/
BLONDEL,JEANDEL,KOIRAN,ANDPORTIERComplexversusrealentries.Throughoutth BLONDEL,JEANDEL,KOIRAN,ANDPORTIERComplexversusrealentries.Throughoutth

BLONDEL,JEANDEL,KOIRAN,ANDPORTIERComplexversusrealentries.Throughoutth - PDF document

jane-oiler
jane-oiler . @jane-oiler
Follow
366 views
Uploaded On 2016-03-11

BLONDEL,JEANDEL,KOIRAN,ANDPORTIERComplexversusrealentries.Throughoutth - PPT Presentation

BLONDELJEANDELKOIRANANDPORTIERwehavethatthesolutionsoftheoriginalPCPproblemarethewordsforwhich0whichisequivalenttoValThevaluestakenbyValarenonnegativeandsotheproblemofdeterminingifthereexistsano ID: 252071

BLONDEL JEANDEL KOIRAN ANDPORTIERwehavethatthesolutionsoftheoriginalPCPproblemarethewordsforwhich=0 whichisequivalenttoValThevaluestakenbyVal)arenonnegativeandsotheproblemofdeterminingifthereexistsano

Share:

Link:

Embed:

Download Presentation from below link

Download Pdf The PPT/PDF document "BLONDEL,JEANDEL,KOIRAN,ANDPORTIERComplex..." 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

BLONDEL,JEANDEL,KOIRAN,ANDPORTIERComplexversusrealentries.Throughoutthepaperwewillassumethattheinitialstate,theunitarymatrices,andtheprojectionmatrixhaverealratherthancomplexentries(i.e.,thesematricesareactuallyorthogonal).Thisisnotasigni“cantrestrictionsinceanyquantumautomaton(withpossiblycomplexentries)canbesimulatedbyanotherquantumautomatonwithrealentriesbydoublingthenumberofstates.Moreprecisely,letbethesetofstatesof.Wereplaceeachelementbytwostates.Letbethe-linearmapwhichsendsacon“guration.Wereplacetheinitialcon“guration).LetbeoneofthematricesofTherowsandcolumnsofareindexedbyelementsof.Letbetheentryatrowandcolumn.Recallthatacomplexnumbercanbeidenti“edtothe22matrixItisthereforenaturaltoreplacethisentrybythe22matrixThetworowsandtwocolumnsofthismatrixareindexed,respectively,by.Byabuseofnotationwealsodenotebythemapwhichsends.Itiseasybutinstructivetocheckthatforanyandforanycomplexmatricesthefollowingrelationshold:,and).Nowrecallthatunitarymatrices,orthogonalmatrices,complexmatricesoforthogonalprojection,andrealmatricesoforthogonalprojectionare,respectively,characterizedbythefollowingrelations:,and.Itfollowsthatsendsunitarymatricestoorthogonalmatrices,andcomplexmatricesoforthogonalprojectiontorealmatricesoforthogonalprojection.Thequantumautomatonde“nedbytheorthogonalmatrices),theprojectionmatrix),andtheinitialcon“gurationforanyword.HenceVal)=Val)foranyword2.Undecidabilityfornonstrictinequality.Weproveinthissectionthattheproblemofdeterminingifaquantumautomatahasawordofvaluelargerthanorequaltosomethresholdisundecidable.TheproofisbyreductionfromPostscorrespondenceproblem(PCP),awell-knownundecidableproblem.AninstanceofPCPisgivenbya“nitealphabetandpairsofwords(.Asolutiontothecorrespondenceisanynonemptywordoverthealphabetsuchthat,where.Thiscorrespondenceproblemisknowntobeundecidable:thereisnoalgorithmthatdecidesifagiveninstancehasasolution[Pos46].Itiseasytoseethattheproblemremainsundecidablewhenthealphabetcontainsonlytwoletters.Theproblemisalsoknowntobeundecidablefor=7pairs[MS05]butisdecidablefor=2pairs;thedecidabilityofthecases27isnotyetknown.Wearenowreadytostateour“rstresult.Theorem2.1.ThereisnoalgorithmthatdecidesforagivenautomatonthereexistsanonemptywordforwhichVal,orifthereexistsoneforVal.Thesetwoproblemsremainundecidableeveniftheautomatonisgivenbyorthogonalmatricesindimension BLONDEL,JEANDEL,KOIRAN,ANDPORTIERwehavethatthesolutionsoftheoriginalPCPproblemarethewordsforwhich=0,whichisequivalenttoValThevaluestakenbyVal)arenonnegativeandsotheproblemofdeterminingifthereexistsanonemptywordsuchthatVal0isundecidable.Noticealso=1andsowithequalityonlyforThus,theproblemofdeterminingifthereexistsanonemptywordsuchthatVal1isundecidabletoo. Theorem2.1dealsonlywithnonemptywords.Weremovethisrestrictioninthenextresult,andwereducethenumberofmatricesfrom7to2.Corollary2.2.ThereisnoalgorithmthatdecidesforagivenautomatonifthereexistsawordforwhichVal,orifthereexistsoneforwhichVal.Theseproblemsremainundecidableeveniftheautomatonisgivenbyorthogonalmatricesindimension,orbyorthogonalmatricesindimensionProof.AsintheproofofTheorem2.1,theundecidabilityresultsfortheconditionVal1followfromthosefortheconditionVal0.Hencewesupplytheproofsforthelatterconditiononly.WeproceedbyreductionfromtheproblemVal0for7matricesindimension6,whichisundecidablefornonemptywordsasshowninTheorem2.1.NotethatthelanguageofthenonemptysuchthatVal0istheunionofthesevenlanguagesde“nedbytheconditionsVal0forpossiblyemptywords.Hencetheemptinessofoneoftheselanguages(say,the“rstone)mustbeundecidable.Thus,theproblemofdeterminingifthereexistsawordsuchthatVal0isundecidable.Foreachautomaton=((,s,P)wecannowconstructthequantum=((,y,P),where.ThenVal0ifandonlyifValThefollowingproblemisthereforeundecidable:givenaquantumautomatonde“nedby7orthogonalmatricesindimension6,istherea(possiblyempty)wordsuchthatValFinally,weshowhowtoreducethenumberofmatricesto2.WeuseaconstructionfromBlondelandTsitsiklis[BT97]andBlondelandCaterini[BC03].Giventheaboveandtheprojectionmatrix,wede“ne...YWhentakingproductsofthesetwomatricesthematrixactsasaselectingmatrixŽontheblocksof.Letusde“ne...P Itisnotdiculttoshowthatthe6otherproblemsmustbeundecidableaswell. BLONDEL,JEANDEL,KOIRAN,ANDPORTIERisclearlysemidecidable.Inordertoshowthatitisdecidable,itremainstoexhibitaprocedurethathaltswhenValforallLetaquantumautomatabegivenbya“nitesetoforthogonaltransition,aninitialcon“gurationofunitnorm,andaprojectionmatrix.ThevalueofthewordisgivenbyVal.Letbethesemigroupgeneratedbythematrices,andletbethefunctionde“nedbysXP.WehavethatValandtheproblemisnowthatofdeterminingifforall.Thefunctionisa(continuous)polynomialmapandsothisconditionisequivalenttoforall ,where istheclosureof.Theset hastheinterestingpropertythatitisalgebraic(seebelowforaproof),andsothereexistpolynomial,suchthat isexactlythesetofcommonzeros.Ifthepolynomialsareknown,theproblemofdeterminingforall canbewrittenasaquanti“ereliminationproblem ··· )=0)=Thisisa“rst-orderformulaovertherealsandcanbedecidedeectivelybyTarski…Seidenbergeliminationmethods(see[Ren92a,Ren92b,Ren92c,BPR96]forasurveyofknownalgorithms).Ifweknewhowtoeectivelycomputethepolynomialsfromthematrices,adecisionalgorithmwouldthereforefollowimmediately.Inthefollowingwesolveasimplerproblem:weeectivelycomputeasequenceofpolynomialswhosezerosdescribethesameset after“nitelymanyterms(butwemayneverknowhowmany).Itturnsoutthatthisissucientforourpurposes.Wewillusesomebasicalgebraicgeometry.Inparticular,wewillusetheNoether(ordescendingchainŽ)property:inany“eld,thesetofcommonzerosofasetof-variatepolynomialsisequaltothesetofcommonzerosofasubsetofthesepolynomials(seeanytextbookonalgebraicgeometry,forinstance,[CLO92,Prop.1,sect.4.6]).Theorem3.1.Letbeorthogonalmatricesofdimensionandlet theclosureofthesemigroup.Theset isalgebraic,andiftherationalentries,wecaneectivelycomputeasequenceofpolynomialssuchthat forallthereexistssomesuchthat Proof.We“rstprovethat isalgebraic.Itisknown(see,e.g.,[OV90])thateverycompactgroupofrealmatricesisalgebraic.Infact,theproofofalgebraicityin[OV90]revealsthatanycompactgroupofrealmatricesofsizeisthezerosetsetX]G={fR[X]:f(I)=0andisthezerosetofthepolynomialsinvariableswhichvanishattheidentityandareinvariantundertheactionof.Wewillusethispropertylaterintheproof.Toshowthat isalgebraic,itsucestoshowthat iscompactandisagroup.Theset isobviouslycompact(boundedandclosedinanormedvectorspaceof“nitedimension).Letusshowthatitisagroup.Itisinfactknownthateverycompactsubsemigroupofatopologicalgroupisasubgroup.Hereisaself-containedproofinoursetting:Foreverymatrix,thesequenceadmitsasubsequencethatisa BLONDEL,JEANDEL,KOIRAN,ANDPORTIERalgebraicnumbersover.Thispurelyalgebraicinformationissucienttocomputethesequenceofpolynomials()inTheorem3.1.Wealsoneedtodecideforeveryinitialsegmentwhether(3.1)holds.Afterquanti“erelimination,thisamountstocomputingthesignofa“nitenumberofpolynomialfunctionsoftheelementsofInordertodothisweneedonlyassumethatwehaveaccesstoanoraclewhichforanyelementandany0outputsarationalnumbersuchthat(suchanoraclecanbeeectivelyimplementediftheentriesarecomputablerealnumbers).Weusethealgebraicinformationtodeterminewhetherapolynomialtakesthevaluezero,andifnotweuseapproximationstodetermineitssign.IntheproofofTheorem3.2wehavebypassedtheproblemofexplicitlycomputinga“nitesetofpolynomialsde“ning .Itisinfactpossibletoshowthatthisproblemisalgorithmicallysolvable[DJK03].Thisimpliesinparticularthatthefollowingtwoproblemsaredecidable:(i)Decidewhetheragiventhresholdisisolated.(ii)DecidewhetheragivenQFAhasanisolatedthreshold.AthresholdissaidtobeisolatedifVal�.Itisknownthatthesetwoproblemsareundecidableforprobabilisticautomata[Ber75,BMT77,BC03].Thealgorithmof[DJK03]forcomputing alsohasapplicationstoquantumcircuits:thisalgorithmcanbeusedtodecidewhetheragivensetofquantumgatesiscomplete(completemeansthatanyorthogonaltransformationcanbeapproximatedtoanydesiredaccuracybyaquantumcircuitmadeupofgatesfromtheset).Mucheorthasbeendevotedtotheconstructionofspeci“ccompletesetsofgates[DBE95,95],butnogeneralalgorithmfordecidingwhetheragivensetiscompletewasknown.Finally,wenotethattheproofofTheorem3.2doesnotyieldanyboundonthecomplexityofproblems(i)and(ii).Wehopetoinvestigatethisquestioninfuturework.Acknowledgment.P.K.wouldliketothankEtienneGhysforpointingoutreference[OV90].Wearealsogratefultotheanonymousrefereeforhisverycarefulreadingofthemanuscript.uscript.M.AmanoandK.IwamaUndecidabilityonquantum“niteautomata,inProceedingsofthe31stACMSymposiumonTheoryofComputing,ACM,NewYork,1999,pp.368…375.368…375.+95]A.Barenco,C.H.Bennett,R.Cleve,D.P.DiVincenzo,N.H.Margolus,P.W.Shor,T.Sleator,J.A.Smolin,andH.WeinfurterElementarygatesforquan-tumcomputation,Phys.Rev.A,52(1995),pp.3457…3467.3457…3467.S.Basu,R.Pollack,andM.-F.RoyOnthecombinatorialandalgebraiccomplexityofquanti“erelimination,J.ACM,43(1996),pp.1002…1045.1002…1045.V.D.BlondelandV.CanteriniUndecidableproblemsforprobabilisticautomataof“xeddimension,TheoryComput.Syst.,36(2003),pp.231…245.231…245.A.BertoniThesolutionofproblemsrelativetoprobabilisticautomataintheframeoftheformallanguagestheory,inVierteJahrestagungderGesellschaftf¨urInformatik,LectureNotesinComput.Sci.26,Springer,Berlin,1975,pp.107…112.107…112.A.Bertoni,G.Mauri,andM.TorelliSomerecursivelyunsolvableproblemsrelatingtoisolatedcutpointsinprobabilisticautomata,inProceedingsofthe4thInternational