ebook img

Two infinite families of nonadditive quantum error-correcting codes PDF

1.2 MB·English
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Two infinite families of nonadditive quantum error-correcting codes

Twoinfinitefamiliesofnonadditivequantumerror-correctingcodes Sixia Yu1,2, Qing Chen1,2, and C. H. Oh2 1HefeiNationalLaboratoryforPhysicalSciencesatMicroscaleandDepartmentofModernPhysics& DepartmentofModernPhysicsUniversityofScienceandTechnologyofChina,Hefei230026,P.R.China 2PhysicsDepartment, NationalUniversityofSingapore, 2ScienceDrive3, Singapore117542 We construct explicitly two infinite families of genuine nonadditive 1-error correcting quantum codes and provethattheircodingsubspacesare50%largerthanthoseoftheoptimalstabilizercodesofthesameparam- etersviathelinearprogrammingbound. Allthesenonadditivecodescanbecharacterizedbyastabilizer-like structureandthustheirencodingcircuitscanbedesignedinastraightforwardmanner. One major family of quantum error-correcting codes whenthelengthtendstoinfinitythatoutperformsallthesta- (QECCs) [1, 2, 3, 4], which are powerful tools to fight the bilizercodesofthesameparameters. 9 quantum noises in various quantum informational processes, InthisLetterweshallconstructtwoinfinitefamiliesofgen- 0 0 are called as additive or stabilizer codes [5, 6, 7, 8]. The uine nonadditive 1-error-correcting codes with coding sub- 2 coding subspace of a stabilizer code is specified by the joint spaces being 50% larger then the corresponding optimal 1- +1eigenspaceagroupofcommutingmultilocal(directprod- error-correcting stabilizer codes of the same parameters to n a uctof)Paulioperators. Usually[[n,k,d]]denotesastabilizer show that the nonadditive error-correcting codes outperform J codeoflengthn,thenumberofphysicalqubits,anddistance the stabilizer codes even when the length n tends to infinity. 14 dic,ail.eq.u,cboitrsre(2ctki-ndgimupentosi(cid:98)odn−a2l1s(cid:99)u-qbuspbaitceer)r.ors,thatencodesklog- Astrllutchtuerneoannadddthiteivreefcooredethseareencchoadriancgt-edreizceoddibnygacsirtacbuiiltiszcear-nlibkee ] The first example of nonadditive codes, codes that cannot designedinastraightforwardmanner. h be described within the framework of stabilizer, was an infi- Two families of nonadditive 1-error correcting codes that p nitefamilyof1-error-detectingcodes[9,10],e.g.,((5,6,2)), weshallconstructhavethefollowingparameters - nt wbiiltihzecrocdoidnegssoufbtshpeacsaemsbeepinagra5m0e%terlsa.rgReerctehnatnlythaenootphteimrfaalmsitlay- D(m,a) =((Nma,322Nma−2m−6,3)) (1) a qu ohfas1-beereronrcdoentesctrtuincgtecdodines[1w1i]thasntdillsllairgghetrlyenimcopdrionvgesdubinsp[a1c2e]s. where Nma = 22m+35−5 +a with a = 0,1 and m ≥ 1. To ensure that they are genuine nonadditive we shall prove that [ Here we have denoted by ((n,K,d)) a nonadditive code of the corresponding optimal 1-error-correcting stabilizer codes 1 lengthnanddistancedthatencodesaK-dimensionallogical ofthesamelengthhaveparameters v subspace(aboutlog K logicalqubits). 2 35 The first example of nonadditive code [13], namely [[Nma,Nma −2m−6,3]] (2) ((9,12,3)), that outperforms all the stabilizer codes of the 9 byworkingoutanalyticallythelinearprogrammingboundfor 1 samelengthwhilecapableofcorrectingarbitrarysinglequbit the lengths Na. Notice that the quantum Hamming bound . errorshasrecentlybeenconstructedviaagraphicalapproach m 1 permits exactly one more logical qubit, i.e., log (3Na + 0 based on graph states. Later on an optimal 10-qubit code (cid:100) 2 m 1) =2m+5. Thefirstnonadditivecodesofthesetwofami- 9 ((10,24,3)) has been found via a comprehensive computer (cid:101) liesare((41,3 232,3))and((42,3 233,3))respectivelywhile 0 search [12]. Recently a family of codes of distance 8 that · · : the optimal stabilizer codes have the parameters [[41,33,3]] v encode 3 more logical qubits than the best known stabilizer and[[42,34,3]]. i codes have been constructed based on nonlinear classical X Ourconstructionisbasedonafamilyofstabilizercodesof codes [14]. However the possibility of being equivalent to ar somesubcodeofanoptimalstabilizercodeorevenastabilizer lengths{22r+3}mr=1[5,8]andtwononadditivecodesoflength 9and10discoveredrecently[12,13]andisakindofpasting code of the same parameters under local unitary transforma- stabilizer codes with nonadditive codes that generalizes the tionshasnotyetbeenexcluded. pastingofstabilizercodesinRef.[16]. Wedenoteby , v v Generally, being without a stabilizer structure, the nonad- X Z three Pauli operators acting nontrivially only on some qubit ditive codes promise larger coding subspaces while they are labeledbyv andby theidentityoperator. Furthermorefor hardertoconstructandidentifythanthestabilizercodes. On I agivenindexsetU wedenote = andsimilarly onehandthereisnosystematicconstructionsofarandallthe forotherPaulioperators. XU v∈UX goodcodesarefoundviacomputersearch[12, 15], whichis Letuslookattheoptimalstabilizero(cid:81)flength22r+3atfirst. impossibleforarelativelylargelength(e.g. n 11). Onthe AccordingtoRef.[5]thestabilizerofthecodehas2r+5gen- ≥ otherhandanobviouscriterionforagenuinenonadditivecode eratorswithtwoofthembeing and wherewehave is to check whether or not its coding subspace is larger than labeled22r+3physicalqubitswiXthUUr = Z1U,r2,3,...,22r+3 . r allthestabilizercodesofthesameparameters. Howeverthe { } Theremaining2r+3generatorsaregivenby exact bound for stabilizer codes is generally unknown. As a resultitisofinteresttofindnonadditiveerror-correctingcode r = hk hk−1+h1+h2r+3 k U . (3) Sk X Z | ∈ r (cid:8) (cid:9) 2 HHeerreehh ddeennootteessaa2222rr++33--ddiimmvveeccttoorrtthhaattiisstthheekk--tthhrrooww wweehhaavveeddeennootteedd ==11((11++ ++ ))..IIffwweeddeennoottee kk VVaabb 22 XXaa XXbb−−XXaaXXbb ooffaa((22rr++33)) 2222rr++33mmaattrriixxHH == [[cc,,cc,,......,,cc ]] == wweehhaavveeTTrr ==2288.. IIttsshhoouullddbbeennootteedd wwhhoosseekk--tthhccooll×u×ummnncc bbeeiinnggtthheebrbriinnaarryyrr0e0epprr1e1esseennttaattii2o2o2n2nrr++oo33f−f−k1k1,, AttAhha0a0tttthheeAAcco1o1AdAdee22iAiAss3n3noonnddeeggeenneerraattVeeV,0,0AwwAh0h0iicchhccaannbbeeeeaassiillyysseeeennffrroomm kk ee..gg..,,ccTT == ((00,,00,,......,,11))aannddcc == ((11,,11,,......,,11))aanndd iittsswweeiigghhttddiissttrriibbuuttiioonnss.. 11 2222rr++33 11 hh00 == 00iisstthheezzeerroovveeccttoorr.. AAnnddffoo−r−raavveeccttoorrhhwwiitthhccoomm-- TThheennoonnaaddddiittiivveeooppttiimmaallccooddee((((1100,,2244,,33))))≡≡DD((00,,11))hhaassaa appanonodndneennththss={={hhvv||vv∈∈UUhhrrvv}}..wweehhaavveeddeennootteeddXXhh== vv∈∈UUrrXXvhvhvv gogornrnaa1p1p0h0h-v-vssetetararttttieieccebebsasasasaisissssschchoooorwrwrrenensspipinnooFnnFdidigigin.n.11ggbbttaoaonntdtdhhtetehhgeegrcracaopoprrhrhreeGsGspp11oon=n=ddii(nn(VgVg1g1g,,rrEaEap1p1h)h) DDZeZessppiitteetthhveev∈i∈irrUUrnrnoZoZnnvvaaddddiittiivveenneesssstthheeccooddeess((((99,,Q(cid:81)1122,,33))))aanndd ssttaatteeiiss GG11.. OObbvviioouussllyytthheeggrraapphhiissuunncchhaannggeedduunnddeerrtthhee ((1100,,2244,,33))))Q(cid:81)aaddmmiittaassttaabbiilliizzeerr--lliikkeessttrruuccttuurreeaannddccaannbbeemmoosstt ffoolllloowwiinn|g|gttwwi(cid:105)ooppeerrmmuuttaattiioonnss ccoonnvveenniieennttllyyffoorrmmuullaatteeddbbyyuussiinnggtthheeggrraapphhssttaatteess[[2200,,2211]].. ππ==((1144))((2233))((6699))((7788)),, ττ==((1122))((3344))((6677))((8899)).. ((66)) WWeeddeennootteebbyyGG==((VV,,EE))aassiimmpplleeuunnddiirreecctteeddggrraapphhwwiitthhaa sseettVV ooffvveerrttiicceessaannddaasseettEEooffeeddggeess..TTwwoovveerrttiicceessaarreeccoonn-- tthhaattaaccttnnaattuurraallllyyoonnVV.. TTwwoosswwaappppiinnggooppeerraattoorrss aanndd ππ onnoneenccttVeVed0d0ww==iitthh99aanananneeddddggVeVe1i1iffff={={aa,1,1b0b0}}vv∈∈eerrEtEtiicc..eeTsTswwaaororeeggsrsrhahapopowhhwssnnGGiinana(F(Faaiigg==..11.0.0,B,B11y)y) MaMarrbbτiτittraraaarrrereyyddCCeefifinneeVdVd.v.viFiFaao,o,rerel.l.gagat.t.e,e,rMrMuussπeπeZZwwCCee||+d+deei(cid:105)fiVxfiVxn1n1ee==ttwwZZooππc(c(oCoCn)n)t|t|r+r+ooil(cid:105)lMlVxlMVxee11ddf--fossowrwraaaanppn llaabbe|e|lliinn|g|g VV qquubbiit|ts|sbby|y|VVwweeccaannddeefifinneetthheeggrraapphhssttaatteeccoorrrree-- ooppeerraattiioonnssww⊆⊆iitthhqquubbiittss00aanndd55aassssoouurrcceessrreessppeeccttiivveellyyaass || || ssppoonnddiinnggttootthheeggrraapphhGGaa GG == ++VVwwhheerree || i(cid:105) UUGG|| i(cid:105)xx 11 == 11++ ++((11 )) ,, ((77aa)) == 11++ZZaa++ZZbb−−ZZaaZZbb,, ((44)) TTππ 22 XX00 −−XX00MMππ UUGG 22 11(cid:0)(cid:0) (cid:1)(cid:1) {{aa,Y,(cid:89)bb}}∈∈EE TTττ == 22 11++XX55++((11−−XX55))MMττ ,, ((77bb)) aanndd||++i(cid:105)VxVx ddeennootteesstthheejjooiinntt++11eeiiggeennssttaatteeooffXXvvffoorrvv∈∈VV.. wwhhiicchhccaannbbeerreeaaddiillyy(cid:0)i(cid:0)immpplleemmeenntteeddvviiaaTTooffflfliiaa(cid:1)n(cid:1)nddccoonnttrrooll--nnoott OObbvviioouussllyyUUG2G2==11aannddtthheeggrraapphhssttaattee||GGi(cid:105)iissaallssootthhee++11jjooiinntt ggaatteess.. AAccccoorrddiinnggttooRReeff..[[1122]],,tthhoossee2244ggrraapphh--ssttaatteebbaasseessooff eeiiggFFeernrnoossmtmtaatttetehhoeoeffgtgthrhraeaeppfhfhoolsllsltotoaawwtteeiinngGgGnnsasattaabbbbaaiislsliiiizszseefrfrososrrGGtvthvhe=e=wwUUhhGGooXlXleevvUsUsyGyGss.t.teemmaa 1tt1hh,,e2e2,c,c.o.o.d.d..ee,,6D6Da(a(0n0n,,d1d1))µµaa,r,rνeνeg=g=iivv0e0e,n,n11bbwwyyhh{{ee|r|CrCeeµiµ2i2νν44i(cid:105)ss=u=ubbsZsZeeCtCtssµiµioνoν|ff|GGVV11i(cid:105)a}a}rrewewiitthhii== || i(cid:105) 11 bbaassiissoofftthheewwhhoolleessyysstteemmccaannbbuuiilltt GG CC VV ..FFoorr CC ttrriivviiaallggrraapphhwwiitthhnnooeeddggeetthheeggrraapphh--ss{t{taZaZtteeb|b|aassii(cid:105)iss||rreeaadd⊆s⊆sZZCC}|}|++i(cid:105)VxVx.. CCµiµiνν==ππµµ◦◦ττνν((CCii))△(cid:52)ννBB△(cid:52)µµττνν((AA)) ((88)) AAnnyyggiivveennccoolllleeccttiioonnooffvveerrtteexxssuubbsseettssooffVVwwiillllddeefifinneeaabbaassiiss wwiitthhAA BB==AA BB AA BBbbeeiinnggtthheessyymmmmeettrriiccddiiffffeerreennccee tthhaattssppaannssaassuubbssppaacceewwhhiicchhiissrreeffeerrrreeddttooaasstthheeggrraapphh--ssttaattee △(cid:52) ∪∪ −− ∩∩ bbeettwweeeennttwwoosseettssAAaannddBBaanndd bbaassiiss.. WWiitthhaaggrraapphhaannddaaccoolllleeccttiioonnooffvveerrtteexxssuubbsseettsswwiillll ddeefifinneeaassuubbssppaaccee.. AA== 00,,22,,33 ,,BB== 55,,11,,22 ,,CC == ,, 11 {{ }} {{ }} ∅∅ FFoorrtthheennoonnaaddddiittiivveeccooddee((((99,,1122,,33))))≡≡DD((00,,00))wweeccoonnssiiddeerr CC22=={{11,,22,,33,,99}},,CC33=={{11,,22,,77,,88}},,CC44=={{11,,22,,66,,77,,99}},,((99)) tthheellooooppggrraapphhGG ==((VV,,EE))oonn99vveerrttiicceessaasssshhoowwnniinnFFiigg..11aa CC == 11,,33,,77,,88,,99 ,,CC == 11,,33,,44,,66,,77,,99 .. 00 00 00 55 66 {{ }} {{ }} aannddccoorrrreessppoonnddiinnggggrraapphhssttaattee GG ..TThheejjooiinntt++11eeiiggeennssppaaccee 00 oofftthheeffoolllloowwiinngg66oobbsseerrvvaabblleess|| i(cid:105) IIttttuurrnnssoouutttthhaatttthheennoonnaaddddiittiivveeccooddeeDD((00,,11))aaddmmiittssaallssooaa ssttaabbiilliizzeerr--lliikkeessttrruuccttuurree.. WWiitthhtthheehheellppoofftthheeffoolllloowwiinnggeenn-- αα == ,, ((55aa)) ccooddiinnggooppeerraattoorr 11 UUGG00XX{{33,,88}}UUGG00 αα == ,, ((55bb)) 22 UUGG00XX{{66,,22}}UUGG00 UUeenncc==ZZ22UUGG11TTττTTππZZ22 ((1100)) αα == ,, ((55cc)) 33 UUGG00XX{{99,,55}}UUGG00 iittccaannbbeerreeaaddiillyycchheecckkeeddtthhaatttthheeffoolllloowwiinngg66mmuuttuuaallllyyccoomm-- == (( )) ,, ((55dd)) AA11 UUGG00XX{{44,,77,,33,,66,,99}}VV6699UUGG00 mmuuttiinnggoobbsseerrvvaabblleess == (( )) ,, ((55ee)) iisseexxaaccttllyytthhee1AA1AA2223-23-ddi=i=mmcUUcUUooGGdGGd00i00in(n(gXXgXXs{{s{{u11u11,,b,,b7474s,,s,,33p33p,,a,,69a69c}}c}}eVVeVVo33o3396f96f)t)tUUhUUheGGeGGc00c00o,o,ddeeDD((00,,00))..HH((e5e5rfrfe)e) ββββ2112 ==== UUUUeeeennnnccccXXXX{{{{2626,,,,7337,,,,7878}}}}UUUUee††ee††nnnncccc,,,, ((((11111111baba)))) ββ33 == UUeennccXX{{33,,44,,66,,99}}UUe†e†nncc,, ((1111cc)) ! 9 ! ! BB00 == UUeenncc((XX66VV6677))UUe†e†nncc,, ((1111dd)) !2 1 !9 7 5 8 ! ! BB11 == UUeenncc XX{{11,,22}}VV3377 UUe†e†nncc,, ((1111ee)) !3 !8 6 ! ! BB22 == UUeenncc(cid:0)(((cid:0)XX44VV3366))UUe†e†(cid:1)n(cid:1)ncc ((1111ff)) 4 ! ! !4 !7 2 ssttaabbiilliizzeetthheeccooddiinnggssuubbssppaacceeooffDD((00,,11)),,ii..ee..,,oonneeoonneehhaanndd (a) !5 !6 (b) 1 0 3 ! oβoβnkn !k||tCtChhµeiµeiννoio(cid:105)tth=h=eerrBBhhlal|a|CnCnddµiµiνtνthih(cid:105)ee==jjoo|i|iCnCntµtiµi+ν+νi1(cid:105)1fefoeoiirgrgeaeanlnllslsppppaoaocscsesesiibobolflfeetthkhke,e,slsle,e,ii6,6,µµoo,,bbννsseaearnrvnvd-d- aabblleess,,wwhhoosseepprroojjeeccttoorriissggiivveennbbyy FFIIGG..11:: TTwwooggrraapphhssGGaa == ((VVaa,,EEaa))aarreesshhoowwnnwwiitthhVV00 == 33 22 {b{ba1a1s,s,e2e2s,s,f.f.o.o.r.r.,t,tw9w9}o}onananonodndnaaVdVd1d1di=it=tiivve{e{0c0c,o,o1d1d,e,e.s.s..(.(.((,9,999,,}1}1.2.2T,T,3h3h)e)e)y)yaapnpnrdrodov(v(i(i(d1d1e0e0,t,th2h2e4e4,g,g3r3ra)a)p)p).h.h--ssttaattee PP11== 11++22ββii 11++22BBii,, ((1122)) ii==11 ii==00 Y(cid:89) Y(cid:89) 3 TABLEI:ThestabilizingobservablesofthenonadditivecodesDam wnoetihcainvgeTrOi(a) =0fori≤2m+5andTrO2(am)+6 =2Nma−1 whosephysicalqubitsarelabeledwithUm ... U1 Va(a=0,1). ∪ ∪ ∪ TheblankentriesstandforsuitableidentityoperatorsIUk orIVa. TrPma = Tr(12+2mO+2(6am)+6) = 232Nma−2m−6. (14) Um Um−1 U2 U1 V0orV1 Thusweobtainthe1-errorcorrectingcodeofparametersex- ··· (0,1) O1 XUm actly as given in Eq.(1). Now we shall demonstrate that its (0,1) coding subspace is 50% larger than the corresponding opti- O2 ZUm (0,1) m malstabilizercodessothatourcodesaregenuinenonadditive O3(0,1) S1m XUm−1 codesthatareneitherequivalenttosomestabilizercodesun- O4 S2 ZUm−1 derlocalunitarytransformationsnorsubcodesofsomelarger (0,1) m m−1 O5 S3 S1 1-errorcorrectingstabilizercodesofthesamelength. ... ... ... ... ThequantumHammingboundfora1-errorcorrectingsta- O2(0m,1−)4 S2mm−6 S2mm−−18 ··· b[[inli,zke,r3c]o],deb,eei.ngg.,inn−trokdu≥ce(cid:100)dloign2i(ti3anlly+1fo)r(cid:101)fthoeransotnab-dileigzeenrecroadtee (0,1) m m−1 O2m−3 S2m−5 S2m−7 ··· XU2 codes,isvalidforbothdegenerateandgeneratecodesofdis- (0,1) m m−1 O2m−2 S2m−4 S2m−6 ··· ZU2 tance 3 and 5 [6] and of a large enough length [19]. In the O2(0m,1−)1 S2mm−3 S2mm−−15 ··· S12 XU1 case of n = Nma we have the quantum Hamming bound (0,1) m m−1 2 n k 2m + 5. This is not enough to prove the nonad- O2m S2m−2 S2m−4 ··· S2 ZU1 − ≥ (0,1) m m−1 2 1 α orβ ditivenessofourcodes. Howeverbyworkingoutanalytically O2m+1 S2m−1 S2m−3 ··· S3 S1 1 1 (0,1) m m−1 2 1 α orβ thelinearprogrammingboundwehave O2m+2 S2m S2m−2 ··· S4 S2 2 2 Theorem If there exists a stabilizer code [[Na,k,3]], de- (0,1) m m−1 2 1 α orβ m O2(0m,1+)3 S2mm+1 S2mm−−11 ··· S52 S31 3 or 3 generate or non-degenerate, with Nma = 22m+35−5 + a and O2m+4 S2m+2 S2m ··· S6 S4 A1 B1 m 0anda=0,1thenNa k 2m+6. (0,1) m m−1 2 1 or ≥ m− ≥ O2m+5 S2m+3 S2m+1 ··· S7 S5 A2 B2 Proof. Givenastabilizercode[[n,k,d]]itsweightdistribu- O2(0m,1+)6 A0orB0 tions{Ai}ni=0isdefinedby 1 A = Tr(P )2 (i=0,1,...,n), (15) i 22k | Eω | hasexactlydimension24, i.e., TrV1P = 24sinceTrV1B0 = |(cid:88)ω|=i 29. An encoding circuit can therefore be designed in a simi- wherethesummationisoverallerrorssupportedon iqubits larmannerasthatof((9,12,3))[13]. Wenotealsothatthis andP istheprojectorontothecodingsubspace. Itisobvious nonadditivecodeisnon-degenerate. thatA 0,A =1,and A =2ssothat A /2s canbe Nowwearereadytopresentourconstruction. Weconsider i ≥ 0 i i { i } regardedasaprobabilitydistributionwiths = n k. Foran NUma qVubiwtsitahndUlabe=lt2h2ekm+3byanddisjVoint=se9t+Uma∪wiUthmk−1∪m..a.n∪d arbitraryfunctionf(x)we(cid:80)denoteitsaverage − 1 a k a ∪ | | | | ≤ a=0,1.Weclaimthatthejoint+1eigenspaceofthose2m+ d 1 6observables{Oi(a)}2i=m1+6witha=0or1asdefinedinTable (cid:104)f(x)(cid:105)≡ 2s f(i)Ai. (16) IisthecodeD(m,a)inEq.(1)withthefollowingprojectoronto (cid:88)i=0 thecodingsubspace Inthefollowingweshallformulateasubsetofthelinearpro- 2m+61+ (a) grammingboundfor1-errorcorrectingcode,whichservesour a = Oi . (13) purposeperfectly. Foracompletesetoflinearprogramming Pm 2 boundseeRef.[8,17]. i=1 (cid:89) Linear Programming bound (Restricted) If there exists a InTableIobservables r aredefinedinEq.(3)and α , , Si { i Aj} stabilizer code [[n,k,3]] then the following conditions hold β , are defined via Eq.(5) and Eq.(11) respectively. { i Bj} true Blank entries represent suitable identity operators. By jux- taposition of some operators in the same row we mean their A = 3n 4x , (17a) 1 (cid:104) − (cid:105) directproduct. 1 Firstofall,these2m+6stabilizingobservablesdetectall A2 = (4x 3n+1)2 3n 1 , (17b) 2(cid:104) − − − (cid:105) 2-qubiterrorsbecausefirstlyallerrorshappenedonU-blocks n orV blockscanbedetectedbecauseallthesubcodesarepure (cid:98)2(cid:99)A2i 2s−1. (17c) 1-errorcorrectingcodesandsecondlyalltwoerrorshappened ≥ i=0 on different qubit blocks can always be detected by the sta- (cid:88) bilizercontaining and forsomek. Thusweobtain Inthecaseofa=0,i.e.,n=N0 withm 0weintroduce XUk ZUk m ≥ a pure 1-error correcting codes of length Na. Secondly, by anonnegativefunctionf(x) = (3n+1 4x)2 anditiseasy m − 4 tocheckthataslongasn 5 happenonsomeU-blockornot. Ifyesweusethedecodings ≥ forGottesmanscodesandifnotthenwehaveonlytodecod- f(0) = (3n+1)2 >(3n+5)(3n 7)+16, (18a) − ingD(0,a) withthedetailedcircuitinthecaseofa = 0being f(1) = (3n 3)2 >4(3n+5), (18b) givenin[12]. − f(2) = (3n 7)2 >2(3n+5)+16. (18c) We acknowledge the financial support of NNSF of China − (Grant No. 90303023, 10675107, and 10705025) and the If there exists a stabilizer code [[n,k,3]] then Eqs.(17a-17c) A*STARgrantR-144-000-189-305. musthold. Asaresult Noteadded. Onfinishingthepaperanotherinfinitefamily ofgenuinenonadditivecodeshasbeenreportedin[22]. f(x) =3n+1+4A +2A , (19a) 1 2 (cid:104) (cid:105) n n (cid:98)2(cid:99) (cid:98)2(cid:99) 16+16A + f(2i)A 16 A 8.2s,(19b) 2 2i 2i ≥ ≥ i=2 i=0 (cid:88) (cid:88) [1] P.W.Shor,Phys.Rev.A,2,2493(1995). wherewehaveusedthatfactthatf(2i) 16since 3n+1,the ≥ 4 [2] C. H. Bennett, D. P. DiVincenzo, J. A. Smolin, andW. uniquezerooff(x),isanoddinteger. Puttingallthesepieces K.Wootters,Phys.Rev.A54,3824(1996). togetherweobtain [3] A.Steane,Phys.Rev.Lett.77,793(1996). [4] E.KnillandR.Laflamme,Phys.Rev.A55,900(1997). n [5] D.Gottesman,Phys.Rev.A541862(1996). 2s f(x) = f(i)A (cid:104) (cid:105) i [6] D.Gottesman,arXive:quant-ph/9705052. (cid:88)i=0 [7] A. Calderbank, E. Rains, P. Shor, and N. Sloane, Phys. Rev. n (cid:98)2(cid:99) Lett.76,405(1997). f(0)+f(1)A1+f(2)A2+ f(2i)A2i [8] A.Calderbank,E.Rains,P.Shor,andN.Sloane,IEEETrans. ≥ i=2 Inform.Theory,44,1369(1998). f(0) 16+f(1)A1+(f(2)(cid:88)16)A2+8.2s [9] E.M.Rains,R.H.Hardin,P.W.Shor,andN.J.A.Sloane,Phys. ≥ − − > (3n+5)(3n 7+4A +2A )+8.2s Rev.Lett.79,953(1997). 1 2 − = (3n+5) f(x) 8 +8.2s, (20) [10] E.M.Rains,IEEETrans.Inf.Theory45,266(1999). (cid:104) − (cid:105) [11] J.A. Smolin, G. Smith and S. Wehner, Phys. Rev. Lett. 99, inwhichthestrictinequalitycomesfromthef(0)term. Tak- 130505(2007). ingintoaccountof f(x) > 8weobtain2s > 3n+5, i.e., [12] S.Yu,Q.Chen,andC.H.Oh,arXiv:0709.1780v1[quant-ph] n k 2m+6. (cid:104) (cid:105) [13] S.Yu,Q.Chen,C.H.Lai,andC.H.Oh,Phys.Rev.Lett.101, − ≥ 090501(2008) Inthecaseofa = 1,i.e.,n = N1 withm 0wedefine m ≥ [14] M.GrasslandM.Roetteler,Proc.2008IEEEInt.Symp.onInf. g(x) = (3n+2 4x)(3n 2 4x) which is nonnegative − − − Theory(ISIT2008),300(Toronto,Canada,July2008). on integers because 3n4+2 is an integer. It is obvious that as [15] A.Cross, G.Smith, J.Smolin, andB.Zeng, IEEETrans.Inf. long as n 5 we have g(i) > 2(3n+2) for i = 1,2 and Theory55,433(2009). ≥ mostimportantlyg(0) > (3n+2)(3n 4). Ifthereexistsa [16] D.Gottesman,arXive:quant-ph/9607027. stabilizercode[[n,k,3]]thenEqs.(17a-1−7c)musthold,which [17] E. Rains, IEEE Trans. Inform. Theory 44, 1388 (1998); ibid, leadsto g(x) =3n 4+2A +2A . Asaresultwehave IEEETrans.Inform.Theory45,2361(1999). 1 2 (cid:104) (cid:105) − [18] P.ShorandR.Laflamme,Phys.Rev.Lett.78,1600(1997). 2s g(x) g(0)+g(1)A +g(2)A [19] A.AshikhminandS.Litsyn, IEEETrans.Inform.Theory45, 1 2 (cid:104) (cid:105) ≥ 1206(1999). > (3n+2)(3n 4+2A +2A ) − 1 2 [20] D. Schlingemann and R.F. Werner, Phys. Rev. A 65, 012308 = (3n+2) g(x) , (21) (cid:104) (cid:105) (2001). [21] M. Hein, J. Eisert, and H.J. Briegel, Phys. Rev. A 69. in which the strict inequality sign is due to the g(0) term. 062311(2004). Since g(x) >0wehave2s >3n+2,i.e.,n k 2m+6. (cid:3) (cid:104) (cid:105) − ≥ [22] M. Grassl, P. Shor, G. Smith, J. Smolin, and B. Zeng, arXiv: 0901.1319[quant-ph]. It should be noted that the optimal stabilizer codes of pa- rametersasgiveninEq.(2)existandconstructisalreadygiven by the stabilizers in Table I with the stabilizers acting on qubits V or V being replaced by 6 stabilizers of the pure 0 1 optimalstabilizercodes[[9,3,3]]or[[10,4,3]]. Thestabilizer-likestructuresofourcodessimplifysignifi- cantlytheencodinganddecodingprocedures. Letussuppose we have already the encoding and decoding circuits for the codesD(0,a) andfortheGottesman’scodes. Withsomeaddi- tionalcontrolled-notgatesinfrontoftheencodingcircuitsof theseindividualcodesweobtaintheencodingsofourcodes. Todecodewehaveonlytocheckatfirstthefirst2mgenera- tors in Table I, from which we can be sure wether the errors

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.