~c2':::(~ct+1hl+2:::hkF)�1)�1(20)=('~c1'~c2':::~c(t+1)1(hl+2:::hkF)�1)�1,whichisoftheform(30)withfewertransformationsintheinnerbrackets.Ifhl+1is'thenweget('~c1'~c2':::('hl+2:::hkF)�1)�1.Iffurtherhl+2is~ct+1,then('~c1'~c2':::('~ct+1hl+3:::hkF)�1)�1.Ifhl+2istheinverseorhl+2='thenweget(30).Indeed,ifhl+2='thenitcancelswithhl+1,andifhl+2istheinversethenweget:('~c1'~c2':::('(hl+3:::hkF)�1)�1)�1(18)=('~c1'~c2':::'('hl+3:::hkF)�1)�1.Continuingtheseprocesswewillclearlytransform(30)to(24)inawaythattheseo-polynomialsproduceEA-equivalentNihobentfunctions.Inthispaper,whenwesaythattwoo-polynomialsFandF0de nepoten-tiallyEA-inequivalentNihobentfunctionsgFandgF0,itmeansthateitherinsomecasesgFandgF0areEA-inequivalent,oritisnotpossibletodeduceEA-equivalencewiththedevelopedtechnicswhichleavesapossibilitythatgFandgF0maybeEA-inequivalent.Belowweconsidersomeparticularcasesofformula(24).Corollary3.LetFbeano-polynomialde nedonF2m.Theno-polynomialsFc(x)= cFxF1x+c+F(c)�1;c2F2m(31)de neasequenceofNihobentfunctionsgFcpotentiallyEA-inequivalenttoeachotherfordi erentc,andEA-inequivalenttoNihobentfunctionsde nedbyF,F�1.Proof.o-polynomial(31)istheexplicitformofo-polynomial(24)forq=1;k1=2.Indeed,((~cF

)0)�1(x)=x~cF1x&
)0)�1(x)=x~cF1x�1= cFxF1x+c+F(c)�1.NotethatFc=(F0)�1forc=0.Hence,theo-polynomial(F0)�1isincludedintheclassofo-polynomialsFc.Forc=1wegetthefunctionF=xF1x+1+1�1studiedin[7]andwhichcande neaNihobentfunctionEA-inequivalenttothosede nedbyF,F�1and(F0)�1.Forinstance,whenF(x)=x2i,gFisEA-inequivalenttogF,gF�1andg(F0)�1[7].Usingtheequality(8)foreveryc2F2mwecanwrite:Fc=((~cF)0)�1=((~1~c+1F)0)�1=(~c+1F):SinceF,F,F�1and(F0)�1cande nefourpotentiallyEA-inequivalentNihobentfunctions,weobtainthatFccande neNihobentfunctionspotentiallyEA-inequivalenttothosede nedby~c+1F,(~c+1F)�1,((~c+1F)0)�1.Itmeansthat,foranyc2F2maNihobentfunctiongFccanbepotentiallyEA-inequivalenttogF,gF�1andgFc+1.21Corollary4.LetFbeano-polynomialde nedonF2m.Theno-polynomials(Fc)�1= cF0(1+cx)Fx1+cx+cxF1c�1;c2F2m(32)de neNihobentfunctionsg(Fc)�1whichcanpotentiallybeEA-inequivalenttoeachotherfordi erentcandEA-inequivalenttoNihobentfunctionsde nedbyF,(F0)�1.Proof.o-polynomial(32)istheexplicitformofo-polynomial(24)forq=1andk1=3.Indeed,((~cF0)0)�1(x)= cF0xF01x+c+F0(c)�1= cF0x1+cxxFx1+cx+cF1c�1= cF0(1+cx)Fx1+cx+cxF1c�1:Notethat(F0)�1=F�1.Sotheo-polynomialF�1isincludedintheclassofo-polynomials(Fc)�1withc=0.Forc=1wegetthefunction(F1)�1=((x+1)F(xx+1)+x)�1alsostudiedin[7],andtheNihobentfunctionassociatedwithitisEA-equivalenttotheonede nedbyF[7].Butinthegeneralcase,forarbitraryc2F2mwecan'tsaythat(Fc)�1de nesano-polynomialEA-equivalenttothosede nedbyFandFc.Usingequalities(8)and(31)notethat(Fc)�1=(F0)c=(~c+1F0).Hence,wecansaythat(Fc)�1=(F0)cde nesaNihobentfunctionpotentiallyEA-inequivalenttoNihobentfunctionsde nedbyF0,(F

0)�1and(F0)c+1=(Fc+1)�1.
0)�1and(F0)c+1=(Fc+1)�1.6.3Thecaseofo-monomialsandtheknowno-polynomialsFurtherwestudytheconsequencesoftheobtainedresultsfortheparticularcasesofo-monomialsandtheknowno-polynomials.Lemma7.Forano-monomialF(x)=xd,theNihobentfunctionsde nedbyFcandFareEA-equivalent,foranyc2F2m.Proof.Wehaveforc6=0Fc(x)=('~cF)�1= cFxF1x+c+F(c)�1= cFx1x+cd+cd�1= cFx1+cxxd+cd�1= cFcdx1+cxcxd+1�1= cFcd�1cx1+cxcxd+1�1=1cF1 cFcd�1x:FromLemma1itfollowsthatNihobentfunctionsde nedbyFcandFareEA-equivalentforanyc6=0.Fromtheproofofthepreviouslemmaitiseasytoseethatforanyo-monomialF'~cF(x)= c'1F(cx);(33)where c= cFcd�1;c2F2m.22Lemma8.Forano-monomialF(x)=xd,theNihobentfunctionsde nedby(Fc)�1,(F)�1andFareEA-equivalent,forc2F2m.Proof.F(x)=(x+1)F(xx+1)+x=(x+1)(xx+1)d+x.Forc6=0wehave(Fc)�1(x)=('c'F)�1= cF0(1+cx)Fx1+cx+cxF1c�1= cF0(1+cx)x1+cxd+cx1cd�1= cF01cd(1+cx)cx1+cxd+cx�1=1c(F)�1cd cF0x:UsingLemma1,weconcludethattheNihobentfunctionsde nedby(F)�1and(Fc)�1areEA-equivalentforc6=0.Accordingto[7],theNihobentfunctionde nedby(F)�1andFareEA-equivalent,andtakingintoaccountLemma7,wegetthatNihobentfunctionsde nedby(Fc)�1,(F)�1andFareEA-equivalenttoeachotherforanyc6=0.Fromtheproofofabovelemmaitiseasytoseethatforanyo-monomialF'~c'F(x)= c'1'F(cx):(34)where c= cF0cd�1;c2F2m;F0='F.Furtherwewillneedthefollowingequality,whichholdsforanyo-polynomialF'1'F=1'1F:(35)Indeed,1'1F(x)=(1+x)F11+x+1+1+1=(1+x)Fx1+x+x='1'F(x):Tokeepnotationsassimpleaspossible,

sinceweareinterestedinEA-equivalenceofNi
sinceweareinterestedinEA-equivalenceofNihobentfunctionsandcoecientsofargumentsofo-polynomialdonota ectonEA-equivalenceofNihobentfunctionsaswellascoecientofo-polynomial,theninsteadofaF(bx)=G(x)wewillwriteFGfora;b2F2m.Lemma9.LetFbeano-monomialde nedonF2m.Then'~c1'~c2':::|{z}kF8���������������&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;:8&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;:1F;ift0mod4;'1F;ift1mod4;1'F;ift2mod4;'1'F;ift3mod4;ifk=2t8&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;:1'F;ift0mod4;'1'F;ift1mod4;1F;ift2mod4;'1F;ift3mod4;ifk=2t+1;wheret1.23Proof.Assumethatk=2t,i.e.theorbitinthestatmentofthislemmahastheform'~c1'~c2':::'~ctF.Then1)Fort=1wehave'~c1F(33)'~1F.2)Fort=2,'~c1'~c2F(33)'~c1'1F(??)'1'~c1F(33)'1'1F(35)''&

#14;~1'F1'F.3)F
#14;~1'F1'F.3)Fort=3,'~c1'~c2'~c3F2)'~c11'F'~c1+1'F(??)'1'F4)Fort=4'~c1'~c2'~c3'~c4F3)'~c1'1'F2)1'('F)1F.Thusforevenk,'~c1:::'~ct�3'~ct�2'~ct�1'~ctF4)'~c1:::'~ct�41F'~c1:::'~ct�4+1F4):::8�����&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;:1F;ift0mod4;'~c11F1)'1F;ift1mod4;'~c1'~c21F2)1'F;ift2mod4;'~c1'~c2'~c31F3)'1'F;ift3mod4;Notethat'Fisano-monomial,thereforewecanapplythepreviousformulatothecaseofoddk.Indeed,'~c1:::'~ct�3'~ct�2'~ct�1'~ct('F)8&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;:1'F;ift0mod4;'1('F);ift1mod4;1'('F)1F;ift2mod4;'1'('F)'1F;ift3mod4;Lemma10.LetFbeano-monomialde nedonF2m.Then'~c1'~c2':::|{z}k('1F)�18����������&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00

;&#x]TJ ;� -1;.93; Td;&#x [00
;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;:8&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;:('1F)�1;ift0mod3;('1('F)�1)�1;ift1mod3;('1'F�1)�1;ift2mod3;ifk=2t8&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;:('1F�1)�1;ift0mod3;('1('F�1)�1)�1;ift1mod3;('1'F)�1;ift2mod3;ifk=2t+1;(36)wheret1.24Proof.Assumethatk=2t,i.e.theorbitinthestatementofthislemmahastheform'~c1'~c2':::'~ct('1F)�1.Then1)Fort=1weget:'~c1('1F)�1(20)'(~c11'1F)�1(18)('('~c11'1F)�1)�1(??)('('1'~c11F)�1)�1(33)('('1'1F)�1)�1(35)('(1'F)�1)�1(20)('1('F)�1)�1:2)Fort=2'~c1'c2('1F)�11)'~c1('1('F)�1)�11)('1('('F)�1)�1)�1(18)('1'F�1)�1:3)Fort=3,'~c1'~c2'c3('1F)�12)'~c1('1'F�1)�11)('1F)�1:Thus,'~c1:::'~ct�2'~ct�1'~ct('1F)�13)'~c1:::'~ct�3('1F)�13):::8��&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;:('1F)�1;ift0mod3;'~c1('1F)�11)('1('F)�1)�1;ift1mod3;'~c1'~c2('1F)�12)('1'F�1)�1;ift2mod3:Notethatfrom(18)followsthat'('1F)�1=('(1F)�1)�1=('1F�1)�1.Thereforthecaseofoddkcomesdowntothepreviouscase.Indeed,'~c1:::'~ct�2'~&

#28;ct�1'~ct'('&#
#28;ct�1'~ct'('1F)�13)'~c1:::'~ct�2'~ct�1'~ct('1F�1)�18&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;:('1F�1)�1;ift0mod3;('1('F�1)�1)�1;ift1mod3;('1'F)�1;if2mod3:Lemma11.LetFbeano-monomial.Thenforq3(H1(H2(:::(HqF)�1:::)�1)�18�&#x]TJ ;� -1;.93; Td;&#x [00;:1G�1;('1G)�1;'1G;whereG2fF;('F)�1;'F�1;F�1;('F�1)�1;'FgandHiarede nedby(23)forall1iq.Proof.Firstconsiderthefollowingcases:1.q=1.ItiseasytoseethatfromLemma9follows(H1F)�18&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;:(1F)�11F�1;('1F)�1;(1'F)�11('F)�1;('1'F)�1;=(1G�1;('1G)�1;(37)25whereG2fF;'Fg2.q=2.ObviouslyfromLemma10wehave(H1('1F)�1)�1='1G;(38)whereG2fF;('F)�1;'F�1;F�1;('F�1)�1;'Fg:Using(37)and(38)weget(H1(H2F)�1)�1378��&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;:(H11G�11)�137(1G�12;('1G2)�1;(H1('1G1)�1)�138'1G2;(39)whereG12fF;'Fg;G22fG�11;'G�11g=A1;G22fG1;('G1)�1;'G�11;G�11;('G�11)�1;'G1g=A2:ItiseasytoseethatA1=fF�1;('F)�1;'F�1;('F�1)�1g;A2=fF;('F)�1;'F�1;F�1;('F�1)�1;'Fg:Indeed,ifwetakeG1=FinA2,thenwegetfF;('F)�1;'F�1;F�1;('F�1)�1;'Fg,ifwetakeG1='F,thenwegetthesamesetofo-polynomials,since('('F)�1)�1(18)=(('F�1)�1)�1='F�1:NotethatallfunctionsinthesetsA1andA2areo-monomials.3.q=3,(H1(H2(H3F)�1)�1)�1(39)8�������&#x]TJ ;� -1;.93; Td;&#x [00;

&#x]TJ ;� -1;.93; Td;&#x [00;
&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;:(H11G�12)�1(37)(1G�13;('1G3)�1;(H1('1G2)�1)�1(38)'1G3(H1'1G2)�1(37)(1~G�13;('1~G3)�1;whereG32fG�12;'G�12g,G32fG2;'G�12;('G2)�1;G�12;('G�12)�1;'G2g,~G32fG2;'G2g,G22A1,G22A2.Substitutinginthecorrespondingsetso-monomialsfromA1andA2,using(18),wegetthatG3;G3;~G3belongtoA2,therefore(H1(H2(H3F)�1)�1)�18�&#x]TJ ;� -1;.93; Td;&#x [00;:1G�13;'1G3;('1G3)�1;whereG32A2=fF;('F)�1;'F�1;F�1;('F�1)�1;'Fg.Wearegoingtoprovethislemmabyinductiononthelengthoforbitq.Forq=3thestatementofthelemmaistrueaswesawabove.Supposethatitis26trueforanylq�1andl3.Byourassumption:(H1(H2(:::(HqF)�1:::)�1)�18�������&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;&#x]TJ ;� -1;.93; Td;&#x [00;:(H11G�1)�1(37)(1G�11;('1G1)�1;(H1('1G)�1)�1(38)'1G1;(H1'1G)�1(37)(1~G�11;('1~G1)�1;whereG2A2,G12fG�1;'G�1g,G12fG;('G)�1;'G�1;G�1;('G�1)�1;'Gg,~G12fG;'Gg.BystraightforwardcomputationsitiseasytoseethatallofthesetsareequaltoA2,thus(H1(H2(:::(HqF)�1:::)�1)�18�&#x]TJ ;� -1;.93; Td;&#x [00;:1G�1;('1G)�1;'1G;whereG2fF;('F)�1;'F�1;F�1;('F�1)�1;'Fg,whichprovesourstate-ment.Proposition2.Themodi edmagicactionandtheinversemapappliedtoo-monomialsgiveatmost4EA-inequivalentfunctions.Forano-monomialFthe4potentiallyEA-inequivalentbentfunctionsarede nedb

yF;F�1;(F0)�1andF.Proof.Weuse
yF;F�1;(F0)�1andF.Proof.WeuseLemma11anddiscussthecasesq=1;2andq3separately.1.q=1.Accordingto(37)(H1F)�1hasthefollowingtwoforms1G�1and('1G)�1,whereG2fF;'Fg.The rstfunctionobviouslyde nesNihobentfunctionsEA-equivalenttoonede nedbyG�1andthereforetothosede- nedbyF�1and('F)�1.Thesecondfunctionde nesNihobentfunctionsEA-equivalenttoonede nedbyF(byLemma8).2.q=2.From(39)wehave:(H1(H2F)�1)�18�&#x]TJ ;� -1;.93; Td;&#x [00;:1G�12;('1G2)�1;'1G2;whereG22fF�1;('F)�1;'F�1;('F�1)�1g;G22fF;('F)�1;'F�1;F�1;('F�1)�1;'Fg:Obviously,1G�12and'1G2de neNihobentfunctionEA-equivalenttothosede nedbyG�12andG2respectively,whichintheirturnde neNihobentfunctionsEA-equivalenttoF;F�1and(F0)�1.('1G2)�1de nesfunctionsEA-equivalenttoonede nedbyF.Indeed,('1G2)�1hasoneofthefol-lowingforms:('1F�1)�1(20)=('(1F)�1)�1(18)='('1F)�1de nesNihobentfunctionEA-equivalentto('1F)�1=F('1'F�1)�1,byLemma8de nesNihobentfunctionsEA-equivalentto('1F�1)�1=('(1F)�1)�1(18)='('1F)�1,whichde nesfunctionsEA-equivalenttoonede nedby('1F)�1=F;('1('F)�1)�1(20)=('(1'F)�1)�1(18)='('1'F)�1de nesNihobentfunctionEA-equivalenttoF(byLemma8);27('1('F�1)�1)�1=('1'('F)�1)�1(35)=(1'1('F)�1)�1(20)=1('1('F)�1)�1de nesNihobentfunctionEA-equivalentto('1('F)�1)�1,whichbythepreviouscasede nesNihobentfunctionEA-equivalenttoF.3.Forq3byLemma11,(H1(H2(:::(HqF)�1:::)�1)�18�&#x]TJ ;� -1;.93; Td;&#x [00;:1G�1;('1G)�1;'1G;whereG2fF;('F)�1;'F�1;F�1;('F�1)�1;'Fg.1G�1and'1Gde neNihobentfunctionEA-equivalenttoG�1andGcorrespondingly,which

intheirturnde neNihobentfunctionsEA-
intheirturnde neNihobentfunctionsEA-equivalenttoF;F�1and('F)�1.('1G)�1de nesNihobentfunctionsEA-equivalenttoF.Indeed,forGequalstoF�1;('F)�1;'F�1;('F�1)�1,wealreadyproveitinthecaseq=2.IfG='F,then('1G)�1=('1'F)�1whichde nesNihobentfunctionEA-equivalenttoonede nedbyF(byLemma8).IfG=F,then('1F)�1=F:Proposition3.Themodi edmagicactionandtheinversemapappliedtotheFrobeniusmap,giveexactly3EA-inequivalentfunctionscorrespondingtoF,F�1,(F0)�1.Proof.FortheFrobeniusmapF(x)=x2iwehave:F=(F0)�1=x11�2i.HencebyProposition2,Fcanpotentiallyde ne3EA-inequivalentNihobentfunctionscorrespondingtoF,F0and(F0)�1.This3o-polynomialsde ne3surlyEA-inequivalentNihobentfunctions[7].ThePayneo-polynomialcanberepresentedviaDicksonpolynomials.LetusrecallDicksonPolynomials.Foreverynon-negativeintegerdDicksonpoly-nomialsDd(x)overF2mcanbede nedbyarecursionrelationinthefollowingway:D0(x)=0;D1(x)=x,Dd+2(x)=xDd+1+Dd(x),forallintegersd0.Itsatis esthefollowingroperties:1.DdDd0=Ddd0.2.Ifdisco-primewith2m�1,thenDdisapermutationalpolynomial.UsingDicksonpolynomialswecanprovethefollowingresultsforthePayneo-polynomials.Lemma12.LetF(x)=x16+x12+x56.ThenFc=(Fc)�1foranyc2F2m.Proof.Note rst,thatF(x)=x16+x12+x56=D5(x16).AlsoitiseasytoseethatF0=F.Indeed,F0(x)=xF(x�1)=xD5(x�16)=x(x�16+x�12+x�56)=x16+x12+x56=D5(x16)=F(x):Therefore(F0)�1=F�1,andhence,(Fc)�1=((cF0)0)�1=((cF)0)�1=Fc;foranyc2F2m:28Proposition4.Themodi edmagicactionandtheinversemapappliedtoo-polynomialF(x)=x16+x12+x56canpotentiallygiveEA-inequivalentNihobentfunctionscorrespondingtoo-polynomialsFandFc,c2F2m.Proof.ImmediatelyfollowsfromLemma12.ExampleForm=5wecheckedcomputationallythattheo-polynomialF(x)=D5(x16)overF2mde nes6EA-inequivalentNihobentfunctionscorrespondingtoo-polynomialsF,F�1andFw;Fw3;Fw5,wherewisaprimitiveelementofF2m.RemarkThemodi edmagicactionandtheinversemapappliedtoSubiaco,Adelaideandx2k+x2k+2+x32k+4o-polinomialsFcangiveasequence

ofEA-inequivalentfunctionsde nedbyo-
ofEA-inequivalentfunctionsde nedbyo-polynomialsontheorbitsF,F�1,Fc,(~cF)c,(~c(F0))candsoon.7TheKnownHyperovals1Overtwodecades, nitegeometersdeterminedthestabilizersofallknownhy-perovals.Inthissectionweprovideanexplicitlistofallo-polynomialswhichprovideEA-inequivalentNihobentfunctionsforeveryoftheknownhyperoval.WestartbygivinganoverviewoverthenumberofEA-inequivalentNihobentfunctionsforeachknownhyperoval.NameHyperovalConditionNumberRef.Regularx2m=11[23,Th.4.1]m=21[23,Th.4.1]m32[23,Th.4.2]IrregularTranslationx2im33[23,Th.4.3]Segrex6m=52[23,Th.4.4]m�5odd4[23,Th.4.4]GlynnIx3+4m7odd=2(m+1)=24Th.7GlynnIIx+m=7=4=2Th.7m�7odd=2(m+1)=2=2kform=4k�1;=23k+1form=4k+14Th.71SomeoftheresultswillrepeatSection6.2results.Wedecidedtokeepbothofthem,sinceweuseamixofalgebraicandgeometricapproach.29Cherowitzox+x+2+x3+4m=510[23,Th.4.6]m�5prime4m+2m�2mTh.9m�5oddnC(m)[23]Paynex1=6+x3=6+x5=6m5isprime3m+2m�1�1mTh.8m5isoddnP(m)Th.8Lunelli-Sce(Subiaco)m=4prim.root4=+11[23,Th.4.1]Subiacom=6jAutj=603[32,p.98]m=6jAutj=156[32,p.98]moddm=712[34]moddm�7nS(m)Th.11m0(mod4)m�6nS(m)Th.11m2(mod4)m�6jAutj=10eTh.12m2(mod4)m�6jAutj=5e=25-mTh.12Adelaidem=68[34]m�6mevennA(m)Th.10O'Keefe-Penttilam=512[22,Case2]2Below,forgiveno-polynomialsF1andF2,wedenoteF1=F2ifF1andF2de neEA-equivalentNihobentfunctionsgF1andgF2.Notethatamatrixcorrespondingtothetransformation'cis011010c1=c110;andthat'~c= cF('c).Hence,byTheorem3thehyperovalde nedbytheo-polynomialFcisobtainedfromthehyperovalde nedbyFusingthe2Noticethatthereferenceclaims1+110insteadof1+11orbitsduetoatypo.30followingtransformationmatrix(the rstmatrixintheproductcorrespondstotheinversetransformation):0@0101000011A0@0010 cF cFF(c)=c1001A=0@0 cF cFF(c)=c00110c1A:Thatis,Fc(x)= cFxF1x+c+F(c)�1correspondstothemapAcF:=0@0 cF cFF

(c)=c00110c1A:Alsorecallthatthechoiceofa
(c)=c00110c1A:Alsorecallthatthechoiceofano-polynomialforagivenhyperovalHonlydependsonwhichpointofHischosenasnucleus,sotheo-polynomialisdeter-minedbythepreimageof(0;1;0).WehaveAcF(c;F(c);1)T=( cFF(c)+ cFF(c)=c;1;c+c)T=(0;1;0):Hence,Fc=Fdifandonlyifh(c;F(c);1)iandh(d;F(d);1)ilieinthesamepointorbitofthestabilizerofH.Tosummarize,wehavethefollowing:(a)Fc=Fdifandonlyifh(c;F(c);1)iandh(d;F(d);1)ilieinthesamepointorbit;(b)F=Fcifandonlyifh(0;1;0)iandh(c;F(c);1)ilieinthesamepointorbit;(c)F�1=Fcifandonlyifh(1;0;0)iandh(c;F(c);1)ilieinthesamepointorbit;(d)F=F�1ifandonlyifh(0;1;0)iandh(1;0;0)ilieinthesamepointorbit.Asguidelinedin[8]weusetheknownresultsonorbitsoftheknownhyper-ovalstogettheexplicitnumbersandrepresentationsforo-polynomialswhichprovideo-equivalentbutEA-inequivalentNihobentfunctionsforeachoftheknownhyperoval.Lemma13.Letm3.Thetwoo-polynomialsobtainedfromtheregularhyperovalH,thatisF(x)=x2,are(uptoEA-equivalenceforthecorrespondingNihobentfunctions)FandF�1.Proof.By[23,Th.4.2],onepointorbitisthenucleusNandtheotherpointorbitisHnfNg.Hence,F�1isarepresentativeofthesecondorbit.Lemma14.Letm3.Thethreeo-polynomialsobtainedfromtheirregulartranslationhyperovalH,thatisF(x)=x2iwithi�1co-primetom,are(uptoEA-equivalenceforthecorrespondingNihobentfunctions)F,F�1andF0.Proof.By[23,Th.4.3],onepointorbitisthenucleusN=(0;1;0),anotherpointorbitisN0:=(1;0;0),andthelastpointorbitisHnfN;N0g.Hence,F,F�1,andF0arerepresentativesofthethreeorbits.31Lemma15.Letm5beodd.ConsidertheSegrehyperovalH,thatisF(x)=x6.(a)Ifm=5,thenthetwoo-polynomialsobtainedfromHare(uptoEA-equivalenceforthecorrespondingNihobentfunctions)FandF1.(b)Ifm�5,thenthetwoo-polynomialsobtainedfromHare(uptoEA-equivalenceforthecorrespondingNihobentfunctions)F,F�1,F0,andF1.Proof.By[23,Th.4.4],form=5thepointorbitsofHaref(1;0;0);(0;1;0);(0;0;1)gandalltheremainingpoints.Hence,(0;1;0)and(1;1;1)arerepresentatives,sowecanchooseFandF1asrepresentatives.Form�5the rstorbitsplitsintothreeorbits,sowehavetoaddF�1andF0tothepreviouslist.Theorem7.Thecollineation

stabiliserofaGlynnhyperovalhas4orbitsunl
stabiliserofaGlynnhyperovalhas4orbitsunlessitisoftypeIIandm=7.Proof.FirstconsiderthecaseGlynnI.By[23,Th.4.4]wehave4orbitsunless(3+4)2�(3+4)+10(mod2m�1).Thissimpli esto92m+1+212(m+1)=2+1331+212(m+1)=20(mod2m�1):Onecaneasilycheckthatthisisneversatis ed.NowconsiderthecaseGlynnII.By[23,Th.4.4]wehave4orbitsunless(+)2�(+)+10(mod2m�1).Form=4k�1,thisis2(3m+7)=4�2(m+1)=4+30(mod2m�1):Equivalityholdsonlyform=7asform�7thelefthandsideissmallerthan2m�1.Thecaluclationform=4k+1issimilar.SimilartoLemma15,weobtainthefollowing.Lemma16.Letm7beodd.ConsiderahyperovalHoftypeGlynnIorGlynnII.(a)Ifm=7,thenthetwoo-polynomialsobtainedfromHare(uptoEA-equivalenceforthecorrespondingNihobentfunctions)FandF1.(b)Otherwise,thefouro-polynomialsobtainedfromHare(uptoEA-equivalenceforthecorrespondingNihobentfunctions)F,F�1,F0,andF1.Theorem8.ThenumberoforbitsofthecollineationstabilizerofthePaynehyperovalHisgivenby3+2m�1mifmisaprime.Moregenerally,thenumberoforbitsaregivenbynP(m):=3+X`jm;�`1 F2`n[hj`;hF2h =(2`):ForwaprimitiveelementofFqandc=w2n,wegetFc=Fdifandonlyifd=w2inord=w�2inforsomei2f1;:::;mg.Theo-polynomialsFandF�1de neNihobentfunctionsEA-inequivalenttothosede nedbyallothero-polynomialsfromH.32Proof.By[23,Th.4.5],theorbitsaref(0;1;0)g,f(1;0;0);(0;0;1)g,andsetsHn:=f(wn2i;f(wn2i);1):i=1;:::;mg[f(1;f(wn2i);wn2i):i=1;:::;mg;wherewisaprimitiveelementofFq.NoticethatH0isf(1;1;1)g.FormprimeitiseasytoseethateachorbitHnhaslengthmforn�1,hencethetotalnumberoforbitsis3+2m�1�1m.Ingeneral,ifwn2F`with`jm,thenf(wn)2ig2F`.Thisyieldsthegeneralformula.ThedescriptionoftheequivalenceofFcandFdfollowsdirectlyfromtheexplicitdescriptionoftheorbits.Forexampleform=5,thepreviousresultgivesthefollowingrepresentativesforall6o-polynomialswhichcanbeobtainedfromthePaynehyperoval:F;F�1;F1;Fw;Fw3;Fw5:Theorem9.ThenumberoforbitsofthecollineationstabilizeroftheCherow-itzohyperovalisgivenby4+22m�1�1mifmisaprime.Moregene

rally,thenumberoforbitsaregivenbynC(m):=
rally,thenumberoforbitsaregivenbynC(m):=3+X`jm F(2`)n[hj`;hF2h =`:ForwaprimitiveelementofFqandc=w2n,wegetFc=Fdifandonlyifd=w2inforsomei2f1;:::;mg.TheNihobentfunctionsgFandgF�1arebothEA-inequivalenttoNihobentfunctionsde nedbyallothero-polynomialsfromH.Proof.Corollary4.5in[2]describesthestabilizerasf(x;y;z)7!(x ;y ;z ): 2Aut(Fq)g:TherestofthecalculationissimilartothePaynehyperoval,justthatthistimethe rstandsecondcoordinatecannotbeinterchanged.Theorem10.Let[1]:=+�1.Forc2Fq,letOc:=fc2h+h�1Xi=1[1]2i:i=0;:::;2m�1g:ThenumberofEA-inequivalentNihobentfunctionsobtainedfromtheAdelaidehyperovalisnA(m):=2+jfOc:c2Fqgj.Inparticular,for xedc2Fq,theNihobentfunctionsde nedbytheo-polynomialsF,F�1,FcarepairwiseEA-inequivalent.Furthermore,gFcandgFdareEA-equivalentifandonlyifd2Oc.Proof.In[31,Eq.(9)](inaslightlydi erentrepresentation)thestabilizeroftheAdelaidepolynomialwasdeterminedasthecyclicgroupgeneratedbythemap:x7!0@10[1]01[1]0011A0@xF(x)11A2:33Fromthisitiseasilyveri edthat xes(0;1;0)and(1;0;0),sogFandgF�1arenotEA-equivalenttothosefunctionsde nedbyanyoftheothero-polynomials.Furthermore,itiseasilycheckedthattheorbitof(c;F(c);1)isf(x;F(x);1):x2Ocg:Theorem11.Letm7withm62(mod4),letOc:=fx(�1)i+12i:i=0;:::;2m�1g:ThenumberofEA-inequivalentNihobentfunctionsobtainedfromtheSubiacohyperovalisnS(m):=2+jfOc:c2Fqgj.Inparticular,for xedc6=0;1,theo-polynomialsF,F�1,F0,FcprovidepairwiseEA-inequivalentNihobentfunctions.Furthermore,gFcandgFdareEA-equivalentifandonlyifd2Oc.Proof.By[24,Th.13,Th.16](seealso[15]),thestabilizeroftheSubiacohyperovalHisgeneratedbythemap:x7!0@0010101001A0@xF(x)11A2:Fromthisitiseasilyveri edthat xes(0;1;0),f(1;0;0);(0;0;1)g,(1;1;1),soNihobentfunctionsde nedbyF,F�1=F0,andF1arenotEA-equivalenttothosede nedbyanyothero-polynomialobtainedfromH.Furthermore,itiseasilycheckedthattheorbitof(c;F(c);1)isf(x;F(x);1):x2Ocg:Form2(mod4)therearetwotypesofnon-equivalenthyperovals,see[33].Inpa

rticular,fromTheorem6.6andTheorem6.7in[3
rticular,fromTheorem6.6andTheorem6.7in[33]weobtainthefollowing.Wearenotawareofanynicedescriptionoftheorbitsofthegivengroups,buttheinformationissucienttocalculateallo-polynomialseciently.Theorem12.Letm6withm2(mod4).(a)IfF(x)=2(x4+x)x4+2x2+1+x1=2,thengFisEA-inequivalenttoallgFcandwehaveF�1=F0.Furthermore,Fc=Fdifandonlyif(c;F(c);1)h=(d;F(d);1)foranelementhofthegroup(ofsize10m)generatedby(i)(x;y;z)7!(z;y;x),(ii)(x;y;z)7!(x+z;y+2z;z),(iii)(x;y;z)7!(z2+2x2;z2+y2;z2).(b)IfF(x)=x3+x2+2xx4+2x2+1+x1=2,thengF,gF�1,andgF0arepairwiseEA-inequivalent.Furthermore,Fc=Fdifandonlyif(c;F(c);1)h=(d;F(d);1)hforanelementhofthegroup(ofsize5m=2)generatedby(i)(x;y;z)7!(x;y;z)for2Aut(F)with=,(ii)(x;y;z)7!(z;y+z;x+z).34TheO'Keefe-Penttilahyperovalform=5,whichisnotknowntobelongtoanyin nitefamily,isstabilizedbythegroupgeneratedby0@1011101001A:Hence,mostorbitshavetheformf(c;F(c);1);(1+c�1;1+c�1F(c);1);((1+c)�1;c�1(1+F(c);1)g.Then,representativesforthe14o-polynomialsobtainedfromthehyperovalandde ningEA-inequivalentNihobentfunctionsareF;F�1;Fw;Fw2;Fw4;Fw5;Fw7;Fw8;Fw10;Fw14;Fw16;Fw19:HerewisaprimtiveelementofF25.AcknowledgementTheauthorswouldliketothankAlexanderKholoshaforusefuldiscussions.ThisresearchwassupportedbyTrondMohnStiftelse(TMS)foundation.TheworkofFerdinandIhringerissupportedbyapostdoctoralfellowshipoftheResearchFoundation{Flanders(FWO).References[1]K.Abdukhalikov,"Bentfunctionsandlineoval",FiniteFieldsAppl.,47,pp.97{124,2017.[2]L.Bayens,W.CherowitzoandT.Penttila."GroupsofhyperovalsinDe-sarguesianplanes",Inn.Inc.Geom.,pp.6{7,2007.[3]L.BudaghyanandC.Carlet,\CCZ-equivalenceofsingleandmultioutputBooleanfunctions",AMSContemporaryMath.518,Post-proceedingsoftheconferenceFq9,pp.43{54,2010.[4]L.BudaghyanandC.Carlet,"OnCCZ-equivalenceanditsuseinsecondaryconstructionsofbentfunctions",PreproceedingsofInternationalWorkshoponCodingandCryptographyWCC2009,pp.19{36,2009.[5]L.Budaghyan,A.Kholosha,C.Carlet,andT.Helleseth,\Nihobentfunc-tion

sfromquadratico-monomials",Proceedingsof
sfromquadratico-monomials",Proceedingsofthe2014IEEEInterna-tionalSymposiumonInformationTheory,pp.1827{1831,2014.[6]L.Budaghyan,C.Carlet,T.Helleseth,A.Kholosha,andS.Mesnager,\FurtherresultsonNihobentfunctions",IEEETrans.Inf.Theory,vol.56,no.11,pp.6979{6985,2012.[7]L.Budaghyan,C.Carlet,T.HellesethandA.Kholosha,"Ono-equivalenceofNihoBentfunctions",WAIFI2014,LectureNotesinComp.Sci.9061,pp.155{168,2015.35[8]L.Budaghyan,C.Carlet,T.Helleseth,A.Kholosha,T.Penttila."Pro-jectiveequivalenceofovalsandEA-equivalenceofNihobentfunc-tions",BookofAbstractsof"FinitegeometriesfourthIrseeconfer-ence",Sept.2014;theslidesfromthepresentationcanbefoundathttps://people.uib.no/lbu061/irsee.pdf[9]C.Carlet,\Booleanfunctionsforcryptographyanderror-correctingcodes",BooleanModelsandMethodsinMathematics,ComputerScience,andEngi-neering,ser.EncyclopediaofMathematicsanditsApplications,Y.CramaandP.L.Hammer,Eds.Cambridge:CambridgeUniversityPress,2010,vol.134,ch.8,pp.257{397.[10]C.CarletandS.Mesnager,\OnDillon'sclassHofbentfunctions,Nihobentfunctionsando-polynomials",J.Combin.TheorySer.A,vol.118,no.8,pp.2392{2410,Nov.2011.[11]C.CarletandS.Mesnager.Fourdecadesofresearchonbentfunctions.Designs,CodesandCryptography78(1),pp.5{50,2016.[12]C.Carlet,T.Helleseth,A.Kholosha,andS.Mesnager,\Onthedualofbentfunctionswith2rNihoexponents",Proceedingsofthe2011IEEEInternationalSymposiumonInformationTheory.IEEE,Jul./Aug.2011,pp.657{661.[13]W.E.CherowitzoandL.Storme,\ -Flockswithovalherdsandmonomialhyperovals",FiniteFieldsAppl.,vol.4,no.2,pp.185{199,Apr.1998.[14]W.Cherowitzo,\HyperovalsinDesarguesianplanesofevenorder",Ann.DiscreteMath.,37,pp.87{94,1988.[15]W.Cherowitzo,T.Penttila,I.Pinneri,andG.Royle."FlocksandOvals",Geom.Dedicata,60,pp.17{37,1996.[16]P.Dembowski,FiniteGeometries,Springer,1968.[17]J.F.Dillon,ElementaryHadamarddi erencesets,Ph.D.dissertation,Univ.Maryland,CollegePark.MD,USA,1974.[18]H.Dobbertin,G.Leander,A.Canteaut,C.Carlet,P.Felke,andP.Gaborit,\ConstructionofbentfunctionsviaNihopowerfunctions",J.Combin.TheorySer.A,vol.113,no.5,pp.779{798,Jul.2006.[19]D.Glynn,\Twonewsequencesofovalsin nitedesarguesianplanesofevenorder",CombinatorialMathematics,LectureNotesinMathe

matics,vol.1036,pp.217{229,1983.[20]T.He
matics,vol.1036,pp.217{229,1983.[20]T.Helleseth,A.Kholosha,andS.Mesnager,\NihobentfunctionsandSubiacohyperovals",TheoryandApplicationsofFiniteFields,ser.Con-temporaryMathematics,M.Lavrauw,G.L.Mullen,S.Nikova,D.Panario,andL.Storme,Eds.,vol.579.Providence,RhodeIsland:AmericanMath-ematicalSociety,pp.91{101,2012.[21]C.M.O'KeefeandT.Penttila,\AutomorphismGroupsofGeneralizedQuadranglesviaanUnusualActionofP�L(2;2h)",EuropJ.Combina-torics,33,pp.213{232,2002.36[22]C.M.O'KeefeandT.Penttila."AnewhyperovalinPG(2;32)",J.Geom.,44,pp.117{139,1992.[23]C.M.O'KeefeandT.Penttila."SymmetriesofArcs",J.Combin.TheorySer.A,66,pp.53{67,1994.[24]C.M.O'KeefeandJ.Thas."CollineationsofSubiacoandCherowitzohyperovals",Bull.Belg.Math.Soc.,3,pp.177{192,1996.[25]A.KholoshaandA.Pott,\Bentandrelatedfunctions"inHandbookofFiniteFields,ser.DiscreteMathematicsanditsApplications,G.L.MullenandD.Panario,Eds.London:CRCPress,2013,ch.9.3,pp.255{265.[26]G.LeanderandA.Kholosha,\Bentfunctionswith2rNihoexponents",IEEETrans.Inf.Theory,vol.52,no.12,pp.5529{5532,Dec.2006.[27]N.Li,T.Helleseth,A.Kholosha,andX.Tang,\OntheWalshtransformofaclassoffunctionsfromNihoexponents",IEEETrans.Inf.Theory,vol.59,no.7,pp.4662{4667,Jul.2013.[28]S.Mesnager,"Bentvectorialfunctionsandlinearcodesfromo-polynomials",JournalDesigns,CodesandCryptography.77(1),pp.99{116,2015.[29]R.L.McFarland,\Afamilyofdi erencesetsinnon-cyclicgroups",J.Combin.TheorySer.A,vol.15,no.1,pp.1{10,Jul.1973.[30]K.Nyberg,"S-boxesandRoundFunctionswithControllableLinearityandDi erentialUniformity",ProceedingsofFastSoftwareEncryption,LNCS1008,1994,pp.111-130,1995.[31]S.PayneandJ.A.Thas."ThestabilizeroftheAdelaideoval",DiscreteMath.,294,pp.161{173,2005.[32]T.PenttilaandI.Pinneri."IrregularHyperovalsinPG(64;2)",J.Geom.,51,pp.89{100,1994.[33]S.E.Payne,T.PenttilaandI.Pinneri."IsomorphismsBetweenSubiacoandq-ClanGeometries",Bull.Belg.Math.Soc.,2,pp.197{222,1995.[34]T.PenttilaandG.Royle."OnHyperovalsinSmallProjectivePlanes",J.Geom.,54,pp.91{104,1995.[35]O.S.Rothaus,\On\bent"functions,"J.Combin.TheorySer.A,vol.20,no.3,pp.300{305,May1976.[36]T.L.Vis,\MonomialhyperovalsinDesarguesianplanes,"Ph.D.disserta-tion,UniversityofColoradoDenver,201

Related Contents


Next Show more