arXiv:1001.0543v2 [quant-ph] 10 Oct 2010 ulycmlmnaymaueet a enpooe by proposed mu been on has Bogdanov based measurements states complementary biphotons tually of pro- reconstruction correlations[5] tomography statistical N-particle this of of measurements implementation requires the cess But standa singl the [4]. in a probabilities cedure six determining the fully than rather for state, qubit needed are probabilities ment em[] eety iia ui oorpypoeshas process tomography by qubit proposed Minimal been Recently, li a of [3]. state de- polarization beam the first determine was to technique Stokes by The o veloped reconstruction 2]. state tomography[1, quantum state called un quantum an is of which matrix state, density the known reconstruct quan- to implementing can is for one computation task main tum information A the system. all that about contains obtain by which described matrix, be can density state a quantum a and mechanics, quantum ∗ etfloecnesgas8.Moreva indepen- nine signals[8]. from fluorescence state dent qutrit mixed unknown an determine he-iesoa qtis tt ae nteefc fre of Allevi effect operators[10]. the measurement tic on based state 2009 (qutrits) In three-dimensional [9]. tomography state Taguchi prot ququarts optimal polarization the for of col realization the of problem experimental ueetrslsaentidpneto ahohr othere so other, each of mea- num- independent These large not system. a are the on results of measurements copies surement prepared of identically series of a ber perform to need photodetectors[11]. switchi avalanche by photon by single states on/off thermal and coherent t for and function matrix Wigner density the of reconstruction the of mentation [email protected] h unu tt fasse safnaetlcnetin concept fundamental a is system a of state quantum The nodrt bantefl nomto bu h ytmwe system the about information full the obtain to order In tal et 2 eateto hsc lcrncEgneig ee Norm Hefei Engineering, Electronic & Physics of Department tal et eeoe h igesa oorpyo spatial of tomography scan single the developed nomto at,w ee oorshm steotmlsche optimal the as s scheme our of to process refer reconstruction we waste, the information Because explicitly. sented ASnmes 36.j 36.a 36.d 03.67.Mn 03.65.Ud, 03.65.Ta, 03.65.Wj, numbers: PACS d rec five economical the most over the realize needed to operations scientists nonlocal experimental of i we number so the deco properties, imize be entanglement different can sy three-qutrit with transformations the structures those For all operations. unitary that two-qutrit found and is mutual It each connects basis. mutual that different transformation those each how decompose to to paid be will attention thi in Special minimum the reaches operations conditional required ae nmtal nisdmaueet,a pia tomogr optimal an measurements, unbiased mutually on Based 6 ] Ivanov 7]. [6, Reh´aˇcek ˇ .INTRODUCTION I. 1 e aoaoyo poeetoi nomto Acquisitio Information Opto-electronic of Laboratory Key tal et pia eosrcino h ttsi urt system qutrits in states the of reconstruction Optimal tal et hr nyfu measure- four only where , iityo dcto,Sho fPyisadMtra Scien Material and Physics of School Education, of Ministry nu nvriy ee 309 epesRpbi fChina of Republic People's 230039, Hefei University, Anhui rpsdamto to method a proposed tal et tal et e Yan, Fei tde h imple- the studied adatninto attention paid 1 igYang Ming dpro- rd .The alis- ght ng he o- e r - - , ∗ , 1 tm hr xs v ifrn uulyunbiased-bases mutually different five exist there stem, n hoLagCao Zhuo-Liang and ntuto fqatmsae nqti systems. qutrit in states quantum of onstruction touetecneto hsclcmlxt omin- to complexity physical of concept the ntroduce nw est arx[4.Te rvdta h maximal the un- that an proved is of They MUBs specification of [14]. numbers complete matrix of way density known optimal as well minimal as a provide bases, (MUMs) of measurements class unbiased Wootters special mutually this by in shown measurements been that has Fields and it by Then system prime-dimension construction. the explicit in an proved bases and such [13], of existence determination the state t quantum MUBs of of problem a concept the doe in the measured state introduced is Ivanovi´c first it basis basis. if particular state other a the equal about over are information detection any bases give the other not the say in to vectors is basis That all and vecto a basis between one overlap de- squared in are the that MUBs property [13–36]. the by mechanics fined quantum in topics of riety a e fmaueet o eoigterdnac sof is opti- processin redundancy an information the quantum design removing in to significance for So, fundamental measurements of one. set optimal mal an reconstructio become the will completely, redundancy process waste. this resources remove a we causes If quantu which used [12], previously the processes in tomography results these in redundancy is eutas ple otepiepwrdmninsystem. prime-power-dimension the to applies also result a rpsdb Klimov by schem reconstruction proposed tomographic was optimal [13–16] an state Recently pro- quantum and a etc. bases determining for measurement way of optimal set an minimal vides a forms it as such esoa qdt ytm a enpooe n h num- the is and measurements proposed required been of ber has di- high systems in (qudit) tomography mea- Quantum mensional existing 39]. distri- currently [38, key the protocol quantum surement than a MUMs be with on to protocol three-level attack proven bution symmetric of been a in use has against systems MUMs the securer two-level However, on of based instead [37]. systems system system quantum ions multiqubit trapped of state a eso fteqdtsse and system qudit the of mension uis[2.Ti oorpypoesi o notmlone, optimal an not is process tomography This [12]. qudits oorpi ceefrtesae fqti systems. qutrit of states the for scheme tomographic s lUiest,Hfi206,Pol' eulco China of Republic People's 230061, Hefei University, al yubae esrmnsaeraie;ta s how is, that realized; are measurements unbiased ly ae ae nmtal nisdsae sfe of free is states unbiased mutually on based tates uulyubae ae MB)hv enue nava- a in used been have (MUBs) bases unbiased Mutually Uspa pca oei eemnn unu states, quantum determining in role special a play MUBs yubae ai ihtesadr computational standard the with basis unbiased ly psdit eea ai mlmnal single- implementable basic several into mposed e yotmlw enta h ubro the of number the that mean we optimal By me. feetsrcue.Ti ceei epu for helpful is scheme This structures. ifferent pi ceefrtemliurtsae spre- is states multiqutrit the for scheme aphic n Manipulation, and n 2 d 1 + ce, tal et npiedmninsse.This system. prime-dimension in d o h aeo determining of case the for 2 n n en h ubro the of number the being − 1 with d en h di- the being i.e. n- g. m o n e s r . 2 and there is a big redundancy among the measurement results II. MUTUALLY UNBIASED BASES AND MUTUALLY there. To remove this redundancy, we will propose an optimal UNBIASED MEASUREMENTS tomography process for qutrits states. This optimal quantum tomography process is the MUBs-based qutrit states tomog- As shown by Wootters, Fields [14] and Klappenecker, raphy, and the number of required measurements is greatly R¨otteler [19], in finite field language, the first MUB in a reduced. A d-dimensional quantum system is represented by d = pn (p = 2)-dimensional quantum system is the standard a positive semidefinite Hermitian matrix ρ with unit trace in 0 6 (0) basis B given by the vector (a )l = δkl,k,l ̥pn , where d-dimensional Hilbert space, which is specified by d2 1 k ∈ − the superscript denotes the basis, k the vector in the basis, l the real parameters. A nondegenerate measurement performed on n component and ̥pn is the field with p elements. The other d such a system provides d 1 independent probabilities. So, in r (r) − MUBs are denoted by B which consists of vectors (ak )l general, one requires at least d +1 different orthogonal mea- 2 (r) T r(r·l +k·l) surements for fully determining an unknown ρ. For n-qutrit defined by[14]: (ak )l = (1/√d)ω ,r,k,l n p ∈ tomography it only needs 3 + 1 measurements. Through ̥pn , r = 0. Here ω = exp(2πi/p) and rθ = θ + θ + 2 6 n−1 T analysis, one can find that the tomography process for qubit θp + + θp . The set of mutually unbiased projec- system can not be generalized to qudit system case in a trivial ··· (r) (r) (r) tors can be given by Pk = ak ak . It is worth notic- way. The MUBs-based tomography process for qubit system (r) | ih | 0 ing that ak contains the computational basis B . Here proposed by Klimov et al can not be directly applied to qutrit (s) | (r) i T r(P P )=(1/d)(1 δsr +dδsrδjk). Then the measure- system [37]. This is because the entanglement feature of the j k − MUBs of qubits system is totally different from that of qutrits (r) (r) ment probabilities given by ωk = T r(Pk ρ) completely system. So, we will study the physical implementation of an determine the unknown density operator of a d-dimensional optimal tomographic scheme for the case of determining the d d−1 (r) (r) system [13]: ρ =Σr=0Σk=0ωk Pk I. states of multi-qutrit system based on the MUMs. For instance, in a qutrit system, there− are three MUBs be- side the computational basis B0 = 0 , 1 , 2 , in the fol- lowing form with ω = exp(2πi/3): {| i | i | i} From the experimental point of view, the physical complex- ity is a key point for the implementation of a scheme. Here, 1 (1) √ B : a0 = (1/ 3)( 0 + 1 + 2 ), for multi-qutrit quantum tomography, the physical complex- {| i | i | i | i a(1) = (1/√3)( 0 + ω 1 + ω∗ 2 ), ity mainly comes from the entanglement bases, i.e. the num- | 1 i | i | i | i ber of the two-qutrit conditional operation needed for the de- (1) √ ∗ a2 = (1/ 3)( 0 + ω 1 + ω 2 ) ; (1a) composition of these entangled bases. In addition, there ex- | i | i | i | i } ists many different MUBs with different entanglement prop- erties in multi-qutrit system. So the physical complexity of 2 (2) √ B : a0 = (1/ 3)(ω 0 + 1 + 2 ), the quantum tomography process here depends on the entan- {| i | i | i | i a(2) = (1/√3)( 0 + ω 1 + 2 ), glement structure of the MUBs used, and it becomes very im- | 1 i | i | i | i (2) portant to optimize the quantum tomography process over all a2 = (1/√3)( 0 + 1 + ω 2 ) ; (1b) the possible MUBs entanglement structures of the system. | i | i | i | i }

B3 : a(3) = (1/√3)(ω∗ 0 + 1 + 2 ), {| 0 i | i | i | i This paper is arranged as follows. In the next section we a(3) = (1/√3)( 0 + ω∗ 1 + 2 ), introduce the MUBs in a d-dimensional system (d = pn | 1 i | i | i | i (3) √ ∗ ) and show how to reconstruct an unknown state by a2 = (1/ 3)( 0 + 1 + ω 2 ) . (1c) (p = 2) | i | i | i | i } MUMs.6 Here p is a prime. Section III briefly reviews the general method to reconstruct a qutrit state, where the number of measurements is 8. However the MUBs-based qutrit to- III. RECONSTRUCTION PROCESS FOR AN ARBITRARY mography proposed here only needs 4 measurements, which SINGLE QUTRIT STATE means the number of measurements is reduced. Here the mea- surement reduction for single qutrit case is not obvious, so in An unknownsingle qutrit state can be expressed as [12, 40]: section IV we will extend the one-qutritcase to two-qutrit sys- ρ = (1/3) 8 r λ , where λ is an identity operator and Pj=0 j j 0 tem. For the two-qutrit system the number of measurements the other λj are the SU(3) generators [41]. The general is only 10 for determining all the elements of the density op- method to reconstruct the qutrit state is to measure the ex- 4 erator rather than 3 1 = 80 measurements in the scheme pectation values of the λ operators [12], where rj = λ = − h j i proposed in Ref. [12]. It means a great reduction of the ex- T r[ρλj ]. Thus one will find that the number of required mea- perimental complexity. So we conclude that the optimal mea- surements is 8. However, if we choose the MUMs to deter- surements on the unknown qutrit states are the MUMs. In sec- mine the qutrit state, the number of needed MUMs is only 4 tion V we discuss the physical complexity for implementing rather than 8 of Ref. [12]. The four optimal set of MUBs have the MUMs in three-qutrit system, and give the optimal MUB been presented by Eqs.(1a,1b,1c) plus the standard computa- for the qutrit system quantum tomography process with mini- tional basis in the preceding section. Each of the three MUBs mized physical complexity. The last section is the conclusion. in Eqs.(1a,1b,1c) is related with the standard computational 3

TABLE I: The transformations connecting the MUBs with the stan- TABLE II: The decompositions of MUBs for the two-qutrit system dard computational basis for a qutrit system based on Fourier trans- based on Fourier transformations, phase operations and controlled- forms and the phase operations. NOT gates (X12) [42] with the first particle as source and the second one as target. The subscript denotes the ith particle, i = 1, 2. Basis Transformation 2 F −1 Basis Decompositions −1 −1 −1 3 F R 2 F1 F2 −1 −1 −1 −1 4 F R 3 F1 R1F2 R2 −1 −1 −1 4 F1 X12F2 R2 −1 −1 −1 −1 5 F1 X12 R1F2 R2 −1 −1 −1 −1 basis by a unitary transformation. These transformations have 6 F1 X12 F2 R1 −1 −1 −1 −1 been listed in Table.I. Here, F denotes the Fourier transfor- 7 F1 R1 F2 R2 −1 −1 −1 mation: 8 F1 X12 F2 R2 −1 −1 −1 2 9 F1 R1 X12F2 R2 √ −1 −1 F j = (1/ 3) X exp(2πilj/3) l , j =0, 1, 2, (2) F R1X12F | i | i 10 1 2 l=0 R denotes a phase operation: R = 0 0 + ω 1 1 + ω 2 2 , (3) The fidelity of a practical CNOT gate can reach a value up to | ih | | ih | | ih | 0.926 for trapped ions system in Lab [43]. Klimov et al have and the Controlled gate is introduced the concept of the physical complexity of each set of MUBs as a function of the numberof nonlocalgates needed X i j = i j i . (4) | i| i | i| ⊖ i for implementing the MUMs [37]. Here the fidelity value of Where denotes the difference j i modulo 3.If there are n the CNOT gates for qubits systems also can be used to eval- qutrits,⊖ the number of MUM is 3n−+1, which is far less than uate the physical complexity of the MUBs of qutrits systems. 32n 1 in Ref. [12]. That is to say the use of MUMs can Why we can say so is because of the following point. Al- represent− a considerable reduction in the operations and time though the systems involved here are three-state ones, all the required for performing the full state determination [37]. operations used in our reconstruction process can be decom- posed into effective two-state operations. So the complexity of the current tomography scheme is proportional to the num- IV. RECONSTRUCTION PROCESS FOR AN ARBITRARY ber of the nonlocal gates used(C 6) for two qutrits system. TWO-QUTRIT STATE As shown in Ref[44], the only MUB∝ structure for a two-qutrit system is (4, 6), where 4 is the number of the separable bases Now if we further extend one-qutrit case to two-qutrit and 6 is the number of the bipartite entangled bases. case, the density matrix can be expressed as: ρ12 = However in three-qutrit case, there are five 8 where . Ifwe sets of MUBs with different structures, namely (1/9) Pj,k=0 rjkλj λk , rjk = λj λk use the general method⊗ in Ref. [12] to fullyh ⊗ determinei the (0, 12, 16),(1, 9, 18),(2, 6, 20),(3, 3, 22),(4, 0, 24) [21, 44]. { } state, d2n 1=34 1 = 80 measurements will be needed. It is easy to see that the (0, 12, 16) set of MUBs has the So much measurements− − will inevitably introduce redundant minimum physical complexity. We say that the optimal set of information of the state, which is obviously a resource waste. the MUBs is (0,12,16). The decompositions of the MUBs in So here we will take advantage of the MUMs to reconstruct the three-qutrit case for (0,12,16) are listed in Table.III. So the two-qutrit state. It is easy to find that the nine elements of the set of MUBs (0,12,16) has a complexity C 44, which ∝ ̥9 (finite field) are 0, α, 2α, 1, 1+α, 1+2α, 2, 2+α, 2+2α is a very important value in experimental realization of it. by using the irreducible{ polynomials θ2 +θ+2=0 [14]. Here} For the multi-qutrit system(n > 3), it is not easy to get all we use the representation 0 , α , 2α 2+2α as the the sets of MUBs, and it is even more difficult to get the ex- standard basis. {| i | i | i · · · | i} plicit decompositions of the optimal set of MUBs. Neverthe- One can find that there will be only d2 +1=32 +1=10 less, the results for the two-qutrit and three-qutrit cases have MUMs to be done, which is much less than 80 of Ref. [12]. It provided the experimentalists valuable references. means that the operations and time needed for the whole state determination is great reduced. The decompositions for all the MUBs of the two-qutrit system have been listed in Table. II. VI. CONCLUSION

We have explicitly presented an optimal tomographic V. THE PHYSICAL COMPLEXITY FOR IMPLEMENTING scheme for the single-qutrit states, two-qutrit states and three- THE MUMS IN THE THREE-QUTRIT SYSTEM qutrit states based on the MUMs. Because the MUBs based state reconstruction process is free of information waste, In general, the fidelity of single logic gates can be greater the minimal number of required conditional operations are than 99%, but nonlocal gates have a relatively lower fidelity. needed. So we call our qutrits tomographic scheme the op- 4

TABLE III: The decompositions of MUBs (0,12,16) for three-qutrit system

Basis Decompositions Basis Decompositions −1 −1 −1 −1 1 F2 X23F1 F3 15 F2 R2X23 −1 −1 −1 −1 −1 −1 −1 −1 −1 −1 2 F1 R1 F2 R2 X23F3 R3 16 F1 R1X13F2 R2X23F3 R3 −1 −1 −1 −1 −1 3 F1 X13R3 17 F1 R1 X13F2 F3 −1 −1 −1 −1 −1 −1 −1 −1 −1 −1 −1 −1 4 F2 R2X23 F1 X12R2F3 18 F1 R1 X13 F2 R2 X23 R3F3 R3 −1 −1 −1 −1 −1 −1 −1 −1 5 F1 X12 F3 19 F1 R1X13 F2 R2 F3 R3 −1 −1 −1 −1 −1 −1 −1 6 F2 R2X23F1 R1X13 R3F3 R3 20 F1 X12F3 R3 −1 −1 −1 −1 −1 −1 −1 −1 −1 −1 −1 −1 7 F2 X23 R3 F1 R1 X12 R3 21 F1 R1 X12 F2 R2F3 R3 −1 −1 −1 −1 −1 −1 −1 −1 −1 −1 −1 8 F1 R1 X13 F2 R2F3 R3 22 F1 X12 F2 X23 F3 R3 −1 −1 −1 −1 −1 −1 −1 −1 −1 −1 9 F1 X13 F2 R2X23 F3 R3 23 F1 R1 X13F2 R2 X23 −1 −1 −1 −1 −1 −1 −1 −1 10 F1 R2X12X13 F2 24 F1 R1 X13 F2 X23R3F3 −1 −1 −1 −1 −1 −1 −1 −1 −1 −1 11 F1 R1X12 F2 R2X23 F3 25 F1 R1X12 X23 F2 R2 −1 −1 −1 −1 −1 −1 −1 −1 −1 −1 −1 12 F1 R1 X13X12 F2 F3 R3 26 F1 R1X12X23 F2 R2 F3 −1 −1 −1 −1 −1 −1 −1 −1 −1 13 F1 X12F2 R2 X23 F3 R3 27 F1 R1 X12F2 R2 −1 −1 −1 −1 −1 −1 14 F1 R1X12F2 X23F3 R3 28 F1 R1F2 X23 timal one. Here, we explicitly decompose each measure- Natural Science Foundation of China (NSFC) under Grants ment into several basic single- and two-qutrit operations. Fur- No. 10704001 and No. 61073048, the Key Project of Chinese thermore, all these basic operations have been proven imple- Ministry of Education.(No.210092), the Key Program of the mentable [42]. The physical complexityof a set of MUBs also Education Department of Anhui Province under Grants No. has been calculated, which is an important threshold in experi- KJ2008A28ZC, No. KJ2009A048Z, No. 2010SQRL153ZD, ment. We hope these decompositions can help the experimen- and No. KJ2010A287, the Talent Foundation of Anhui Uni- tal scientists to realize the most economical reconstruction of versity, the personnel department of Anhui province, and An- quantum states in qutrits system in Lab. hui Key Laboratory of Information Materials and Devices Acknowledgments. This work is supported by National (Anhui University).

[1] G. M. D’Ariano, M. G. A. Paris, and M. F. Sacchi, Advances in [22] M. Planat and H. Rosu, Eur. Phys. J. D 36, 133 (2005). Imaging and Electron Physics 128, 205 (2003). [23] T. Durt, J. Phys. A 38, 5267 (2005). [2] G. M. D’Ariano, M. G. A. Paris, and M. F. Sacchi, Lecture [24] A. O. Pittenger and M. H. Rubin, J. Phys. A 38, 6005 (2005). Notes in Physics, vol. 649, Springer- Verlag, Berlin, 7 (2004). [25] A. B. Klimov et al., J. Phys. A 39, 14471 (2006). [3] G. G. Stokes, Trans. Cambridge Philos. Soc. 9, 399(1852). [26] W. K. Wootters, Found. Phys. 36, 112 (2006). [4] J. Reh´aˇcekˇ et al., Phys. Rev. A 70, 052321 (2004). [27] A. J. Scott, J. Phys. A 39, 13507 (2006). [5] A. Ling et al., Phys. Rev. A 74, 022309 (2006). [28] A. B. Klimov et al., J. Phys. A 40, 3987 (2007). [6] Y. I. Bogdanov et al., JETP Letters, 78 (6), 352 (2003). [29] I. Bengtsson et al., J. Math. Phys. 48, 052106 (2007). [7] Y. I. Bogdanov et al., Phys. Rev. A 70, 042303 (2004). [30] M. A. Jafarizadeh et al., arXiv:0801.3100v1 (2008). [8] P. A. Ivanov and N. V. Vitanov, Opt. Commun. 264, 368 (2006). [31] S. Brierley and S. Weigert, Phys. Rev. A 78, 042312 (2008). [9] E. V. Moreva et al., arXiv:0811.1927v2 (2008). [32] M. R. Kibler, J. Phys. A 42, 353001 (2009). [10] G. Taguchi et al., Phys. Rev. A 80, 062102 (2009). [33] P. Jaming et al., arXiv:0902.0882v2 (2009) [11] A. Allevi et al., Phys. Rev. A 80, 022114 (2009). [34] S. Brierley, S. Weigert, I. Bengtsson, arXiv:0907.4097 (2009). [12] R. T. Thew et al., Phys. Rev. A 66, 012303 (2002). [35] A. Ambainis, arXiv:0909.3720v1 (2009). [13] I. D. Ivanovi´c, J. Phys. A: Math. Gen, 14, 3241 (1981). [36] D. M. Appleby, arXiv:0909.5233v1 (2009). [14] W. K. Wootters, B. D. Fields, Ann. Phys. (NY) 191, 363 (1989). [37] A. B. Klimov et al., Phys. Rev. A 77, 060303(R) (2008). [15] S. Bandyopadhyay et al., Algorithmica, 34, 512 (2002). [38] D. Bruss, C. Macchiavello, Phys. Rev. Lett. 88, 127901 (2002). [16] M. Revzen, arXiv:0912.5433v1(2009). [39] N. J. Cerf et al., Phys. Rev. Lett. 88, 127902 (2002). [17] J. Lawrence et al., Phys. Rev. A 65, 032320 (2002). [40] D. F. V. James et al., Phys. Rev. A 64, 052312 (2001). [18] S. Chaturvedi, Phys. Rev. A 65, 044301 (2002). [41] C. M. Caves and G. J. Milburn, Opt. Commun. 179, 439 (2000). [19] A. Klappenecker and M. R¨otteler, 7th International Conference [42] A. B. Klimov et al., Phys. Rev. A 67, 062313 (2003). on Finite Fields and Applications, Fq7, Lecture Notes in Com- [43] M. Riebe et al., Phys. Rev. Lett. 97, 220407 (2006). puter Science 2948, 262 (2004). [44] J. Lawrence, Phys. Rev. A 70, 012302 (2004). [20] A. B. Klimov et al., J. Phys. A 38, 2747 (2005). [21] J. L. Romero et al., Phys. Rev. A 72, 062310 (2005).